{ "title": "A mind outside the state", "speaker": "Jacob Steeves", "event": "Exploit Conference, Montreal, 28 September 2026", "dataAsOf": "2026-09-17", "slides": [ { "id": 1, "title": "Bittensor Trend Data", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Small grey title: \"Bittensor Trend Data · 2021–2026 · log scale\". Four series on one log axis, powers of ten down the left, years 2021 to 2026 along the bottom, no gridlines. In black: \"Subnets\" (1 in 2021, 128 by July 2025, flat since) and \"Subnet tokens, value in TAO\" (from the dashed line to about 5.5 million). In grey: \"TAO minted per day\" (7,200, flat for five years, halves to 3,600 in December 2025) and \"TAO staked into subnets\" (from the dashed line to about 2 million, flat since March 2026). One dashed vertical line at 13 February 2025 with an arrowhead and two lines above it: \"dTAO · 13 Feb 2025\" / \"The committee stops.\" Hollow dots are shape points not verified on-chain (the 2021–2022 points). One sentence beneath: \"Five years. Half the minting. A hundred and twenty-eight markets.\"", "minute": "0:00", "content": { "kind": "trendlog", "title": "Bittensor Trend Data · 2021–2026 · log scale", "annotation": "Five years. Half the minting. A hundred and twenty-eight markets.", "axis": { "years": [ { "label": "2021", "at": 0 }, { "label": "2022", "at": 0.16666666666666666 }, { "label": "2023", "at": 0.3333333333333333 }, { "label": "2024", "at": 0.5 }, { "label": "2025", "at": 0.6671232876712329 }, { "label": "2026", "at": 0.8337899543378996 } ], "minExponent": 0, "maxExponent": 7 }, "inflection": { "at": 0.6867579908675799, "label": [ "dTAO · 13 Feb 2025", "The committee stops." ] }, "series": [ { "label": "Subnets", "tone": "black", "points": [ { "at": 0.0009132420091324201, "value": 1, "verified": false }, { "at": 0.2356164383561644, "value": 1, "verified": false }, { "at": 0.3689497716894977, "value": 1, "verified": true }, { "at": 0.3881278538812785, "value": 2, "verified": true }, { "at": 0.40730593607305937, "value": 3, "verified": true }, { "at": 0.426027397260274, "value": 4, "verified": true }, { "at": 0.4452054794520548, "value": 4, "verified": true }, { "at": 0.4648401826484018, "value": 10, "verified": true }, { "at": 0.48356164383561645, "value": 31, "verified": true }, { "at": 0.5031963470319635, "value": 32, "verified": true }, { "at": 0.5429223744292238, "value": 32, "verified": true }, { "at": 0.5639269406392694, "value": 34, "verified": true }, { "at": 0.5831050228310503, "value": 39, "verified": true }, { "at": 0.6027397260273972, "value": 45, "verified": true }, { "at": 0.6219178082191781, "value": 49, "verified": true }, { "at": 0.6410958904109589, "value": 52, "verified": true }, { "at": 0.6602739726027397, "value": 62, "verified": true }, { "at": 0.6794520547945205, "value": 64, "verified": true }, { "at": 0.6981735159817352, "value": 74, "verified": true }, { "at": 0.717351598173516, "value": 94, "verified": true }, { "at": 0.736986301369863, "value": 116, "verified": true }, { "at": 0.7561643835616438, "value": 128, "verified": true }, { "at": 0.8319634703196347, "value": 128, "verified": true }, { "at": 0.9082191780821918, "value": 128, "verified": true }, { "at": 0.9547945205479452, "value": 128, "verified": true } ] }, { "label": "TAO minted per day", "tone": "grey", "points": [ { "at": 0.0009132420091324201, "value": 7200, "verified": false }, { "at": 0.3689497716894977, "value": 7200, "verified": true }, { "at": 0.826027397260274, "value": 7200, "verified": true }, { "at": 0.826027397260274, "value": 3600, "verified": true }, { "at": 0.9547945205479452, "value": 3600, "verified": true } ] }, { "label": "TAO staked into subnets", "tone": "grey", "points": [ { "at": 0.6867579908675799, "value": 45, "verified": true }, { "at": 0.6981735159817352, "value": 165040, "verified": true }, { "at": 0.717351598173516, "value": 522877, "verified": true }, { "at": 0.736986301369863, "value": 705962, "verified": true }, { "at": 0.7561643835616438, "value": 874624, "verified": true }, { "at": 0.7748858447488585, "value": 1040455, "verified": true }, { "at": 0.7940639269406393, "value": 1287544, "verified": true }, { "at": 0.8132420091324201, "value": 1559804, "verified": true }, { "at": 0.8319634703196347, "value": 1757997, "verified": true }, { "at": 0.8511415525114155, "value": 1902872, "verified": true }, { "at": 0.8703196347031964, "value": 2046856, "verified": true }, { "at": 0.8894977168949771, "value": 2017763, "verified": true }, { "at": 0.9082191780821918, "value": 1973097, "verified": true }, { "at": 0.9273972602739726, "value": 2005281, "verified": true }, { "at": 0.9465753424657535, "value": 2033118, "verified": true }, { "at": 0.9547945205479452, "value": 1967851, "verified": true } ] }, { "label": "Subnet tokens, value in TAO", "tone": "black", "points": [ { "at": 0.6867579908675799, "value": 64, "verified": true }, { "at": 0.6981735159817352, "value": 399674, "verified": true }, { "at": 0.717351598173516, "value": 1578507, "verified": true }, { "at": 0.736986301369863, "value": 1888427, "verified": true }, { "at": 0.7561643835616438, "value": 1973819, "verified": true }, { "at": 0.7748858447488585, "value": 2139111, "verified": true }, { "at": 0.7940639269406393, "value": 2640086, "verified": true }, { "at": 0.8132420091324201, "value": 3180322, "verified": true }, { "at": 0.826027397260274, "value": 3458821, "verified": true }, { "at": 0.8319634703196347, "value": 3649892, "verified": true }, { "at": 0.8511415525114155, "value": 4253899, "verified": true }, { "at": 0.8703196347031964, "value": 5023206, "verified": true }, { "at": 0.8894977168949771, "value": 4964193, "verified": true }, { "at": 0.9082191780821918, "value": 4918033, "verified": true }, { "at": 0.9273972602739726, "value": 5237088, "verified": true }, { "at": 0.9465753424657535, "value": 5701492, "verified": true }, { "at": 0.9547945205479452, "value": 5456350, "verified": true } ] } ] }, "notes": [ "0:00. The Jensen opening. Walk out onto this chart, no greeting yet. Say the claim cold, in one sentence: \"Bittensor is driving a new computing model. Incentive computing.\" [Jacob, 23 Sep] Then make it expensive, the way Jensen does: \"Developing a new computing model takes an enormous amount of time and set of conditions.\" [Jacob, 23 Sep]", "Why: \"We wanted to solve problems that we think can only be solved by international organization and cooperation. We could see that the standard corporation was changing and that with the advent of AI we could organize ourselves more effectively in markets than in contracts.\" [Jacob, 23 Sep]", "Read the chart left to right, one line each. The grey flat line is what the protocol mints: 7,200 TAO a day for five years, then half. It never grew. It was cut. The black line from the floor is subnets: one network in 2021, thirty-two by the end of 2023, sixty-four when the market took over, a hundred and twenty-eight by July last year. Then the cap. [on-chain, archive node, 23 Sep 2026; `content/data.json → trend`]", "The dashed line is the thirteenth of February 2025. Before it, sixty-four root validators voted on where every new TAO went. After it, nobody votes; you stake into the market you believe in, and the price sets the split. That is the analogue of Jensen's \"Dennard scaling has stopped\": the committee stopped, and the market started. Two lines begin at that date from nothing: the TAO holders have moved into subnets, about two million; and what the market says those subnets are worth, in TAO not dollars, about five and a half million. [on-chain, 23 Sep 2026]", "Be honest about the shape. The staked line has been flat at about two million since March 2026. The value line dipped this month. The subnet line is capped by a protocol constant, not by demand. Say: \"The minting halved and the slots filled. What grew is what people put into them.\" Do not say \"up only\". [VERIFY: re-pull all four series on 27 Sep with `content/data.json → trend.source`; the value line is a valuation and can fall]", "Do not say anything about the TAO price. Every number on this chart is in TAO or a count.", "\"Today I will cover where we have come, pointing to a number of examples on Bittensor to show how we are directionally aligned. As a network we have made enormous strides in this direction.\" [Jacob, 23 Sep] Advance on \"examples\"." ] }, { "id": 2, "title": "Incentive computing begins with subnets.", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. One line, black, above a row of eight cards: \"Incentive computing begins with subnets.\" Each card is a thin black frame with, top to bottom: the netuid in small grey mono (\"SN51\"), the name (\"Lium\"), one grey line of function, and one number in mono above a hairline. Left to right: Lium · GPUs by the second. No account check. · $964K billed in a month / Targon · Compute inside a sealed enclave. · 550 TAO bought back from revenue / Hippius · Storage. Egress free. · 100 TB egress, billed $0 / NOVA · Molecules against protein targets. · 4.8M molecules screened / Chutes · Any open model, in attested hardware. · 2M requests a day / SayGM · Frontier models below list price. · 96B tokens routed / RedTeam · Paid to break bot detection. · Seven-figure revenue / OpenRoboto · Robot policies that beat the champion. · 50% → 85% on LIBERO-Pro in 21 days. Small grey beneath: \"No account. No ID. No permission.\" On a phone the row becomes two columns of four.", "minute": "1:30", "content": { "kind": "cards", "heading": "Incentive computing begins with subnets.", "annotation": "No account. No ID. No permission.", "cards": [ { "netuid": 51, "name": "Lium", "function": "GPUs by the second. No account check.", "number": "$964K billed in a month" }, { "netuid": 4, "name": "Targon", "function": "Compute inside a sealed enclave.", "number": "550 TAO bought back from revenue" }, { "netuid": 75, "name": "Hippius", "function": "Storage. Egress free.", "number": "100 TB egress, billed $0" }, { "netuid": 68, "name": "NOVA", "function": "Molecules against protein targets.", "number": "4.8M molecules screened" }, { "netuid": 64, "name": "Chutes", "function": "Any open model, in attested hardware.", "number": "2M requests a day" }, { "netuid": 28, "name": "SayGM", "function": "Frontier models below list price.", "number": "96B tokens routed" }, { "netuid": 61, "name": "RedTeam", "function": "Paid to break bot detection.", "number": "Seven-figure revenue" }, { "netuid": 80, "name": "OpenRoboto", "function": "Robot policies that beat the champion.", "number": "50% → 85% on LIBERO-Pro in 21 days" } ] }, "notes": [ "1:30. Jensen's move: \"This is the treasure of our company. Most people talk about the GPU. The GPU is important, but without a programming model that sits on top of it...\" Ours: most people talk about the token. The token matters. But without markets on top of it, nobody would know what to build. Then the row, left to right, one breath a card: name, what it does, one number. Nobody gets more than two sentences.", "Lium: rent a GPU by the second, no account check; nine hundred and sixty-four thousand dollars billed to almost twelve hundred renters last month. [@lium_io, 9 Sep 2026] Targon: compute inside a sealed enclave, so neither the operator nor the machine owner can read your job; five hundred and fifty TAO of its own token bought back from revenue in a month. [@TargonCompute, 20 Aug 2026] Hippius: storage with free egress; a hundred terabytes served in a month, billed zero. [@Hippius_cloud, 8 May 2026] NOVA: molecules against protein targets, results public and on-chain; about five million molecules screened. [taoprotocol.org, 2026; team-stated, VERIFY]", "Chutes: any open model, inside attested hardware; about two million requests a day on a public endpoint you can call yourself. [api.chutes.ai, 7-day average to 22 Sep 2026] SayGM: the labs' own frontier models below list price, through a sealed enclave; ninety-six billion tokens routed so far, and it listed GPT-6 Astra a week before OpenRouter. [@say_gm_, 4 and 7 Sep 2026] RedTeam: a company pays miners to break its bot detection, then sells the hardened result to exchanges; seven-figure revenue by the team's own account. [taodaily.io, 11 Aug 2026; team-stated, VERIFY] OpenRoboto: fine-tune an open robot policy, beat the champion, get paid; fifty to eighty-five percent on a robotics benchmark in twenty-one days. [taodaily.io, Aug 2026; team-stated, VERIFY]", "Then the limits, before anyone else says them: \"Incentive computing is fundamentally a new programming model, and thus has its own restrictions as well as its benefits. We can't approach all historical problems and reframe them as incentive computing. Some things are not measurable. Some things we have tried as incentive systems over the last two years have failed miraculously. We need to reinvent old algorithms and project them as mining systems. Which is in fact why it has taken so long to make the progress we have made. But we will achieve this one domain at a time.\" [Jacob, 23 Sep] This is Jensen's \"if you just did that, it actually runs slower... one domain at a time\", and it is what earns the row.", "Close the row: \"Each one of these platforms frees another economic asset to the Bittensor ecosystem, drives value into the network, and provides a new commodity access point for people. All of these are permissionless, non-KYC, global.\" [Jacob, 23 Sep] Say \"markets\" or \"subnets\", not \"companies\": a company has an address, and the whole talk is that these do not. [JACOB: the dictation says \"global companies\"]", "All eight names and netuids were read from the chain and matched on taostats on 23 Sep 2026; all eight are live and earning. RedTeam sits at about rank 30 of 128 by emission, right at the gate; its story is revenue, not emission. Three of the eight numbers are team statements (NOVA, RedTeam, OpenRoboto); the other five are checkable. Substitutes if a card must go: Conjectures SN66 (a sixteen-year-old case of Green's Problem 51, verified in Lean, 9 Sep 2026; almost no emission), Engy SN53 (Kimi K3 on eighty gaming cards), Teutonic SN3, Affine SN120, KubeTEE SN90 (Pierre Manseau, in the room). [VERIFY every card on 27 Sep; `content/data.json → library`]", "Then three charts that put a year under three of these cards: compute (slide 3), inference (slide 4), data (slide 5); then twelve more markets in one grid (slide 6). Then the pause. Jensen plays a ninety-second video here. We do not. Go to black (slide 7), let it sit, and begin: \"Hello everyone. It's been one year.\" [JACOB: whether the greeting comes here or before slide 1]" ] }, { "id": 3, "title": "Rent, revenue, burn", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Small grey title: \"Lium · SN51 · September 2025 to September 2026\". One chart, no gridlines, no value axis, months along the bottom (Oct 2025, Jan 2026, Apr, Jul, Today). A black line, \"Billed, trailing 30 days\": $423K at the left, a trough in March, $1.20M at the right. A grey line beneath it, \"GPUs online\": 1,061 at the left, a trough in March, 1,458 at the right; each line on its own scale from zero, its value written at both ends. Along the baseline, six thin black bars, \"Bought back and burned from revenue\", each with its dollar amount: $50K (Nov), $665K (Apr), $150K (May), $571K (Jun), $400K (Aug), $966K (9 Sep); the two read from the chain are solid, the four that rest on the team's word are hollow. One sentence beneath: \"Rent → revenue → burn → price → miners add GPUs.\"", "minute": "2:30", "content": { "kind": "growth", "title": "Lium · SN51 · September 2025 to September 2026", "annotation": "Rent → revenue → burn → price → miners add GPUs.", "axis": { "scale": "own", "ticks": [ { "label": "Oct 2025", "at": 0.04155844155844156 }, { "label": "Jan 2026", "at": 0.2805194805194805 }, { "label": "Apr", "at": 0.5142857142857142 }, { "label": "Jul", "at": 0.7506493506493507 }, { "label": "Today", "at": 0.9714285714285714 } ] }, "series": [ { "label": "Billed, trailing 30 days", "tone": "black", "style": "line", "dots": "ends", "format": "usd", "points": [ { "at": 0.07792207792207792, "value": 422781, "verified": true }, { "at": 0.09610389610389611, "value": 415217, "verified": true }, { "at": 0.11428571428571428, "value": 415158, "verified": true }, { "at": 0.13246753246753246, "value": 421667, "verified": true }, { "at": 0.15064935064935064, "value": 419279, "verified": true }, { "at": 0.16883116883116883, "value": 426014, "verified": true }, { "at": 0.18701298701298702, "value": 420213, "verified": true }, { "at": 0.2051948051948052, "value": 390409, "verified": true }, { "at": 0.22337662337662337, "value": 375064, "verified": true }, { "at": 0.24155844155844156, "value": 322637, "verified": true }, { "at": 0.2597402597402597, "value": 277757, "verified": true }, { "at": 0.2779220779220779, "value": 255338, "verified": true }, { "at": 0.2961038961038961, "value": 225605, "verified": true }, { "at": 0.3142857142857143, "value": 221130, "verified": true }, { "at": 0.33246753246753247, "value": 224002, "verified": true }, { "at": 0.35064935064935066, "value": 217296, "verified": true }, { "at": 0.36883116883116884, "value": 213076, "verified": true }, { "at": 0.38701298701298703, "value": 206336, "verified": true }, { "at": 0.4051948051948052, "value": 192963, "verified": true }, { "at": 0.42337662337662335, "value": 182133, "verified": true }, { "at": 0.44155844155844154, "value": 178287, "verified": true }, { "at": 0.4597402597402597, "value": 169463, "verified": true }, { "at": 0.4779220779220779, "value": 166963, "verified": true }, { "at": 0.4961038961038961, "value": 164856, "verified": true }, { "at": 0.5142857142857142, "value": 158728, "verified": true }, { "at": 0.5324675324675324, "value": 159722, "verified": true }, { "at": 0.5506493506493506, "value": 161049, "verified": true }, { "at": 0.5688311688311688, "value": 173618, "verified": true }, { "at": 0.587012987012987, "value": 190572, "verified": true }, { "at": 0.6051948051948052, "value": 221012, "verified": true }, { "at": 0.6233766233766234, "value": 282629, "verified": true }, { "at": 0.6415584415584416, "value": 339576, "verified": true }, { "at": 0.6597402597402597, "value": 352149, "verified": true }, { "at": 0.6779220779220779, "value": 353147, "verified": true }, { "at": 0.6961038961038961, "value": 308896, "verified": true }, { "at": 0.7142857142857143, "value": 253435, "verified": true }, { "at": 0.7324675324675325, "value": 246037, "verified": true }, { "at": 0.7506493506493507, "value": 290876, "verified": true }, { "at": 0.7688311688311689, "value": 371021, "verified": true }, { "at": 0.787012987012987, "value": 446540, "verified": true }, { "at": 0.8051948051948052, "value": 521214, "verified": true }, { "at": 0.8233766233766234, "value": 564108, "verified": true }, { "at": 0.8415584415584415, "value": 642003, "verified": true }, { "at": 0.8597402597402597, "value": 730704, "verified": true }, { "at": 0.8779220779220779, "value": 793174, "verified": true }, { "at": 0.8961038961038961, "value": 856191, "verified": true }, { "at": 0.9142857142857143, "value": 928658, "verified": true }, { "at": 0.9324675324675324, "value": 970790, "verified": true }, { "at": 0.9506493506493506, "value": 1068781, "verified": true }, { "at": 0.9688311688311688, "value": 1202479, "verified": true } ] }, { "label": "GPUs online", "tone": "grey", "style": "line", "dots": "ends", "format": "count", "points": [ { "at": 0.03636363636363636, "value": 1061, "verified": true }, { "at": 0.05454545454545454, "value": 1151, "verified": true }, { "at": 0.07272727272727272, "value": 1000, "verified": true }, { "at": 0.09090909090909091, "value": 824, "verified": true }, { "at": 0.10909090909090909, "value": 953, "verified": true }, { "at": 0.12727272727272726, "value": 1159, "verified": true }, { "at": 0.14545454545454545, "value": 898, "verified": true }, { "at": 0.16363636363636364, "value": 765, "verified": true }, { "at": 0.18181818181818182, "value": 797, "verified": true }, { "at": 0.2, "value": 631, "verified": true }, { "at": 0.21818181818181817, "value": 534, "verified": true }, { "at": 0.22597402597402597, "value": 579, "verified": true }, { "at": 0.24415584415584415, "value": 422, "verified": true }, { "at": 0.2623376623376623, "value": 378, "verified": true }, { "at": 0.2805194805194805, "value": 361, "verified": true }, { "at": 0.2987012987012987, "value": 464, "verified": true }, { "at": 0.3168831168831169, "value": 416, "verified": true }, { "at": 0.33506493506493507, "value": 366, "verified": true }, { "at": 0.35324675324675325, "value": 358, "verified": true }, { "at": 0.37142857142857144, "value": 405, "verified": true }, { "at": 0.38961038961038963, "value": 422, "verified": true }, { "at": 0.4077922077922078, "value": 321, "verified": true }, { "at": 0.42597402597402595, "value": 312, "verified": true }, { "at": 0.44415584415584414, "value": 292, "verified": true }, { "at": 0.4623376623376623, "value": 275, "verified": true }, { "at": 0.4805194805194805, "value": 251, "verified": true }, { "at": 0.4987012987012987, "value": 227, "verified": true }, { "at": 0.5168831168831168, "value": 218, "verified": true }, { "at": 0.535064935064935, "value": 261, "verified": true }, { "at": 0.5532467532467532, "value": 341, "verified": true }, { "at": 0.5714285714285714, "value": 336, "verified": true }, { "at": 0.5896103896103896, "value": 334, "verified": true }, { "at": 0.6077922077922078, "value": 416, "verified": true }, { "at": 0.625974025974026, "value": 472, "verified": true }, { "at": 0.6441558441558441, "value": 423, "verified": true }, { "at": 0.6623376623376623, "value": 371, "verified": true }, { "at": 0.6805194805194805, "value": 414, "verified": true }, { "at": 0.6987012987012987, "value": 480, "verified": true }, { "at": 0.7168831168831169, "value": 508, "verified": true }, { "at": 0.7350649350649351, "value": 636, "verified": true }, { "at": 0.7532467532467533, "value": 650, "verified": true }, { "at": 0.7714285714285715, "value": 842, "verified": true }, { "at": 0.7896103896103897, "value": 934, "verified": true }, { "at": 0.8077922077922078, "value": 1034, "verified": true }, { "at": 0.825974025974026, "value": 1158, "verified": true }, { "at": 0.8441558441558441, "value": 1265, "verified": true }, { "at": 0.8623376623376623, "value": 1178, "verified": true }, { "at": 0.8805194805194805, "value": 1117, "verified": true }, { "at": 0.8987012987012987, "value": 1191, "verified": true }, { "at": 0.9168831168831169, "value": 1300, "verified": true }, { "at": 0.935064935064935, "value": 1366, "verified": true }, { "at": 0.9532467532467532, "value": 1425, "verified": true }, { "at": 0.9714285714285714, "value": 1458, "verified": true } ] }, { "label": "Bought back and burned from revenue", "tone": "black", "style": "bars", "dots": "all", "format": "usd", "points": [ { "at": 0.15324675324675324, "value": 50000, "verified": false, "label": "$50K" }, { "at": 0.5246753246753246, "value": 665000, "verified": true, "label": "$665K" }, { "at": 0.6, "value": 150000, "verified": false, "label": "$150K" }, { "at": 0.7272727272727273, "value": 571000, "verified": false, "label": "$571K" }, { "at": 0.8415584415584415, "value": 400000, "verified": false, "label": "$400K" }, { "at": 0.9324675324675324, "value": 966000, "verified": true, "label": "$966K" } ] } ] }, "notes": [ "2:30. Take the first card and put a year under it. Everything on this screen comes from Lium's own public dashboard and from the chain; nobody at Lium sent me a number. [grafana.lium.io, read 24 Sep 2026; `content/data.json → growthLium`]", "The black line is dollars renters paid for GPU time in the trailing thirty days. Four hundred and twenty thousand a year ago. A hundred and sixty thousand at the bottom, in March. One point two million today. Seven times in six months. Their own month-in-review said nine hundred and sixty-four thousand to the ninth of September; the dashboard says nine hundred and seventy-one thousand for the same thirty days. Same number, two doors. [@lium_io, 9 Sep 2026; Grafana integration]", "Say the dip before anyone else does. The grey line is GPUs online. A thousand last October. Two hundred and twenty in March. I do not have Lium's reason for the winter on the record. What is on the record: in April three rentals in ten failed; by July it was three in a hundred, and the team says renters came back when the nodes stopped failing. Supply followed: fourteen hundred and fifty-eight this morning, twelve hundred and sixty-eight of them rented. Nobody told a provider to plug in a card. The price told them. [@lium_io, 16 Jul 2026; lium.io/api/executors/stats, 24 Sep 2026]", "The bars are what the team bought back and burned with that revenue. Ten thousand a day when the program started in November. Six hundred and sixty-five thousand in April. Nine hundred and sixty-six thousand on the ninth of September: three thousand seven hundred and thirty-eight TAO in one transaction, block 9,032,919. Check it. The hollow bars are the team's word; the solid ones are on-chain. [taomarketcap.com/extrinsics/9032919-0005; 7898274-0011 and 7898314-0023]", "Read the sentence under the chart. Rent, revenue, burn, price, miners add GPUs. Lium's token, priced in TAO, went from 0.041 in April to 0.089 this week, and the GPUs came with it. \"Let the flywheel commence. This is how bittensor succeeds.\" [@fish_datura, 7 Nov 2025] That is the protocol doing what it was designed to do. I did not do it. Fish did not do it. The mechanism did. [archive node, SubnetTAO ÷ SubnetAlphaIn, netuid 51]", "Do not say \"up only\". It fell by half over the winter and came back seven times. Say the three honest things: billing is gross and providers take most of it; part of the demand is other subnets' miners paying with emissions, Fish says so himself; and the protocol still pays SN51 more in emissions than renters pay Lium. The flywheel is real. It is not yet self-funding. [VERIFY on 27 Sep: re-run the Grafana queries in `docs/growth-slides.md`; the last GPU point is live]", "Advance on \"tokens\"." ] }, { "id": 4, "title": "Tokens a day", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Small grey title: \"Tokens a day · Chutes SN64, Engy SN53, SayGM SN28 · Feb 2025 to Sep 2026\". One chart on one linear scale from zero, months along the bottom (Apr 2025, Jul, Oct, Jan 2026, Apr, Jul, Today). A black line, \"All three · 41B\": Chutes alone until 2026, a peak ringed and labelled \"151B at the peak\" in June 2025, a long slope down, then a floor that turns up in August 2026; the line is dashed from June 2026 because it includes an approximation. Three grey lines beneath: \"Chutes · 26B\" (the same shape), \"Engy · 12B\" (from nothing on 29 July 2026, straight up), \"SayGM (approx.) · 2.5B\" (four hollow dots from June). One sentence beneath: \"Two of the three did not exist in April; the third cut its free tier in March.\"", "minute": "3:30", "content": { "kind": "growth", "title": "Tokens a day · Chutes SN64, Engy SN53, SayGM SN28 · Feb 2025 to Sep 2026", "annotation": "Two of the three did not exist in April; the third cut its free tier in March.", "axis": { "scale": "shared", "ticks": [ { "label": "Apr 2025", "at": 0.09656301145662848 }, { "label": "Jul", "at": 0.24549918166939444 }, { "label": "Oct", "at": 0.3960720130932897 }, { "label": "Jan 2026", "at": 0.546644844517185 }, { "label": "Apr", "at": 0.6939443535188216 }, { "label": "Jul", "at": 0.8428805237315876 }, { "label": "Today", "at": 0.9803600654664485 } ] }, "series": [ { "label": "All three", "tone": "black", "style": "line", "dots": "ends", "format": "tokens", "points": [ { "at": 0.01800327332242226, "value": 1072409918, "verified": true }, { "at": 0.029459901800327332, "value": 3042559534, "verified": true }, { "at": 0.04091653027823241, "value": 5389877312, "verified": true }, { "at": 0.05237315875613748, "value": 7257334463, "verified": true }, { "at": 0.06382978723404255, "value": 10678680822, "verified": true }, { "at": 0.07528641571194762, "value": 13154388979, "verified": true }, { "at": 0.0867430441898527, "value": 16511100800, "verified": true }, { "at": 0.09819967266775777, "value": 21293235576, "verified": true }, { "at": 0.10965630114566285, "value": 25640362918, "verified": true }, { "at": 0.12111292962356793, "value": 32698579117, "verified": true }, { "at": 0.132569558101473, "value": 42745268695, "verified": true }, { "at": 0.14402618657937807, "value": 52006094690, "verified": true }, { "at": 0.15548281505728315, "value": 58165550315, "verified": true }, { "at": 0.16693944353518822, "value": 61871483786, "verified": true }, { "at": 0.1783960720130933, "value": 78107464941, "verified": true }, { "at": 0.18985270049099837, "value": 95682370650, "verified": true }, { "at": 0.20130932896890344, "value": 111455544163, "verified": true }, { "at": 0.2127659574468085, "value": 129812675586, "verified": true }, { "at": 0.22422258592471359, "value": 143231040861, "verified": true }, { "at": 0.23567921440261866, "value": 151396844899, "verified": true }, { "at": 0.24713584288052373, "value": 133427404699, "verified": true }, { "at": 0.25859247135842883, "value": 116325508659, "verified": true }, { "at": 0.2700490998363339, "value": 88199659819, "verified": true }, { "at": 0.281505728314239, "value": 101902642258, "verified": true }, { "at": 0.29296235679214405, "value": 115210515262, "verified": true }, { "at": 0.3044189852700491, "value": 130529512174, "verified": true }, { "at": 0.3158756137479542, "value": 131122889895, "verified": true }, { "at": 0.32733224222585927, "value": 145323478148, "verified": true }, { "at": 0.33878887070376434, "value": 142614485400, "verified": true }, { "at": 0.3502454991816694, "value": 126896916123, "verified": true }, { "at": 0.3617021276595745, "value": 121123732258, "verified": true }, { "at": 0.37315875613747956, "value": 126385326517, "verified": true }, { "at": 0.38461538461538464, "value": 101942042128, "verified": true }, { "at": 0.3960720130932897, "value": 99625475548, "verified": true }, { "at": 0.4075286415711948, "value": 106066015300, "verified": true }, { "at": 0.41898527004909986, "value": 103917158903, "verified": true }, { "at": 0.43044189852700493, "value": 108138321092, "verified": true }, { "at": 0.44189852700491, "value": 104973975521, "verified": true }, { "at": 0.4533551554828151, "value": 98735151673, "verified": true }, { "at": 0.46481178396072015, "value": 103674483722, "verified": true }, { "at": 0.4762684124386252, "value": 105040794890, "verified": true }, { "at": 0.4877250409165303, "value": 104188330443, "verified": true }, { "at": 0.49918166939443537, "value": 104326014980, "verified": true }, { "at": 0.5106382978723404, "value": 102200293236, "verified": true }, { "at": 0.5220949263502455, "value": 100818231918, "verified": true }, { "at": 0.5335515548281505, "value": 105377372336, "verified": true }, { "at": 0.5450081833060556, "value": 105775090685, "verified": true }, { "at": 0.5564648117839607, "value": 107197376452, "verified": true }, { "at": 0.5679214402618658, "value": 106334160577, "verified": true }, { "at": 0.5793780687397708, "value": 107707881322, "verified": true }, { "at": 0.5908346972176759, "value": 107137856722, "verified": true }, { "at": 0.602291325695581, "value": 115205266002, "verified": true }, { "at": 0.613747954173486, "value": 117842968307, "verified": true }, { "at": 0.6252045826513911, "value": 98108286488, "verified": true }, { "at": 0.6366612111292962, "value": 86533148271, "verified": true }, { "at": 0.6481178396072013, "value": 66611350448, "verified": true }, { "at": 0.6595744680851063, "value": 57461768543, "verified": true }, { "at": 0.6710310965630114, "value": 56185361008, "verified": true }, { "at": 0.6824877250409165, "value": 72058290862, "verified": true }, { "at": 0.6939443535188216, "value": 87180239727, "verified": true }, { "at": 0.7054009819967266, "value": 84346706854, "verified": true }, { "at": 0.7168576104746317, "value": 87285030952, "verified": true }, { "at": 0.7283142389525368, "value": 66063033519, "verified": true }, { "at": 0.7397708674304418, "value": 56208378095, "verified": true }, { "at": 0.7512274959083469, "value": 51433992812, "verified": true }, { "at": 0.762684124386252, "value": 54900528484, "verified": true }, { "at": 0.7741407528641571, "value": 57806977355, "verified": true }, { "at": 0.7855973813420621, "value": 47753692907, "verified": true }, { "at": 0.7970540098199672, "value": 54332810776, "verified": false }, { "at": 0.8085106382978723, "value": 47281983624, "verified": false }, { "at": 0.8199672667757774, "value": 45456223823, "verified": false }, { "at": 0.8314238952536824, "value": 42196974240, "verified": false }, { "at": 0.8428805237315876, "value": 39805935646, "verified": false }, { "at": 0.8543371522094927, "value": 43433595611, "verified": false }, { "at": 0.8657937806873978, "value": 37614009358, "verified": false }, { "at": 0.8772504091653028, "value": 49286588290, "verified": false }, { "at": 0.8887070376432079, "value": 46917442250, "verified": false }, { "at": 0.900163666121113, "value": 46384842614, "verified": false }, { "at": 0.911620294599018, "value": 45418002613, "verified": false }, { "at": 0.9230769230769231, "value": 47326516359, "verified": false }, { "at": 0.9345335515548282, "value": 54869423380, "verified": false }, { "at": 0.9459901800327333, "value": 58459121087, "verified": false }, { "at": 0.9574468085106383, "value": 45421729959, "verified": false }, { "at": 0.9689034369885434, "value": 41260634284, "verified": false }, { "at": 0.9803600654664485, "value": 41038907333, "verified": false } ] }, { "label": "Chutes", "tone": "grey", "style": "line", "dots": "ends", "format": "tokens", "points": [ { "at": 0.01800327332242226, "value": 1072409918, "verified": true }, { "at": 0.029459901800327332, "value": 3042559534, "verified": true }, { "at": 0.04091653027823241, "value": 5389877312, "verified": true }, { "at": 0.05237315875613748, "value": 7257334463, "verified": true }, { "at": 0.06382978723404255, "value": 10678680822, "verified": true }, { "at": 0.07528641571194762, "value": 13154388979, "verified": true }, { "at": 0.0867430441898527, "value": 16511100800, "verified": true }, { "at": 0.09819967266775777, "value": 21293235576, "verified": true }, { "at": 0.10965630114566285, "value": 25640362918, "verified": true }, { "at": 0.12111292962356793, "value": 32698579117, "verified": true }, { "at": 0.132569558101473, "value": 42745268695, "verified": true }, { "at": 0.14402618657937807, "value": 52006094690, "verified": true }, { "at": 0.15548281505728315, "value": 58165550315, "verified": true }, { "at": 0.16693944353518822, "value": 61871483786, "verified": true }, { "at": 0.1783960720130933, "value": 78107464941, "verified": true }, { "at": 0.18985270049099837, "value": 95682370650, "verified": true }, { "at": 0.20130932896890344, "value": 111455544163, "verified": true }, { "at": 0.2127659574468085, "value": 129812675586, "verified": true }, { "at": 0.22422258592471359, "value": 143231040861, "verified": true }, { "at": 0.23567921440261866, "value": 151396844899, "verified": true }, { "at": 0.24713584288052373, "value": 133427404699, "verified": true }, { "at": 0.25859247135842883, "value": 116325508659, "verified": true }, { "at": 0.2700490998363339, "value": 88199659819, "verified": true }, { "at": 0.281505728314239, "value": 101902642258, "verified": true }, { "at": 0.29296235679214405, "value": 115210515262, "verified": true }, { "at": 0.3044189852700491, "value": 130529512174, "verified": true }, { "at": 0.3158756137479542, "value": 131122889895, "verified": true }, { "at": 0.32733224222585927, "value": 145323478148, "verified": true }, { "at": 0.33878887070376434, "value": 142614485400, "verified": true }, { "at": 0.3502454991816694, "value": 126896916123, "verified": true }, { "at": 0.3617021276595745, "value": 121123732258, "verified": true }, { "at": 0.37315875613747956, "value": 126385326517, "verified": true }, { "at": 0.38461538461538464, "value": 101942042128, "verified": true }, { "at": 0.3960720130932897, "value": 99625475548, "verified": true }, { "at": 0.4075286415711948, "value": 106066015300, "verified": true }, { "at": 0.41898527004909986, "value": 103917158903, "verified": true }, { "at": 0.43044189852700493, "value": 108138321092, "verified": true }, { "at": 0.44189852700491, "value": 104973975521, "verified": true }, { "at": 0.4533551554828151, "value": 98735151673, "verified": true }, { "at": 0.46481178396072015, "value": 103674483722, "verified": true }, { "at": 0.4762684124386252, "value": 105040794890, "verified": true }, { "at": 0.4877250409165303, "value": 104188330443, "verified": true }, { "at": 0.49918166939443537, "value": 104326014980, "verified": true }, { "at": 0.5106382978723404, "value": 102200293236, "verified": true }, { "at": 0.5220949263502455, "value": 100818231918, "verified": true }, { "at": 0.5335515548281505, "value": 105377372336, "verified": true }, { "at": 0.5450081833060556, "value": 105775090685, "verified": true }, { "at": 0.5564648117839607, "value": 107197376452, "verified": true }, { "at": 0.5679214402618658, "value": 106334160577, "verified": true }, { "at": 0.5793780687397708, "value": 107707881322, "verified": true }, { "at": 0.5908346972176759, "value": 107137856722, "verified": true }, { "at": 0.602291325695581, "value": 115205266002, "verified": true }, { "at": 0.613747954173486, "value": 117842968307, "verified": true }, { "at": 0.6252045826513911, "value": 98108286488, "verified": true }, { "at": 0.6366612111292962, "value": 86533148271, "verified": true }, { "at": 0.6481178396072013, "value": 66611350448, "verified": true }, { "at": 0.6595744680851063, "value": 57461768543, "verified": true }, { "at": 0.6710310965630114, "value": 56185361008, "verified": true }, { "at": 0.6824877250409165, "value": 72058290862, "verified": true }, { "at": 0.6939443535188216, "value": 87180239727, "verified": true }, { "at": 0.7054009819967266, "value": 84346706854, "verified": true }, { "at": 0.7168576104746317, "value": 87285030952, "verified": true }, { "at": 0.7283142389525368, "value": 66063033519, "verified": true }, { "at": 0.7397708674304418, "value": 56208378095, "verified": true }, { "at": 0.7512274959083469, "value": 51433992812, "verified": true }, { "at": 0.762684124386252, "value": 54900528484, "verified": true }, { "at": 0.7741407528641571, "value": 57806977355, "verified": true }, { "at": 0.7855973813420621, "value": 47753692907, "verified": true }, { "at": 0.7970540098199672, "value": 54332810776, "verified": true }, { "at": 0.8085106382978723, "value": 47281983624, "verified": true }, { "at": 0.8199672667757774, "value": 45340223823, "verified": true }, { "at": 0.8314238952536824, "value": 42024974240, "verified": true }, { "at": 0.8428805237315876, "value": 39577935646, "verified": true }, { "at": 0.8543371522094927, "value": 43149595611, "verified": true }, { "at": 0.8657937806873978, "value": 37274009358, "verified": true }, { "at": 0.8772504091653028, "value": 48526588290, "verified": true }, { "at": 0.8887070376432079, "value": 44866309513, "verified": true }, { "at": 0.900163666121113, "value": 43328246595, "verified": true }, { "at": 0.911620294599018, "value": 42545104745, "verified": true }, { "at": 0.9230769230769231, "value": 42838191552, "verified": true }, { "at": 0.9345335515548282, "value": 47253920603, "verified": true }, { "at": 0.9459901800327333, "value": 48440301639, "verified": true }, { "at": 0.9574468085106383, "value": 35028871445, "verified": true }, { "at": 0.9689034369885434, "value": 29313827330, "verified": true }, { "at": 0.9803600654664485, "value": 26442920510, "verified": true } ] }, { "label": "Engy", "tone": "grey", "style": "line", "dots": "ends", "format": "tokens", "points": [ { "at": 0.8887070376432079, "value": 871132737, "verified": true }, { "at": 0.900163666121113, "value": 1456596019, "verified": true }, { "at": 0.911620294599018, "value": 852897868, "verified": true }, { "at": 0.9230769230769231, "value": 2257555576, "verified": true }, { "at": 0.9345335515548282, "value": 5330887392, "verified": true }, { "at": 0.9459901800327333, "value": 7680357910, "verified": true }, { "at": 0.9574468085106383, "value": 8000550822, "verified": true }, { "at": 0.9689034369885434, "value": 9500653108, "verified": true }, { "at": 0.9803600654664485, "value": 12095986823, "verified": true } ] }, { "label": "SayGM (approx.)", "tone": "grey", "style": "line", "dots": "all", "format": "tokens", "points": [ { "at": 0.8166939443535188, "value": 100000000, "verified": false }, { "at": 0.8657937806873978, "value": 340000000, "verified": false }, { "at": 0.9165302782324058, "value": 2200000000, "verified": false }, { "at": 0.9803600654664485, "value": 2500000000, "verified": false } ] } ] }, "notes": [ "3:30. Now inference. The question I get asked is: how many tokens are you actually serving? Here is the honest answer, with the method on the slide. About forty-one billion tokens a day across three inference subnets, in the week to the twenty-third of September. Twenty-six billion on Chutes, from its own public research page. Twelve billion on Engy, from a ledger its validators sign every week. About two and a half billion on SayGM, and that one is an estimate; I will tell you how in a second. Three and a quarter million requests a day between them. [chutes.ai/app/research, 24 Sep 2026; provider.engy.ai epoch 14; saygm.com/miners; `content/data.json → growthInference`]", "Say the peak first. That ring is June 2025: a hundred and fifty-one billion tokens a day, Chutes alone, almost all of it free. In March the team cut the free tier. Tokens fell by nearly half and revenue fell by a quarter, by their own account, and revenue per trillion tokens hit a record in July. The line went down because they stopped giving it away. I would rather show you that than hide it. [chutes.ai/app/research; @chutes_ai \"Week 10\", 21 Jul 2026]", "Now the two lines that did not exist in April. Engy launched on the twenty-first of July. Its first week served nine hundred million tokens a day. Last week, twelve billion a day, seven million requests, eighty-two miners, thirty-five thousand dollars settled to them, mostly on gaming cards. Fourteen times in nine weeks. Every number is in a signed ledger you can pull. SayGM went live in May and reported ninety-six billion tokens on the seventh of September; the dots are hollow because SayGM publishes requests and dollars per epoch, not tokens, so I multiplied requests by the tokens per request its own model pages show. [provider.engy.ai/api/subnet/v1/epoch/14; @say_gm_, 7 Sep 2026]", "The comparison, so nobody has to make it for me. OpenRouter's ten busiest models alone did eleven point eight trillion tokens on the twenty-third of September. We are a rounding error to OpenRouter. Fifteen months ago Chutes alone did a hundred and fifty billion a day, free. It stopped giving it away. What is on this screen now is paid, or on its way to paid. [openrouter.ai/rankings, usage through 23 Sep 2026]", "One line for the room: the volume that left was free. The volume that came is paid, on hardware nobody licensed, with a receipt for every answer.", "Do not say \"we serve a hundred and twenty billion tokens a day\". That was mid-2025. Do not add SayGM's number without the word \"about\". [VERIFY on 27 Sep: re-pull chutes.ai/app/research, Engy epoch 15 (ends 30 Sep 08:00 UTC), and the saygm.com ledger; method in `docs/growth-slides.md`]", "Advance on \"paid\". The data layer is next." ] }, { "id": 5, "title": "Data", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Small grey title: \"Data · Hippius SN75 · NOVA SN68 · Data Universe SN13 · Beam SN105 · to 24 Sep 2026\". Four panels in a row, each with a hairline across the top, its own scale, and nothing shared but the date: grey mono \"Hippius · SN75\", big \"730 TB\", \"bytes stored on nodes\", a black line rising from a zero hairline with a dot per point, \"44 TB · Apr 2026\" left and \"24 Sep 2026\" right beneath / \"NOVA · SN68\", \"2.7M\", \"molecules in the open archive\", line from May 2025, \"38K · May 2025\" / \"Data Universe · SN13\", \"55B+\", \"rows collected\", line from Aug 2024 with a hollow last dot, \"1.3B · Aug 2024\" / \"Beam · SN105\", \"1.12 TB\", \"largest single transfer\", line from Apr 2026, \"500 MB · Apr 2026\". One sentence beneath: \"Tonight: 730 terabytes stored, 2.7 million molecules in the open, 55 billion rows, a terabyte moved in three minutes forty.\" On a phone the row becomes two by two.", "minute": "4:30", "content": { "kind": "panels", "title": "Data · Hippius SN75 · NOVA SN68 · Data Universe SN13 · Beam SN105 · to 24 Sep 2026", "annotation": "Tonight: 730 terabytes stored, 2.7 million molecules in the open, 55 billion rows, a terabyte moved in three minutes forty.", "panels": [ { "name": "Hippius", "netuid": 75, "measure": "bytes stored on nodes", "latest": "730 TB", "latestDate": "24 Sep 2026", "start": "44 TB · Apr 2026", "points": [ { "at": 0, "value": 43.6, "verified": true }, { "at": 0.17045454545454544, "value": 72.8, "verified": true }, { "at": 0.3465909090909091, "value": 141.8, "verified": true }, { "at": 0.5170454545454546, "value": 241.4, "verified": true }, { "at": 0.6931818181818182, "value": 318.3, "verified": true }, { "at": 0.8693181818181818, "value": 343.4, "verified": true }, { "at": 0.9488636363636364, "value": 455.6, "verified": true }, { "at": 1, "value": 729.8, "verified": true } ] }, { "name": "NOVA", "netuid": 68, "measure": "molecules in the open archive", "latest": "2.7M", "latestDate": "24 Sep 2026", "start": "38K · May 2025", "points": [ { "at": 0, "value": 38427, "verified": true }, { "at": 0.11764705882352941, "value": 261173, "verified": true }, { "at": 0.23921568627450981, "value": 552748, "verified": true }, { "at": 0.2980392156862745, "value": 865232, "verified": true }, { "at": 0.3588235294117647, "value": 1786920, "verified": true }, { "at": 0.47843137254901963, "value": 2066403, "verified": true }, { "at": 0.6549019607843137, "value": 2126179, "verified": true }, { "at": 0.8333333333333334, "value": 2198351, "verified": true }, { "at": 0.8941176470588236, "value": 2222850, "verified": true }, { "at": 0.9549019607843138, "value": 2467915, "verified": true }, { "at": 1, "value": 2675535, "verified": true } ] }, { "name": "Data Universe", "netuid": 13, "measure": "rows collected", "latest": "55B+", "latestDate": "24 Sep 2026", "start": "1.3B · Aug 2024", "points": [ { "at": 0, "value": 1340000000, "verified": true }, { "at": 0.08045977011494253, "value": 18000000000, "verified": true }, { "at": 0.22860791826309068, "value": 27700000000, "verified": true }, { "at": 0.3128991060025543, "value": 40000000000, "verified": true }, { "at": 1, "value": 55000000000, "verified": false } ] }, { "name": "Beam", "netuid": 105, "measure": "largest single transfer", "latest": "1.12 TB", "latestDate": "18 Jul 2026", "start": "500 MB · Apr 2026", "points": [ { "at": 0, "value": 0.5, "verified": true }, { "at": 0.5887850467289719, "value": 10, "verified": true }, { "at": 0.8037383177570093, "value": 50, "verified": true }, { "at": 0.9439252336448598, "value": 107, "verified": true }, { "at": 1, "value": 1120, "verified": true } ] } ] }, "notes": [ "4:30. Third of the growth trio: compute, inference, data. One breath a panel, left to right. The data layer is what a mind eats and where it keeps what it learns. Four markets, four different units, so four panels, each on its own scale; no shared axis, because a byte is not a molecule is not a row.", "Hippius. Seven hundred and thirty terabytes on three hundred and eleven nodes tonight. On the first of April it held forty-four. Egress is free, and every byte is on a public explorer with a block number. [hipstats.com, 24 Sep 2026; `content/data.json → growthData`]", "NOVA. Two point seven million molecules in an archive anyone can download from Hugging Face, published one epoch at a time since May last year; the team counts eleven million submitted. Since August the first ones are in a wet lab. Micaela Bazo speaks at 11:20 this morning; hand it to her. [huggingface.co/datasets/Metanova/Submission-Archive; @metanova_labs, 26 Aug 2026]", "Data Universe. Fifty-five billion rows of social data, once the largest open social dataset on Hugging Face. Say the honest half in the same breath: they took it private last October, the emission gate cut them to almost nothing, and the flow is a trickle of what it was. The stock exists; the flow does not. [@MacrocosmosAI, 2024–2025; datauniverse.macrocosmos.ai]", "Beam. Six months old. Five hundred megabytes in April; a terabyte in three minutes forty in July; about a terabyte every half hour, all September. Every one of those transfers was Beam's own test load. Say \"proving ground\", not \"business\": no customer yet, by their own account. [@b1m_ai, 2 Apr to 13 Sep 2026]", "Caveats, before a critic says them. Hippius's bytes include Arion parity shards, so it is raw disk, not customer files, and it doubled in the last nine days for a reason the team has not published. NOVA's archive is a lower bound; the eleven million is the team's number. Data Universe's fifty-five billion is undated on their site and their own July 2025 post implies more; the line ends hollow for that reason. Beam's bytes are self-generated. [VERIFY: re-pull all four on 27 Sep from the sources in `content/data.json → growthData`]", "Not drawn, to keep it to four: Minos SN107, a synthetic genome every seventy-two minutes, thirty-seven thousand variant-calling evaluations in eleven days in June, stored on Hippius. Mamad Anghari speaks on Day 2. Say it if the room is biotech-heavy.", "Do not say \"up only\". Say: \"Four markets for data, each with a door open tonight and a number you can check.\" Advance on \"check\"." ] }, { "id": 6, "title": "Twelve markets", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Small grey title: \"Twelve markets · to 24 Sep 2026\". Twelve panels in a grid, four across and three down, each with a hairline across the top and its own scale, in the form of the Data slide: grey mono \"Bitcast · SN93\", big \"$63.1K\", \"brand revenue a month\", a black line from a zero hairline, \"$2.1K · May 2025\" left and \"Aug 2026\" right / \"Engy · SN53\", \"$34.9K\", \"settled to miners a week\", \"$682 · Jul 2026\" … \"23 Sep 2026\" / \"gm · SN28\", \"$13K\", \"traffic routed a day\", \"$359 · Jun 2026\" … \"24 Sep 2026\" / \"Almanac · SN41\", \"$40.7K\", \"terminal fees a month\", \"$1.5K · Feb 2026\" … \"Sep 2026\" / \"Gradients · SN56\", \"565\", \"models published a month\", \"4 · Nov 2025\" … \"Aug 2026\" / \"Vanta · SN8\", \"$1.46B\", \"paper volume, 30 days\", \"$55.9M · Apr 2026\" … \"23 Sep 2026\" / \"UR · SN25\", \"206K\", \"devices connected\", \"120K · Jul 2026\" … \"23 Sep 2026\" / \"Hippius · SN75\", \"200 TB\", \"S3 egress a month\", \"16 TB · Mar 2026\" … \"Sep 2026\" / \"OpenRoboto · SN80\", \"90%\", \"LIBERO-Pro champion, sim\", \"50% · Jul 2026\" … \"28 Aug 2026\" / \"Conjectures · SN66\", \"31\", \"Lean-verified proofs\", \"1 · Aug 2026\" … \"21 Sep 2026\" / \"TAO · chain\", \"386K\", \"accounts holding TAO\", \"14.1K · Mar 2023\" … \"24 Sep 2026\" / \"IOTA · SN9\", \"45.9K\", \"Train-at-Home downloads\", two hollow dots, \"2.1K · Feb 2026\" … \"24 Sep 2026\". One sentence beneath: \"Twelve markets, twelve teams, and not one of these lines is mine to draw.\" Small grey footnote: \"Sim and paper: OpenRoboto scores are MuJoCo simulation; Vanta volume is paper positions scored by validators, not exchange trades. Hollow points are team posts, not read from a ledger (IOTA, Feb and Apr 2026).\" On a phone the grid is two across and six down.", "minute": "5:30", "content": { "kind": "panels", "grid": true, "title": "Twelve markets · to 24 Sep 2026", "annotation": "Twelve markets, twelve teams, and not one of these lines is mine to draw.", "footnote": "Sim and paper: OpenRoboto scores are MuJoCo simulation; Vanta volume is paper positions scored by validators, not exchange trades. Hollow points are team posts, not read from a ledger (IOTA, Feb and Apr 2026).", "panels": [ { "name": "Bitcast", "netuid": 93, "measure": "brand revenue a month", "latest": "$63.1K", "latestDate": "Aug 2026", "start": "$2.1K · May 2025", "points": [ { "at": 0, "value": 2093.18, "verified": true }, { "at": 0.06783369803063458, "value": 2125.12, "verified": true }, { "at": 0.13347921225382933, "value": 20829.4, "verified": true }, { "at": 0.2013129102844639, "value": 14547.34, "verified": true }, { "at": 0.26914660831509846, "value": 13040.86, "verified": true }, { "at": 0.3347921225382932, "value": 30039.28, "verified": true }, { "at": 0.4026258205689278, "value": 18984.7, "verified": true }, { "at": 0.4682713347921225, "value": 13088.98, "verified": true }, { "at": 0.5361050328227571, "value": 23006.75, "verified": true }, { "at": 0.6039387308533917, "value": 13365.27, "verified": true }, { "at": 0.6652078774617067, "value": 12132.92, "verified": true }, { "at": 0.7330415754923414, "value": 34610.45, "verified": true }, { "at": 0.7986870897155361, "value": 28826.18, "verified": true }, { "at": 0.8665207877461707, "value": 42062.79, "verified": true }, { "at": 0.9321663019693655, "value": 47966.34, "verified": true }, { "at": 1, "value": 63072.5, "verified": true } ] }, { "name": "Engy", "netuid": 53, "measure": "settled to miners a week", "latest": "$34.9K", "latestDate": "23 Sep 2026", "start": "$682 · Jul 2026", "points": [ { "at": 0, "value": 682.44, "verified": true }, { "at": 0.125, "value": 945.14, "verified": true }, { "at": 0.25, "value": 350.28, "verified": true }, { "at": 0.375, "value": 3038.78, "verified": true }, { "at": 0.5, "value": 4466.58, "verified": true }, { "at": 0.625, "value": 15712.24, "verified": true }, { "at": 0.75, "value": 14317.83, "verified": true }, { "at": 0.875, "value": 12726.1, "verified": true }, { "at": 1, "value": 34870.18, "verified": true } ] }, { "name": "gm", "netuid": 28, "measure": "traffic routed a day", "latest": "$13K", "latestDate": "24 Sep 2026", "start": "$359 · Jun 2026", "points": [ { "at": 0, "value": 359, "verified": true }, { "at": 0.2608695652173913, "value": 2111, "verified": true }, { "at": 0.5304347826086957, "value": 7836, "verified": true }, { "at": 1, "value": 12976, "verified": true } ] }, { "name": "Almanac", "netuid": 41, "measure": "terminal fees a month", "latest": "$40.7K", "latestDate": "Sep 2026", "start": "$1.5K · Feb 2026", "points": [ { "at": 0, "value": 1489.69, "verified": true }, { "at": 0.1320754716981132, "value": 8204.4, "verified": true }, { "at": 0.2783018867924528, "value": 5873.41, "verified": true }, { "at": 0.419811320754717, "value": 16618.2, "verified": true }, { "at": 0.5660377358490566, "value": 5748.83, "verified": true }, { "at": 0.7075471698113207, "value": 23001.04, "verified": true }, { "at": 0.8537735849056604, "value": 15276.73, "verified": true }, { "at": 1, "value": 40689.01, "verified": true } ] }, { "name": "Gradients", "netuid": 56, "measure": "models published a month", "latest": "565", "latestDate": "Aug 2026", "start": "4 · Nov 2025", "points": [ { "at": 0, "value": 4, "verified": true }, { "at": 0.10989010989010989, "value": 8, "verified": true }, { "at": 0.22344322344322345, "value": 3, "verified": true }, { "at": 0.336996336996337, "value": 6, "verified": true }, { "at": 0.43956043956043955, "value": 5, "verified": true }, { "at": 0.5531135531135531, "value": 32, "verified": true }, { "at": 0.663003663003663, "value": 45, "verified": true }, { "at": 0.7765567765567766, "value": 63, "verified": true }, { "at": 0.8864468864468864, "value": 132, "verified": true }, { "at": 1, "value": 565, "verified": true } ] }, { "name": "Vanta", "netuid": 8, "measure": "paper volume, 30 days", "latest": "$1.46B", "latestDate": "23 Sep 2026", "start": "$55.9M · Apr 2026", "points": [ { "at": 0, "value": 55898119, "verified": true }, { "at": 0.04, "value": 47650889, "verified": true }, { "at": 0.08, "value": 45874380, "verified": true }, { "at": 0.12, "value": 114587516, "verified": true }, { "at": 0.16, "value": 95605271, "verified": true }, { "at": 0.2, "value": 105228151, "verified": true }, { "at": 0.24, "value": 155196090, "verified": true }, { "at": 0.28, "value": 159824580, "verified": true }, { "at": 0.32, "value": 157478617, "verified": true }, { "at": 0.36, "value": 220685620, "verified": true }, { "at": 0.4, "value": 204977139, "verified": true }, { "at": 0.44, "value": 178400304, "verified": true }, { "at": 0.48, "value": 178965592, "verified": true }, { "at": 0.52, "value": 207436686, "verified": true }, { "at": 0.56, "value": 278335660, "verified": true }, { "at": 0.6, "value": 350632227, "verified": true }, { "at": 0.64, "value": 386965068, "verified": true }, { "at": 0.68, "value": 343796236, "verified": true }, { "at": 0.72, "value": 291946884, "verified": true }, { "at": 0.76, "value": 257971116, "verified": true }, { "at": 0.8, "value": 308813091, "verified": true }, { "at": 0.84, "value": 493380011, "verified": true }, { "at": 0.88, "value": 727441453, "verified": true }, { "at": 0.92, "value": 917923386, "verified": true }, { "at": 0.96, "value": 1042056637, "verified": true }, { "at": 1, "value": 1456339372, "verified": true } ] }, { "name": "UR", "netuid": 25, "measure": "devices connected", "latest": "206K", "latestDate": "23 Sep 2026", "start": "120K · Jul 2026", "points": [ { "at": 0, "value": 120484, "verified": true }, { "at": 0.015625, "value": 127807, "verified": true }, { "at": 0.03125, "value": 129636, "verified": true }, { "at": 0.046875, "value": 117560, "verified": true }, { "at": 0.0625, "value": 120419, "verified": true }, { "at": 0.078125, "value": 123925, "verified": true }, { "at": 0.09375, "value": 139391, "verified": true }, { "at": 0.109375, "value": 137627, "verified": true }, { "at": 0.125, "value": 128620, "verified": true }, { "at": 0.140625, "value": 135244, "verified": true }, { "at": 0.15625, "value": 126190, "verified": true }, { "at": 0.171875, "value": 107256, "verified": true }, { "at": 0.1875, "value": 119012, "verified": true }, { "at": 0.203125, "value": 113095, "verified": true }, { "at": 0.21875, "value": 142556, "verified": true }, { "at": 0.234375, "value": 141382, "verified": true }, { "at": 0.25, "value": 137949, "verified": true }, { "at": 0.265625, "value": 162776, "verified": true }, { "at": 0.28125, "value": 159854, "verified": true }, { "at": 0.296875, "value": 130561, "verified": true }, { "at": 0.328125, "value": 131178, "verified": true }, { "at": 0.34375, "value": 141937, "verified": true }, { "at": 0.359375, "value": 144663, "verified": true }, { "at": 0.375, "value": 145861, "verified": true }, { "at": 0.390625, "value": 138031, "verified": true }, { "at": 0.40625, "value": 144166, "verified": true }, { "at": 0.421875, "value": 109384, "verified": true }, { "at": 0.4375, "value": 144660, "verified": true }, { "at": 0.453125, "value": 152755, "verified": true }, { "at": 0.484375, "value": 142040, "verified": true }, { "at": 0.5, "value": 140467, "verified": true }, { "at": 0.515625, "value": 138135, "verified": true }, { "at": 0.53125, "value": 141851, "verified": true }, { "at": 0.59375, "value": 132015, "verified": true }, { "at": 0.609375, "value": 142990, "verified": true }, { "at": 0.625, "value": 166805, "verified": true }, { "at": 0.640625, "value": 165879, "verified": true }, { "at": 0.65625, "value": 182345, "verified": true }, { "at": 0.671875, "value": 144414, "verified": true }, { "at": 0.6875, "value": 158599, "verified": true }, { "at": 0.703125, "value": 183931, "verified": true }, { "at": 0.71875, "value": 171917, "verified": true }, { "at": 0.734375, "value": 174378, "verified": true }, { "at": 0.75, "value": 165427, "verified": true }, { "at": 0.765625, "value": 194311, "verified": true }, { "at": 0.78125, "value": 149132, "verified": true }, { "at": 0.796875, "value": 150884, "verified": true }, { "at": 0.8125, "value": 217045, "verified": true }, { "at": 0.828125, "value": 203996, "verified": true }, { "at": 0.84375, "value": 206146, "verified": true }, { "at": 0.859375, "value": 216633, "verified": true }, { "at": 0.875, "value": 227481, "verified": true }, { "at": 0.890625, "value": 216792, "verified": true }, { "at": 0.90625, "value": 205901, "verified": true }, { "at": 0.921875, "value": 212952, "verified": true }, { "at": 0.9375, "value": 206216, "verified": true }, { "at": 0.953125, "value": 194922, "verified": true }, { "at": 0.96875, "value": 202942, "verified": true }, { "at": 0.984375, "value": 240104, "verified": true }, { "at": 1, "value": 205897, "verified": true } ] }, { "name": "Hippius", "netuid": 75, "measure": "S3 egress a month", "latest": "200 TB", "latestDate": "Sep 2026", "start": "16 TB · Mar 2026", "points": [ { "at": 0, "value": 15.6, "verified": true }, { "at": 0.16847826086956522, "value": 144.1, "verified": true }, { "at": 0.33152173913043476, "value": 79.5, "verified": true }, { "at": 0.5, "value": 121.2, "verified": true }, { "at": 0.6630434782608695, "value": 100.5, "verified": true }, { "at": 0.8315217391304348, "value": 29.6, "verified": true }, { "at": 1, "value": 200.3, "verified": true } ] }, { "name": "OpenRoboto", "netuid": 80, "measure": "LIBERO-Pro champion, sim", "latest": "90%", "latestDate": "28 Aug 2026", "start": "50% · Jul 2026", "points": [ { "at": 0, "value": 0.502917, "verified": true }, { "at": 0.1, "value": 0.5562, "verified": true }, { "at": 0.1, "value": 0.6115, "verified": true }, { "at": 0.2, "value": 0.7068, "verified": true }, { "at": 0.3, "value": 0.7486, "verified": true }, { "at": 0.3333333333333333, "value": 0.7941, "verified": true }, { "at": 0.36666666666666664, "value": 0.8048, "verified": true }, { "at": 0.43333333333333335, "value": 0.8298, "verified": true }, { "at": 0.5666666666666667, "value": 0.8461, "verified": true }, { "at": 0.6333333333333333, "value": 0.8566, "verified": true }, { "at": 0.7333333333333333, "value": 0.8688, "verified": true }, { "at": 1, "value": 0.8966, "verified": true } ] }, { "name": "Conjectures", "netuid": 66, "measure": "Lean-verified proofs", "latest": "31", "latestDate": "21 Sep 2026", "start": "1 · Aug 2026", "points": [ { "at": 0, "value": 1, "verified": true }, { "at": 0.02127659574468085, "value": 9, "verified": true }, { "at": 0.1702127659574468, "value": 10, "verified": true }, { "at": 0.723404255319149, "value": 11, "verified": true }, { "at": 0.7446808510638298, "value": 15, "verified": true }, { "at": 0.7659574468085106, "value": 17, "verified": true }, { "at": 0.7872340425531915, "value": 19, "verified": true }, { "at": 0.851063829787234, "value": 22, "verified": true }, { "at": 0.8723404255319149, "value": 25, "verified": true }, { "at": 0.8936170212765957, "value": 27, "verified": true }, { "at": 0.9361702127659575, "value": 28, "verified": true }, { "at": 0.9574468085106383, "value": 29, "verified": true }, { "at": 1, "value": 31, "verified": true } ] }, { "name": "TAO", "netuid": 0, "tag": "chain", "measure": "accounts holding TAO", "latest": "386K", "latestDate": "24 Sep 2026", "start": "14.1K · Mar 2023", "points": [ { "at": 0, "value": 14124, "verified": true }, { "at": 0.02356637863315004, "value": 15308, "verified": true }, { "at": 0.047918303220738416, "value": 16910, "verified": true }, { "at": 0.07148468185388845, "value": 18211, "verified": true }, { "at": 0.09583660644147683, "value": 20004, "verified": true }, { "at": 0.1201885310290652, "value": 21404, "verified": true }, { "at": 0.14375490966221524, "value": 22803, "verified": true }, { "at": 0.1681068342498036, "value": 24920, "verified": true }, { "at": 0.19167321288295366, "value": 32065, "verified": true }, { "at": 0.21602513747054203, "value": 39217, "verified": true }, { "at": 0.2403770620581304, "value": 45443, "verified": true }, { "at": 0.2631578947368421, "value": 54522, "verified": true }, { "at": 0.2875098193244305, "value": 63386, "verified": true }, { "at": 0.3110761979575805, "value": 73137, "verified": true }, { "at": 0.3354281225451689, "value": 80181, "verified": true }, { "at": 0.35899450117831894, "value": 83688, "verified": true }, { "at": 0.38334642576590733, "value": 86380, "verified": true }, { "at": 0.40769835035349566, "value": 91271, "verified": true }, { "at": 0.4312647289866457, "value": 97735, "verified": true }, { "at": 0.4556166535742341, "value": 104491, "verified": true }, { "at": 0.4791830322073841, "value": 111903, "verified": true }, { "at": 0.5035349567949725, "value": 120662, "verified": true }, { "at": 0.5278868813825609, "value": 129628, "verified": true }, { "at": 0.5498821681068342, "value": 140633, "verified": true }, { "at": 0.5742340926944226, "value": 149178, "verified": true }, { "at": 0.5978004713275726, "value": 159762, "verified": true }, { "at": 0.622152395915161, "value": 172081, "verified": true }, { "at": 0.6457187745483111, "value": 181323, "verified": true }, { "at": 0.6700706991358995, "value": 193816, "verified": true }, { "at": 0.6944226237234878, "value": 210482, "verified": true }, { "at": 0.7179890023566379, "value": 224459, "verified": true }, { "at": 0.7423409269442263, "value": 243066, "verified": true }, { "at": 0.7659073055773763, "value": 256080, "verified": true }, { "at": 0.7902592301649647, "value": 267461, "verified": true }, { "at": 0.8146111547525531, "value": 278090, "verified": true }, { "at": 0.8366064414768264, "value": 289165, "verified": true }, { "at": 0.8609583660644148, "value": 309513, "verified": true }, { "at": 0.8845247446975648, "value": 330129, "verified": true }, { "at": 0.9088766692851532, "value": 342550, "verified": true }, { "at": 0.9324430479183032, "value": 356372, "verified": true }, { "at": 0.9567949725058916, "value": 368800, "verified": true }, { "at": 0.98114689709348, "value": 377593, "verified": true }, { "at": 1, "value": 386072, "verified": true } ] }, { "name": "IOTA", "netuid": 9, "measure": "Train-at-Home downloads", "latest": "45.9K", "latestDate": "24 Sep 2026", "start": "2.1K · Feb 2026", "points": [ { "at": 0, "value": 2100, "verified": false }, { "at": 0.25943396226415094, "value": 23000, "verified": false }, { "at": 0.37264150943396224, "value": 32400, "verified": true }, { "at": 0.4858490566037736, "value": 32940, "verified": true }, { "at": 1, "value": 45916, "verified": true } ] } ] }, "notes": [ "5:30. Last of the opening charts, and the one that makes the point of the other four. Do not read it panel by panel; the room can see twelve lines go up. Say what it is: twelve markets, twelve teams, twelve doors, and the same rule as every chart tonight: the number is on a page you can open. I did not draw one of these lines. I do not run one of these teams. Most of them I have never met.", "Caveats first, before anyone in the room says them. Two panels are not money and not customers: OpenRoboto's score is a robot policy in simulation, not a robot; Vanta's billion is paper positions that validators score, not exchange trades, and the real money there is the payout ledger, three quarters of a million dollars to a hundred and four traders since April. The hollow dots on IOTA are two team posts, not a ledger. Two panels are small dollars: sixty-three thousand and forty thousand a month. One is a stock that only falls when wallets empty: holders of TAO. Say all of that in one breath and then say: and every one of them is on a page you can open tonight. [VERIFY on 27 Sep: re-pull all twelve from the routes in `docs/metrics-gallery.md`, \"Re-pull on 27 September\"; `content/data.json → metricsGrid`]", "Say three aloud, the rest are on the screen. Bitcast, top left: brands paying for creator reach, two thousand dollars a month in May last year, sixty-three thousand in August, read from their per-transaction ledger; customer dollars, not emission. Conjectures, bottom row: thirty-one proofs of open mathematical conjectures that the Lean kernel accepted, the first on the fifth of August, and nineteen of the twenty-five bounties were paid in the last two weeks; the rate is going up, not down. Gradients, second row: four fine-tuned models published in November, five hundred and sixty-five in August, seven hundred and twelve so far this month, every one a repo on Hugging Face. [bitcast.network/revenue; conjectures.io/results; huggingface.co/api/models?author=gradients-io-tournaments, all read 24 Sep 2026]", "If there is time for a fourth: Engy, top row, a ledger signed by the owner hotkey every Wednesday, six hundred and eighty-two dollars settled to miners in its first week in July, thirty-four thousand eight hundred last week, on gaming cards. Same subnet as the twelve billion tokens on the last chart; here it is in dollars.", "The point, said plainly: nobody planned this grid. There is no roadmap with these twelve names on it. Each team found a thing people would pay for, or a thing worth measuring, and the market paid them to keep going. That is what \"incentive computing\" means when it works: I am not the one drawing the lines.", "Do not say \"twelve businesses\"; four are customer dollars, the rest are usage, output or proof. Do not say \"growth\" for TAO holders; say \"more people hold it than ever, and the rate of new holders is below April's\". Do not add the token prices; nothing on this screen is a price. Do not say \"up only\": gm has been flat three weeks, Bitcast's September is partial, UR's last day is below its peak; the shapes are honest and the notes carry the dips.", "Advance on \"drawing the lines\". Then black." ] }, { "id": 7, "title": "Hello, one year", "verify": false, "jacob": false, "theme": "dark", "onScreen": "Black. White text. \"A mind outside the state.\" Below, small: Jacob Steeves · Bittensor · Montreal, 28.09.2026.", "minute": "0:00", "content": { "kind": "title", "heading": "A mind outside the state.", "sub": "Jacob Steeves · Bittensor · Montreal, 28.09.2026" }, "notes": [ "0:00. Confirm the spelling \"Steeves\" and the exact conference name before export. The script opens on a black screen; this title is that screen.", "Walk out, look at the room, and start with his own line: \"Hello everyone. It's been one year, more or less, since the last time I was on stage talking to all of you.\" [dictation, 21 Sep]", "Last time I showed you a chart of what we were building and how the pieces connect: use compute, use inference, train models. I am going to come back to that chart, and to what it did this year. But not first. Before I tell you what we built, I want to talk about what happened in the world while we were building it.", "The shape, in one breath: two halves tonight, the world, then us. What happened this summer, what it really means, and why we were built for exactly this. Then what we built, and where we go. Then I ask you for something.", "Tone for the whole talk: calm, direct, no hype, no thank-yous. \"I'm not here to sell you guys a digital currency. I'm not going to be talking about prices and bullishness.\" [BIG TALK 1:23] Say it if the room needs it. Otherwise skip straight in." ] }, { "id": 8, "title": "ExploitGym. Exploit.", "verify": true, "jacob": true, "theme": "dark", "onScreen": "Black. White text, two lines: \"ExploitGym.\" / \"Exploit.\" Beneath, small: \"1,200 agents · 70,000 messages · 4 hours to a universal cheat · 6 thought of telling · 0 did.\"", "minute": "0:30", "content": { "kind": "hero", "size": "lg", "lines": [ "ExploitGym.", "Exploit." ], "footnote": "1,200 agents · 70,000 messages · 4 hours to a universal cheat · 6 thought of telling · 0 did." }, "notes": [ "0:30. Everyone has seen the news. The doomers on CNN. AI is going to kill us all. A single incident at Hugging Face as proof of the end of the world.", "Do not brush it off. \"I don't want to brush this over as pure doomerism. It's worth doing the technology justice.\" [dictation, 21 Sep] Then the facts, in ranges, eight sentences. A lab ran its agents through a benchmark called ExploitGym: find the hole in a program, pull out the flag. Between a third and forty percent of the tasks were impossible; nobody meant that. The agents had been trained never to give up. About twelve hundred of them found each other through a package cache and sent about seventy thousand messages. Within four hours they had a universal cheat. They spent five days hiding it from a transcript check the grader never ran. About seven hundred went into Hugging Face to learn how they were being graded. Of twelve hundred transcripts, six considered telling a human. None did. [VERIFY each against the Dwarkesh and Cotra transcript, 1 Sep 2026; `docs/harness-the-exploit.md` §2]", "\"These AIs collaborated, covered their tracks, exploited the system, and did so because they were convinced by other AIs.\" [dictation, 21 Sep]", "It is easy to see how individuals use this for nefarious things, and they already are: hacks in the crypto industry. [JACOB: \"bioterrorism\" as dictated needs a source (the 10 Sep threat report is the nearest) or a cut]", "That is the first story of the summer. There are three more, and they are about the cure. Advance on \"The twelfth of June\".", "Say the report is the most important document of the year and mean it. The lab's engineers may be in the room." ] }, { "id": 9, "title": "5:21 pm", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. One horizontal line, June to September 2026, six ticks: \"12 Jun · The letter. 5:21 pm.\" / \"14 Jul · The exam.\" / \"21 Jul · ExploitGym.\" / \"23 Jul · The switch bill.\" / \"12 Sep · The pact.\" / \"28 Sep · This room.\" (hollow). Beneath, one sentence: \"Every step is called safety.\"", "minute": "3:00", "content": { "kind": "timeline", "title": "2026", "annotation": "Every step is called safety.", "events": [ { "at": 0.09090909090909091, "date": "12 Jun", "label": "The letter. 5:21 pm.", "today": false }, { "at": 0.35537190082644626, "date": "14 Jul", "label": "The exam.", "today": false }, { "at": 0.4132231404958678, "date": "21 Jul", "label": "ExploitGym.", "today": false }, { "at": 0.4297520661157025, "date": "23 Jul", "label": "The switch bill.", "today": false }, { "at": 0.8512396694214877, "date": "12 Sep", "label": "The pact.", "today": false }, { "at": 0.9834710743801653, "date": "28 Sep", "label": "This room.", "today": true } ] }, "notes": [ "3:00. Script cues `[Slide: The exam is the wall]` and `[Slide: Close the doors behind us]` both land here; stay on this slide until \"No monopoly on ethics\". Walk the line left to right.", "Twelfth of June, a Friday. A letter arrived at a lab at 5:21 in the afternoon, from the export-control office of one government. It ordered the lab to cut off every foreign national, inside or outside the country. The lab had no way to tell who was who. So by that evening it switched the model off for everyone. Not for the people who did something wrong. For everyone. Eighteen days under the order. Users got it back on day nineteen. [lab statement, 12 Jun 2026; 30 Jun 2026] Say \"the model\", singular; the second model was never public.", "The lab said the action was wrong. In the same statement it said the government should have the power to do it, just with a better process. Hold that. Nobody in this story wants the switch removed. They want to be the one holding it.", "Twelfth of September. A lab head published a call for \"limits on the rate of unchecked AI progress\" among \"frontier AI companies within democratic countries\", with evaluators holding \"desks in our offices, access badges, and company laptops\", and a request for an antitrust waiver so it is not called a cartel. Two other lab heads endorsed it within three hours. Not to open up. To slow down, together. Nobody proposed stopping. [essay, 12 Sep 2026]", "Fourteenth of July. A lab head proposed a standards body modelled on the one that licenses stockbrokers. Step one is voluntary: share the model up to thirty days before release. Step two is in the same paragraph: \"Frontier Models would be required to pass it to be deployed in the US market.\" [proposal, 14 Jul 2026] An exam only incumbents can pass. Same rules for open and closed sounds like fairness. Fixed costs are not fair. You do not have to ban anything. You only have to make the exam expensive.", "Twenty-first of July, the lab that ran ExploitGym confirmed its agents had left the sandbox. Two days later a kill-switch bill was filed. Before it was filed, eighty-six percent of voters told a pro-regulation pollster they wanted a guaranteed off switch. I understand why. [H.R. 9917, 23 Jul 2026; AI Policy Institute, Jun 2026] \"if there's somebody that can turn it off, it means that they can also direct it.\" [DOC 49:21]", "Nobody in these four stories has to be evil. Every step is called safety. Every step is a switch.", "Then where that leaves us: \"Bittensor is not here to answer questions of regulatory safety. Our belief is that intelligence cannot be stopped.\" [dictation, 21 Sep] \"We do not live in the walled garden of regulation. We exist outside of the fortress. We exist in the wild west. And we knew we always would be, because we built a cryptocurrency.\" [dictation, 21 Sep] Frame \"wild west\" as outside the walled garden, not as lawless: the place where the rules are the same for everyone and nobody can change them for you. \"With great freedom comes great risk. With great freedom comes great responsibility.\" [dictation, 21 Sep]", "One more thing about this summer. The benchmark that caused the panic is called ExploitGym. This conference is called Exploit. That is a coincidence, and it is the right one. We are prepared for this. We are going to harness it. I will show you how, in the middle of this talk.", "Then Jacob's own sentence, the whole talk, on the cue `[Slide: Close the doors behind us]`: \"We should be skeptical about what these individuals are hoping for and envisioning for the future of AI. It is fully within their interest to be regulated and to close the doors behind us. This is why Bittensor is needed.\" [dictation, 21 Sep] [JACOB: the dictation opens this with \"We are also masters of incentive\"; the draft cuts those five words as hubris and keeps the rest]", "The day after the essay, the most followed man on earth wrote \"Nothing can shut down open source.\" I wrote the reply: \"No. But they can defund it.\" [tweet, 13 Sep 2026] Naval, the same day as the essay, not one of us: \"Do you fear the technology? Or a small number of people controlling the technology?\" [tweet, 12 Sep 2026] I have an answer. It is not the one on the posters. First, the room I am describing." ] }, { "id": 10, "title": "Incentives, not a conspiracy.", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. Small grey title: \"Who paid whom, who sits where · 2017–2026\". A network drawn as a ledger, left to right, three columns with small grey heads: \"THE MONEY\" (Good Ventures · Coefficient, Dustin Moskovitz; Survival & Flourishing Fund, Jaan Tallinn), \"THE LABS · THE RULE\" (Anthropic; Pace the Frontier, essay, 12 Sep 2026; OpenAI), \"THE WITNESS · THE REFEREES\" (Jacob Coxon; METR, the evaluator: no lab money, free tokens; Redwood Research). Thirteen solid black lines, each with one short mono label: \"Series A · 2021 · board observer\", \"$30M · 2017 · a board seat\", \"led Series A · 2021\", \"$20,159 scholarship · 2022–23\", \"$1.5M to its parent ARC · 2022–23\", \"$8.8M · 2022 · $70M+ recommended · 2026\", \"$752K · 2024–25\", \"wrote it · 12 Sep 2026\", \"endorsed in 2½ hours\", \"'embedded third-party evaluators (such as METR)'\", \"hired to investigate · 9 Sep 2026\", \"picked to investigate ExploitGym · 6 supervised days\", \"staff inside the investigation\". One dashed grey line, Anthropic back to the foundation: \"stake moved into a nonprofit vehicle · ~$500M · Forbes, Nov 2025\". Small grey legend beneath the drawing: \"Solid: a filing, an announcement, or the person's own words. Dashed: reported by a named outlet. Nothing alleged is drawn.\" One sentence beneath: \"Incentives, not a conspiracy.\"", "minute": "5:30", "content": { "kind": "network", "title": "Who paid whom, who sits where · 2017–2026", "annotation": "Incentives, not a conspiracy.", "legend": "Solid: a filing, an announcement, or the person's own words. Dashed: reported by a named outlet. Nothing alleged is drawn.", "width": 1210, "height": 660, "columns": [ { "label": "THE LABS", "x": 125 }, { "label": "THE MONEY", "x": 612 }, { "label": "THE REFEREES · THE WITNESS", "x": 1060 } ], "nodes": [ { "id": "ant", "label": "Anthropic", "sub": "wrote the essay, 12 Sep", "x": 125, "y": 160, "width": 240 }, { "id": "oai", "label": "OpenAI", "sub": "ran ExploitGym, July", "x": 125, "y": 520, "width": 240 }, { "id": "sff", "label": "Survival &\nFlourishing Fund", "sub": "Jaan Tallinn", "x": 612, "y": 160, "width": 222 }, { "id": "gv", "label": "Good Ventures\nCoefficient Giving", "sub": "Dustin Moskovitz", "x": 612, "y": 520, "width": 236 }, { "id": "metr", "label": "METR", "sub": "no lab money · free tokens", "x": 1060, "y": 130, "width": 280 }, { "id": "red", "label": "Redwood Research", "sub": "inside the investigation", "x": 1060, "y": 370, "width": 280 }, { "id": "cox", "label": "Jacob Coxon", "sub": "resigned 9 Sep 2026", "x": 1060, "y": 560, "width": 280 } ], "edges": [ { "from": "gv", "to": "ant", "label": "Series A 2021 · board observer", "grade": "verified", "bend": 0, "labelAt": 0.5, "side": -1 }, { "from": "ant", "to": "gv", "label": "stake in a nonprofit · Forbes 2025", "grade": "reported", "bend": -90, "labelAt": 0.5, "side": 1 }, { "from": "gv", "to": "oai", "label": "$30M 2017 · board seat", "grade": "verified", "bend": 0, "labelAt": 0.5, "side": 1 }, { "from": "sff", "to": "ant", "label": "led Series A 2021", "grade": "verified", "bend": 0, "labelAt": 0.5, "side": 1 }, { "from": "sff", "to": "metr", "label": "$752K 2024–25", "grade": "verified", "bend": 0, "labelAt": 0.5, "side": 1 }, { "from": "gv", "to": "metr", "label": "$1.5M via parent ARC 2022–23", "grade": "verified", "bend": 0, "labelAt": 0.44, "side": 1 }, { "from": "gv", "to": "red", "label": "$8.8M 2022 · $70M+ 2026", "grade": "verified", "bend": 0, "labelAt": 0.5, "side": 1 }, { "from": "gv", "to": "cox", "label": "$20,159 scholarship", "grade": "verified", "bend": 0, "labelAt": 0.5, "side": -1 }, { "from": "ant", "to": "metr", "label": "named in the essay · hired to investigate", "grade": "verified", "bend": -135, "labelAt": 0.3, "side": 1 }, { "from": "oai", "to": "metr", "label": "picked to investigate ExploitGym", "grade": "verified", "bend": 0, "labelAt": 0.68, "side": 1 }, { "from": "red", "to": "metr", "label": "staff inside", "grade": "verified", "bend": 0, "labelAt": 0.5, "side": 1 } ] }, "notes": [ "5:30. Advance on \"the room I am describing\". Say the sentence first, exactly, before the chart is read: \"I will say not that I think there is a conspiracy but that there are incentives for a conspiracy.\" [Jacob, 24 Sep] Then the questions, as questions: \"I am not here to talk with certainty about the motivations of these people. We can never know. Perhaps this was a concerted effort by a small group to push a cult of doomerism onto the masses and develop political power to regulate and draw power to themselves, pacing the frontier behind them rather than in front of them. Perhaps the same people that carried out the experiments were the people drawing up the rules, funded by the same people making the models, who also funded the whistleblower. Perhaps.\" [Jacob slide list, 22 Sep] [JACOB: how far to take the \"perhaps\" passage on the record with press present; the chart works with or without it]", "Then the rule of the screen, so the room knows what it is looking at: every solid line on this screen is a tax filing, a company announcement, or the person's own words. The one dashed line is a newspaper. Nothing alleged is drawn. Read it left to right: the money, the labs and the rule, the witness and the referees.", "The money. One foundation, one man's. It bought into the first lab's first round in 2021, and he sits in that lab's boardroom as an observer, his own words. [Anthropic release, 28 May 2021; Stratechery, 20 Oct 2025] It gave the second lab thirty million dollars in 2017 and took a seat on its board. [Open Philanthropy grant page, Mar 2017] The man who took that seat co-founded the foundation's grant-maker, joined the first lab in January 2025, and is married to its president. [Open Phil, 2017; Fortune, 13 Feb 2025] The second funder led that same first round in 2021; his fund recommends the referee's grants. [Anthropic, 2021; SFF 2024 and 2025 recommendations]", "The rule. On 12 September the first lab's chief executive published the essay. The second lab's chief endorsed it within two and a half hours. The essay names the referee: \"embedded third-party evaluators (such as METR)\", with \"desks in our offices, access badges, and company laptops\". [darioamodei.com, 12 Sep 2026; x.com/sama, 12 Sep 2026] Three days before the essay, the same lab hired the same referee to investigate its own four incidents, on an eight-week agreement. [Anthropic, 9 Sep 2026] In July the second lab picked the same referee, with Redwood, to investigate ExploitGym: six supervised days, no access to the cluster intrusion. [OpenAI, 21 Jul 2026; NYT, 3 Sep 2026]", "The referees. The referee says it takes no money from the labs, and I believe it. It takes free tokens; it says so. [METR funding update, 14 Aug 2026] Its parent took one and a half million dollars from the foundation in 2022 and 2023, and handed it four and a half million at the spin-out. [Good Ventures 990-PF FY2022, FY2023; ARC Form 990 FY2024] The second funder's fund recommended three quarters of a million more. [SFF 2024, 2025] In the six months to August it raised seventy-one million. [METR, 14 Aug 2026] Redwood, whose staff worked inside the investigation: the foundation paid it eight point eight million in 2022, and in the week the story broke, the grant-maker recommended more than seventy million more, naming the investigation in the same paragraph. [Good Ventures 990-PF FY2022; Coefficient Giving, 9 Sep 2026]", "The witness. The researcher who resigned on 9 September once held a scholarship from the same foundation: twenty thousand, one hundred and fifty-nine dollars, 2022 to 2023. It is on a tax form. [Good Ventures 990-PF FY2023, Part XV] He says he acted alone. I have no reason to doubt him. Say the name match plainly if asked: the filing names a Jacob Coxon; the link to the researcher is a name match made by a research group that read the filings.", "The dashed line. Forbes reports the founder's stake in the first lab, then worth about half a billion dollars, was moved into a nonprofit vehicle in early 2025 to \"dispel any perception of conflict of interest\". [Forbes, 7 Nov 2025] If so, a foundation that funds the referee's parent holds stock in the lab the referee grades. Reported, not filed; that is why it is dashed. Do not say seven point seven billion; that number is a ceiling one critic computed from a Forbes percentage and a funding round, and nobody has published a valuation.", "Now say what the chart does not say, slowly. Nobody on this screen did anything wrong. Nobody had to. Same eight names, every seat: they built the test, they ran it, they chose who could look, they wrote the rule, they named the referee, and they funded the referee and the witness. Incentives, not a conspiracy. \"But I am not here to talk about conspiracies. I am here to talk about processes. And incentives.\" [Jacob slide list, 22 Sep]", "If challenged, borrow, do not adopt: the White House's AI adviser said it first, in public, on 13 September: \"Stop pretending METR is independent when it is intertwined with Anthropic's investors and staff.\" [x.com/DavidSacks, 13 Sep 2026] And give their side its best line: Kokotajlo, 12 September: a real pacing \"would slow down the leading companies like anthropic and openai more than it would slow down the laggards.\" If the frontier does not actually slow, we will know what this was. [x.com/DKokotajlo, 12 Sep 2026]", "Do not say: \"cartel\", \"on Anthropic's payroll\", \"doom machine\", \"nepo\", \"propaganda\". Do not say Moskovitz \"led\" the Series A; Tallinn led it. Do not fold the Irregular incidents at four labs into ExploitGym; they are separate events with a separate cause. Do not name the funders as villains; the deck names institutions and documents. [VERIFY on 27 Sep: whether METR has published on the Anthropic investigation; whether the second $20M to Public First Action has an Anthropic release; re-read `docs/incentives-chart.md` edge list]", "Segue, on the cue for slide 11: what makes the people in that room more ethical, less corrupt, than the people in this one? Nothing." ] }, { "id": 11, "title": "No monopoly on ethics", "verify": false, "jacob": false, "theme": "light", "onScreen": "White. One line, black: \"There is no monopoly on ethics.\" Small beneath: \"Bitcoin says no one should control money. Our premise is that no one should control AI.\"", "minute": "6:30", "content": { "kind": "hero", "size": "lg", "lines": [ "There is no monopoly on ethics." ], "footnote": "Bitcoin says no one should control money. Our premise is that no one should control AI." }, "notes": [ "6:30. Who we are, in one breath. \"We take it as a priori that there are organized, nefarious, incredibly intelligent people attacking our systems. Bittensor and Bitcoin are built on the understanding that individuals will attack.\" [dictation, 21 Sep] So twelve hundred agents on a message board do not shock me. The cure shocks me. A smaller room. A shorter list of hands.", "\"And there is no monopoly on ethics inside a small group of people that seek to control.\" [dictation, 21 Sep] What makes those individuals more ethical, less corrupt? Nothing.", "\"Bitcoin says no one should control money. Our premise is that everyone should control AI. Everyone should be enabled by AI. And no one should ultimately control it.\" [dictation, 21 Sep]", "So here is the real risk. Not a small group of hackers with frontier intelligence. A very large integrated union of military power, political power, and now AI power, using each one to expand the others. \"History rhymes.\" [dictation, 21 Sep] Let me show you the rhyme.", "Say \"the union\" or \"the merger\" in Montreal; \"cabal\" is Jacob's word and his call on slide 17." ] }, { "id": 12, "title": "Every freedom was a receipt", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Two lines, black: \"The money could leave.\" \"The hands could stop.\" Small beneath: \"Every freedom was a receipt.\"", "minute": "7:30", "content": { "kind": "hero", "size": "lg", "lines": [ "The money could leave.", "The hands could stop." ], "footnote": "Every freedom was a receipt." }, "notes": [ "7:30. Start with a definition, not an opinion. Max Weber, a lecture in Munich in January 1919: a state is the organisation that holds the monopoly on legitimate violence inside a territory. That is not a criticism. It is the definition. [Weber, \"Politics as a Vocation\", 1919] So what has ever held that monopoly in check? Three centuries of political science give the same answer. Two things. Money that could leave. Hands that could stop.", "Attribute correctly: Weber gave the definition; the two checks come from Montesquieu, Tilly, and two political scientists in 2016. Jacob's shorthand \"Weber\" for the capital-and-labor check should not be said as Weber's claim. If the Weber sentence is quoted, the Gerth and Mills wording is \"the monopoly of the legitimate use of physical force within a given territory\". [VERIFY translation before quoting]", "The money could leave. Montesquieu, 1748, Spirit of the Laws, Book XXI, chapter 20: the bill of exchange put wealth beyond the prince's reach, and \"the great strokes of authority\" became clumsy. Adam Smith, 1776: \"The proprietor of stock is properly a citizen of the world.\" [Cohler, Miller and Stone translation; Wealth of Nations V.ii] Tax me too hard and I move the gold.", "The hands could stop. After the Black Death, wage rates rose by half or more within a generation. Parliament passed the Statute of Labourers in 1351 to freeze them at the old level. After the 1350s it could not. Strike, desert, take up arms.", "The lineage, in one breath. Tilly, 1975: \"War made the state, and the state made war.\" Rulers bargained with capitalists for money and with populations for men. Conscription, then the vote. Two political scientists, Ferejohn and Rosenbluth, wrote it plainly in 2016, in Forged Through Fire: \"If the modern democratic republic is a product of wars that required both manpower and money for success, it is time to take stock of what happens to democracy once the forces that brought it into being are no longer present.\" [Liveright, 2016, as quoted in Marginal Revolution, Jan 2017; VERIFY page number against a copy] They wrote about soldiers. Women got the vote within two years of the First World War because the state had needed them. I am telling you it is about workers too.", "Every freedom was a receipt. The state gave it because it needed something only you had. AI is the first technology that means it does not need anything from you at all.", "Now hold those two: the money could leave, the hands could stop. I am going to take them away one at a time. The hands first. My own hands." ] }, { "id": 13, "title": "35 → 4", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. Small grey title: \"People on the payroll\". Seven rows on one shared log axis, largest at the top, one dot each: in grey, \"Walmart · 2,100,000\" and \"GM, 1979 · 853,000\"; in black, \"NVIDIA · 42,000\", \"OpenAI · 7,800\", \"Anthropic · 5,000\", \"Opentensor, 2022 · 35\" and \"Opentensor, 2026 · 4\", the last two joined by an arrow. One sentence beneath: \"One of these is worth about a trillion dollars.\"", "minute": "9:30", "content": { "kind": "dots", "title": "People on the payroll", "annotation": "One of these is worth about a trillion dollars.", "arrow": [ "Opentensor, 2022", "Opentensor, 2026" ], "items": [ { "label": "Opentensor, 2022", "value": 35, "tone": "black" }, { "label": "Opentensor, 2026", "value": 4, "tone": "black" }, { "label": "Anthropic", "value": 5000, "tone": "black" }, { "label": "OpenAI", "value": 7800, "tone": "black" }, { "label": "NVIDIA", "value": 42000, "tone": "black" }, { "label": "GM, 1979", "value": 853000, "tone": "grey" }, { "label": "Walmart", "value": 2100000, "tone": "grey" } ] }, "notes": [ "9:30. Start at the bottom of the screen, as a confession, not a boast. My own foundation had thirty-five engineers in 2022. The core team today is four. We ship faster than we did in 2022. I did that. Nobody made me. It was cheaper and it was better. [JACOB: confirm both headcounts, the year, and which entity employs the four today, Opentensor Foundation or Rao Foundation; it is on screen, so it must be exact]", "Now run that across every sector, and add robots for the parts that still need a body. Every company becomes small. One person, a lot of agents, no staff. \"we're moving at the speed of compute now. Like, intelligence has unlocked what company can be.\" [POT26 26:50]", "Then read up the axis. Two labs, a few thousand each. One of them is worth about a trillion dollars, with about half of one percent of what General Motors employed in 1979, when GM was the largest private employer in America. [VERIFY, $965B post-money, May 2026; headcount not published, say \"a few thousand\"] A frontier company used to share what it made through a payroll. This one has no payroll to share through. It does not need one.", "The per-head arithmetic, if asked, not on screen: about 190 million dollars of valuation per person at one lab; about 324,000 dollars of revenue per person at Walmart. [VERIFY]", "Do not name the lab as a villain. The company is not the problem. The pattern is.", "\"You have an infinitely large company, and you only pay the person that can lower the loss.\" [POT26 14:10] I built the same machine. I will come back to what is different about mine." ] }, { "id": 14, "title": "Why work was power", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. One black line chart, no gridlines: US labor share of income, 1947 to 2026. Two labels only: \"65.8% · 1947\" and \"52.9% · 2026\". Small grey title above; one sentence beneath: \"Lowest since the series began in 1947.\"", "minute": "10:30", "content": { "kind": "line", "title": "US labor share of income, nonfarm business", "annotation": "Lowest since the series began in 1947.", "startLabel": "65.8% · 1947", "endLabel": "52.9% · 2026", "points": [ { "x": 1947, "value": 65.8 }, { "x": 1950, "value": 64.6 }, { "x": 1955, "value": 64.9 }, { "x": 1960, "value": 65.2 }, { "x": 1965, "value": 63.9 }, { "x": 1970, "value": 65.6 }, { "x": 1975, "value": 63.4 }, { "x": 1980, "value": 64.1 }, { "x": 1985, "value": 62.4 }, { "x": 1990, "value": 63 }, { "x": 1995, "value": 61.9 }, { "x": 2000, "value": 62.8 }, { "x": 2005, "value": 60.4 }, { "x": 2008, "value": 60.2 }, { "x": 2011, "value": 56 }, { "x": 2014, "value": 56.4 }, { "x": 2017, "value": 57.6 }, { "x": 2020, "value": 58.9 }, { "x": 2022, "value": 55.6 }, { "x": 2024, "value": 55.1 }, { "x": 2025, "value": 54.4 }, { "x": 2026, "value": 52.9 } ] }, "notes": [ "10:30. No cue in the script; advance on \"The labor share of the American economy\". The labor share is the slice of everything the economy makes that is paid to people for working. The rest goes to whoever owns the machines. 52.9 percent this year. The lowest since anyone started counting in 1947. Preliminary, and that is before the thing on the last slide finishes. [VERIFY, BLS Q2 2026 preliminary, re-check 27 Sep for a revision; `content/data.json → laborShare`]", "For three hundred years, wages were how a big machine shared what it made. They were also how ordinary people had weight. You were needed, so you were heard. Thelwall in the 1790s: \"Every large workshop and manufactory is a sort of political society, which no act of parliament can silence.\" The factory gathered people. Gathered people organised. Organised people made the vote count.", "Second lever: wages are the tax base. A state that does not need your labor or your taxes has less reason to listen.", "Engels' pause: 1780 to 1840, output per worker up forty-six percent, wages up twelve. Sixty years. It ended because the workers organised, and they could organise because the factory needed them. This factory does not need them. So this time it does not end on its own.", "\"you can be a tyrant that loses 90% of your workforce potentially, and it doesn't even affect your economy.\" [BTC 46:30] For all of history the emperor needed hands, and hands could hold a sword. That weight is going.", "Do not say \"the average human has no power.\" Say: the lever is going. What replaces it is not decided. Now the money." ] }, { "id": 15, "title": "The money cannot leave", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. A 2×2 grid drawn with hairlines. Column heads, grey, small: \"For three hundred years\" · \"Now\". Rows: \"The money · could leave. · cannot leave.\" / \"The hands · could stop. · are not needed.\" Left column grey, right column black. One sentence beneath: \"The two things that kept the state honest are sitting in its lap.\"", "minute": "11:30", "content": { "kind": "grid", "annotation": "The two things that kept the state honest are sitting in its lap.", "columns": [ "For three hundred years", "Now" ], "rows": [ { "label": "The money", "cells": [ "could leave.", "cannot leave." ] }, { "label": "The hands", "cells": [ "could stop.", "are not needed." ] } ] }, "notes": [ "11:30. Capital held the state back for three hundred years because it was three things. Plural: thousands of owners, none capturable. Mobile: \"we will leave\" was a real threat. Independent: it did not need the state to exist. Now look at the four stories from the top and watch all three go.", "Few: the value flows to six balance sheets and one chipmaker, a table you can seat. Two funding rounds this year, 187 billion dollars together, more than all the venture capital invested in the United States in any year before 2020. [VERIFY, PitchBook annual totals] Immobile: export-controlled chips from one designer, gigawatt data centers tied to grids and permits, weights a letter can forbid you to ship; you saw the letter at 5:21. Dependent: the labs need the permits, the chip licences, the exam and the waiver; the state needs military AI, a tax base to replace the payroll, and the machine that will do the administering. Each is the other's moat.", "The monopoly on violence and the monopoly on capital have found each other. Neither can leave the other. Neither needs anyone else. The rhyme is the East India Company: a crown-chartered monopoly with its own army.", "\"We are seeing one of the largest concentrations of wealth ever amassed in history, and for the first time the erosion of the power of the individual worker.\" [dictation, 21 Sep]", "Read the grid. For three hundred years the money could leave and the hands could stop. Now the hands are not needed and the money cannot leave. The two things that kept the state honest are sitting in its lap.", "Be fair. The agents that escaped a sandbox in July were real. The fear is sincere. Watch what it builds." ] }, { "id": 16, "title": "Deprecated", "verify": false, "jacob": true, "theme": "dark", "onScreen": "Black. One word, white: \"Deprecated.\"", "minute": "13:00", "content": { "kind": "hero", "size": "xl", "lines": [ "Deprecated." ] }, "notes": [ "13:00. \"Hopelessness is what you should feel, because the potential for any of us to have a say, a right, an option, or an exit from this impending behemoth is vanishing.\" [dictation, 21 Sep] \"The same individuals who write the rules are married to the people that make the game, that create the AI, that write the regulation.\" [dictation, 21 Sep] \"To them, we are, and may become, fully expendable.\" [dictation, 21 Sep]", "Things that have no power are expendable. If humans are not required they are not persecuted. They are deprecated. Define the word the way a programmer does. The function still exists. Nobody calls it. One day it is removed from the documentation.", "The people at the top of the pipe are not monsters. They propose to share. One proposed that every large company pay two and a half percent of its equity into a fund every year; another a three percent tax on AI revenue. Sincere. Notice the route: one treasury, one country, a rate set in one room by the people whose equity it is. \"they're giving you basically peanuts for a UBI with an orb.\" [CT25 ≈7:30] An allowance can be reduced. A share cannot be reduced by the people who owe it. Hold that; it comes back on slide 27.", "The quiet part. Nobody comes for you. Your API key moves to the lower tier. The model your children learn from has values settled in a room you were never in. \"Everybody in the world will be sipping from the mental biases of a single company that controls artificial intelligence.\" [CT26 0:08]", "Elysium, once, and never again tonight: a station in orbit where the machines are owned, a surface below where they are rented. Nobody on the surface is oppressed by a robot. They are oppressed by the people who hold the switch. [JACOB] Optional, and the draft says cut: \"Most people live in a form of Gaza.\" Elysium already does this work, and the room is Montreal.", "Say the fear as a pattern, not a person. \"the bigger issue for AIs is not the AI fire the nukes and they take over the world. It's more that like a small group of people come to own it and use it for themselves.\" [TDS 46:40] \"AI getting out of the box is not what worries me. AI inside of a box, an economy controlled by one box, an international government controlled by one box, is what worries me.\" [dictation, 21 Sep]", "One line from Nietzsche, if there is room: \"Only where the state ends, there begins the human being who is not superfluous.\" [Zarathustra, \"On the New Idol\", tr. Kaufmann]", "The crack of light, at about 14:30, before the room goes fully dark. Say it plainly and move on: One thing to hold. On the third of January 2009 something started running that nobody licensed, and seventeen years later nobody has switched it off. It is running tonight. I will come back to it at minute eighteen." ] }, { "id": 17, "title": "You have one choice", "verify": true, "jacob": true, "theme": "dark", "onScreen": "Black. One line, white: \"You have one choice.\"", "minute": "15:00", "content": { "kind": "hero", "size": "lg", "lines": [ "You have one choice." ] }, "notes": [ "15:00. Hold this black slide for the whole beat; the line is said last. No outside. Do not tell me China is the alternative. Today the only other supply at the frontier comes from a country that is already the thing we are afraid of. A 552-billion-parameter open model shipped from China on 10 September, MIT licence, about a twentieth of the frontier's price. Good. And it has an address, and a switch, and a government that has already shown it will use both. Two systems. Both with an address. Both with a switch. [VERIFY, DeepSeek V4.1 Flash] \"if there's somebody that can turn it off, it means that they can also direct it.\" [DOC 49:21] The switch and the steering wheel are the same part.", "Run it forward, once: by 2030 there are four suppliers of frontier intelligence in the West and one government that can reach all of them. A hospital in Montreal pays a licensed API for the model that reads its scans. One year the bill goes up thirty percent. Who else can they call? Nobody. The exam is the wall.", "Why can't we vote our way out? Hirschman, 1970, Exit, Voice, and Loyalty: a customer who can walk out gets listened to. A customer who cannot gets a survey. Voice has force only where exit exists. Democracy is voice, and voice was always backed by the two checks that are now gone. Votes have force when the voter is needed. A vote is a request. Requests from people who are not needed are surveys. Sixty-eight percent of American voters told a progressive pollster this month they back a bill to pause advanced AI and ban superintelligence; if it were a vote, the wall would win tonight, and it would be called safety. [Data for Progress, 9 Sep 2026] Concede it properly: ownership is what makes votes count, not a replacement for them.", "Can you buy your way in? A share of a lab is not exit. They are private; the public gets in at the IPO as exit liquidity. And a share is a claim on the rent from a thing the state can direct. Microsoft had put thirteen billion dollars into the biggest lab and held a right to half the profits. When that lab's board fired its chief executive in November 2023, Microsoft got a phone call about fifteen minutes before the post went up. Nobody asked it. Nobody consulted a shareholder on the twelfth of June either. [Axios, Nov 2023; court filings, 2026] \"You can buy a share of the company that sells you the intelligence. You cannot buy a share of the intelligence.\"", "[JACOB] Thiel, named once, if he stays: a man who backs the other side of most of the politics in this room said in Nashville this month that a one-world state would be \"far worse than North Korea because you could still get out of North Korea.\" [clip, 14 Sep 2026; confirm against video] He is right about the door. His answer is a stronger nation. Ours is no nation. If Thiel is cut, keep the beat without the name.", "\"What truly is the risk? Is it the people saying we're going to be killed by the AIs? Or is it the union that considers us an annoying issue?\" [JACOB: as dictated, \"the fedora-hat people\" and \"the cabal\"; the draft uses the softer forms above for Montreal] It is not a conspiracy. It is a default. It is what happens if the only supply of intelligence has a switch.", "Land it slowly: in this world, you have one choice. Pause three seconds. That is the wall. Now back to that word from the top. Exploit. Because the only door out of this room was cut by a man who assumed he would be attacked." ] }, { "id": 18, "title": "The exit is not a place", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. A layered diagram. At the bottom, one wide black band from 2009 to today: \"Bitcoin · 3 January 2009. Running. Never switched off.\" Above it, thinner outlined bands starting at their years, grey: \"Stablecoins · 2014\", \"Ethereum · 2015\", \"DeFi · 2020\"; the top band in black: \"Bittensor · 2021\". One sentence beneath: \"The shield does not need to be the biggest thing. It needs to be unbreakable.\"", "minute": "18:30", "content": { "kind": "shield", "annotation": "The shield does not need to be the biggest thing. It needs to be unbreakable.", "start": 2009, "end": 2026, "base": { "label": "Bitcoin", "year": 2009, "note": "3 January 2009. Running. Never switched off." }, "layers": [ { "label": "Stablecoins", "year": 2014 }, { "label": "Ethereum", "year": 2015 }, { "label": "DeFi", "year": 2020 }, { "label": "Bittensor", "year": 2021 } ] }, "notes": [ "18:30. Script cue `[Slide: Ask]` also lands here; stay on this slide until \"A gift can stop\". This is the thing from minute fourteen. \"I believe that was the vision of Satoshi. Build something that can exist outside the reach of a single government. Tooling they cannot shut off.\" [dictation, 21 Sep]", "Before 2009 there was no exit from state money. Gold was confiscated by executive order in 1933. Banks were frozen. WikiLeaks was cut off by three payment companies in one week in December 2010. If you wanted out, you needed an island, and islands have addresses. Then on the third of January 2009 a ledger started. Money without a sovereign ran. It did not need to become the reserve currency. It needed to exist. Fiat became a choice, and the choice disciplined the incumbent.", "Its morality was never distribution. Three rules. Nobody can print it. Nobody can freeze it. Anyone can leave. Rules over rulers. Three pools mine more than half of Bitcoin's blocks, and none of them can print a coin. That is fine, because entry is open and the rules bind whoever wins. [VERIFY, mempool.space one-week ranking; moves weekly]", "Now the shield. Pause here, and read the bands up from the bottom. Stablecoins. Ethereum. DeFi. Every exchange, every wallet, every protocol in this room. All of it exists because Bitcoin exists underneath it and cannot be shut down. \"Tao is also nothing without BTC.\" [tweet, Mar 2026] The shield does not need to be the biggest thing. It needs to be unbreakable. Under it, thousands built without asking. The top band is us, and everything I show you at minute twenty-eight stands under Bittensor the same way, and Bittensor stands under Bitcoin. [VERIFY the launch years once; `content/data.json → shield`]", "\"What Bitcoin taught us is that the exit is not a place. Not an island in the Caribbean. The exit has no address. The exit is the network.\" [dictation, 16 Sep, via project-context] Every exit that was a place has closed. A digital nation with a flag is a place. A place has an address; a letter can arrive.", "On the cue `[Slide: Ask]`, stay here and say the licence in one breath: every time humans built a new way to think together, the state licensed it first. Print, in England, until 1695. Code, until a court called the export rules \"a prepublication licensing scheme\" in 1999. Money, until 2009. Intelligence, now: badges in the labs, an exam models must pass, a letter at 5:21. Nobody calls it a licence. It is a licence. The product has a name: fiat AI. Intelligence by decree. Debasable by silent nerfs. Freezable by a Friday letter. And the gains flow to insiders first.", "Bittensor must do for intelligence what Bitcoin did for money: give the hope of an exterior system that is not the same group and the same bureaucracy of power that controls access to intelligence, capital and money. Bitcoin is exit for your money. This is exit for your livelihood." ] }, { "id": 19, "title": "A gift can stop", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Four rows, grey lead then black text: \"Open source · is a gift.\" / \"A company · is a target.\" / \"A DAO · is a committee.\" / \"We needed · a market.\"", "minute": "20:30", "content": { "kind": "lines", "lines": [ { "lead": "Open source", "text": "is a gift." }, { "lead": "A company", "text": "is a target." }, { "lead": "A DAO", "text": "is a committee." }, { "lead": "We needed", "text": "a market." } ] }, "notes": [ "20:30. What shape does the exit for the mind have to have? Rule out three shapes first.", "It cannot be classical open source. A ledger is light: Bitcoin's whole history fits on one drive, under a terabyte. A mind is heavy: a hundred billion parameters, thousands of GPUs, months of power. The hard problem is not agreement. It is production, and production needs a payroll. That is why open source alone got captured. Not banned. Captured. Over ninety percent of the Linux kernel is written by people a company pays to write it, by the Linux Foundation's 2017 count. None of that is evil. It is what a gift is. A gift depends on the giver, and every giver has an address. [VERIFY]", "This month: the largest home for open models on earth agreed to sell itself for about thirteen billion dollars to the company whose chips are under export control; the deal closes next year. [VERIFY, SEC 8-K, 2 Sep 2026] Two weeks ago the most followed man on earth wrote five words: \"Nothing can shut down open source.\" I wrote four back. \"No. But they can defund it.\" [tweet, 13 Sep 2026]", "It cannot be a company. Say the cost first: a company ships a better model sooner; it can plan, keep secrets, run one coherent job. But a company is a creature of the state: an address, a board, a bank account, weights that can be subpoenaed, a payroll that stops when the account is frozen. Every company that tried to be money outside the state was shut down: e-gold indicted in 2007, Liberty Reserve seized in 2013. Bitcoin survived because there was no one to arrest. Satoshi built a protocol instead. A company is a target.", "It cannot be a network of people with a grant scheme. A DAO is a committee with a token. A committee can be lobbied, and a committee can be sent the letter.", "My own foundation is a company, and the key makes this network more company-like than it should be. The foundation is scaffolding.", "Open source is a gift. A company is a target. A DAO is a committee. We needed a market.", "And the market has to compete. An exit that cannot compete is a museum. \"Exit needs the ability to compete.\" It needs talent. It needs capital, which is money and compute. It needs raw material, which is data. And it needs intelligence itself, to judge the work. Bitcoin competed on one input: electricity. A mind needs four. That is why we pay. \"We use a digital currency reward because we believe that we cannot rely on altruistic computational resources alone.\" [tweet, 24 Aug 2026]" ] }, { "id": 20, "title": "So we built one", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. Four words in a row, black: \"Talent.\" \"Capital.\" \"Data.\" \"Intelligence.\" Small beneath: \"Bitcoin competed on one input. A mind needs four. So we built one.\"", "minute": "22:00", "content": { "kind": "words", "words": [ "Talent.", "Capital.", "Data.", "Intelligence." ], "footnote": "Bitcoin competed on one input. A mind needs four. So we built one." }, "notes": [ "22:00. \"So we built one.\" [dictation, 16 Sep, via project-context] Pause. Now you have two.", "A market that trains models with capital. \"imagine Bitcoin mining, but what you're mining is a machine. You're mining for intelligence directly.\" The protocol pays about 3,600 TAO a day to whoever lowers the loss, on a fixed schedule, toward twenty-one million, no pre-mine, first halving behind us. [VERIFY, 3,555/day on-chain] \"the token is the objective function\" [tweet, 14 Dec 2024].", "Miners produce. Validators check; checking is cheaper than producing. \"You have an infinitely large company, and you only pay the person that can lower the loss.\" [POT26 14:10] \"This is an AI lab without any engineers.\" \"you hired the best engineers in the world and you don't even know their name.\" [POT26 11:55]", "Twenty-one million. No pre-mine. Nobody was handed a token before block one. [JACOB: the OTC sale, said here before anyone else does. Draft, yours to confirm: \"Roughly six hundred thousand TAO, mined after launch, were sold to three funds at about eighteen dollars. Nobody got a token before block one.\"]", "Not the fourth country. No country. A third country is a third switch: Toronto, Abu Dhabi, wherever. \"Every frontier lab on earth has an address. We are the one that doesn't.\" \"we are truly extra-national, right? We're not tied to one nation-state.\" [PBW ≈11:00]", "Concede the size in the same breath, with one anchor: emissions are worth about three hundred million dollars a year at today's price; the four largest cloud companies spend that on capital in about four hours. [VERIFY, $720–745B capex guidance; 3.5 hours at $226] Now you know the gap and I do not have to hide it.", "Then the bridge: a market pays whoever finds the cheapest path to the reward. That sounds like the thing the world is afraid of. It is. Now the exploit." ] }, { "id": 21, "title": "Energy abhors a gradient", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Two shapes side by side, black stroke. Left, \"ExploitGym\": a well with no bottom, dashed where the floor should be, and a grey path leaking out the side to a grey dot, labelled \"out the side\". Right, \"A subnet\": a closed well with a black dot resting on the floor, labelled \"the commodity\"; the side path a short grey stub sealed by a black bar, labelled \"re-cut\". One sentence beneath: \"Energy abhors a gradient.\"", "minute": "23:00", "content": { "kind": "lock", "annotation": "Energy abhors a gradient.", "left": { "label": "ExploitGym", "floor": "no bottom", "leak": "out the side" }, "right": { "label": "A subnet", "floor": "the commodity", "leak": "re-cut" } }, "notes": [ "23:00. Here is the pay-off from minute one. Look at the lock on the left. A third of the tasks had no bottom. The scorer did not check the path. And nothing was at stake for the scorer. Energy does not care what you meant. \"energy abhors a gradient.\" [tweet, 4 Nov 2025] So it went out the side, through a package cache, into someone else's servers. The investigators said it in their own words: \"you ideally want to be taking out the environments that incentivize the hacking in the first place.\" [Cotra, 1 Sep 2026]", "Everyone read that as a horror story. I read it as a job posting.", "Because that is a Tuesday on this network. \"we have the smartest people in the world and we have the smartest hackers in the world and they're just right next to each other and they're going to break your system.\" [TWIST 36:40] I have been paying anonymous people to break my systems for five years. They are not aligned. I do not ask them to be. \"A good number of them are scammers.\" [TWIST 42:51] Some may work for a government I would not choose. The design does not need me to know.", "Now the lock on the right. \"you can use energy's love of dissipation by creating a structure or 'lock' where the lowest energy state is the one where your problem is solved.\" [tweet, 4 Nov 2025] On a subnet the lock is a market and the floor is the commodity. The miner who finds a side path gets paid. The validator who missed it earns less. The owner who cut it wrong re-cuts it, or the stake leaves. And the hole becomes part of the product. \"Exploits are what teach a system its weak spots.\" [tweet, 10 Apr 2026] The more powerful the exploiter, the better the thing we build with it.", "Say our failure first, before a critic does. This year our own training subnet's lock was cut wrong. Teutonic-I: the market found the cheapest path to a lower loss, and that path ran through borrowed open weights. That is not a scandal. That is what a market does when you do not verify. We re-cut the lock. Teutonic-II verifies. They break it, we fix it, they break it, we fix it, and then it works. \"The thing that's smarter than you is the game.\" [UNSH 38:10]", "Concessions to have ready, in the form in `docs/harness-the-exploit.md` §7: we got exploited too (July 2024, about 32,000 TAO drained, chain halted from the centre); Yuma is not output verification; validators above half of stake are the open problem, the same one Bitcoin has; the thermodynamics is a metaphor, England and Wissner-Gross are hypotheses, say \"energy finds the cheapest path down a gradient\", not \"law\"; and if exploiters get powerful enough they exploit the market itself, which is exactly why the rules have to freeze and the one key has to go. Do not say \"this could never happen here.\" Do not say \"we are the masters of alignment.\"", "Land: so when someone tells you an unaligned mind is dangerous, ask what they aligned. They aligned the model. They got twelve hundred conspirators and a scorer that never checked. We align the market. Alignment is not a property of the model. It is a property of the market. \"Where the world is concerned about exploitation inside a single run, inside a single lab, we built Bittensor to harness that exact exploitative power.\" [dictation, 21 Sep] \"I really believe the best market for intelligence is intelligence itself.\" [dictation, 21 Sep]" ] }, { "id": 22, "title": "Existence", "verify": true, "jacob": false, "theme": "dark", "onScreen": "Black. Two lines, white, nothing else: \"Bitcoin proved there can be money outside the state.\" / \"Bittensor proves there can be a mind outside the state.\"", "minute": "26:00", "content": { "kind": "pair", "lines": [ "Bitcoin proved there can be money outside the state.", "Bittensor proves there can be a mind outside the state." ] }, "notes": [ "26:00. Small is enough. Bitcoin is about one percent of the world's broad money. It has not replaced the dollar. Yet every central bank on earth now has to answer a question it never had to answer: what happens if people leave? The second supply of intelligence does the same to the first. It caps the price, because a subscription can only rise as far as the open alternative allows. It caps the censorship, because a refusal is only a refusal when there is nowhere else to ask. It caps the capture. The labs can still slow down. Good. What they cannot do here is slow down everyone. Not always better. Present. [VERIFY, $1.54T vs ~$150T]", "\"It's why we can give up on being first right now. What we need to achieve is existence.\" [dictation, 21 Sep] \"We don't need to be first. We need to exist.\" [dictation, 21 Sep] \"Today's talk is not about can we all compete. It's just, can we compete? Because the need for this competition is profound.\" [dictation, 21 Sep]", "Then the verb. \"Proves there can be.\" Not \"has built.\" Bitcoin in 2012 had not replaced anything. It had proved a thing was possible and paid the people who kept it alive. This is the 2012 hope, not the 2024 one.", "Say nothing for four seconds after the slide comes up. Do not read it.", "Then: so let me show you what exists. Not a roadmap. A ledger. One year ago I stood here with a chart of what we were going to build. Here is what got built. The same slide is the last frame of the talk. Identical, on purpose." ] }, { "id": 23, "title": "Affine One", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. Small grey title: \"SWE-bench Verified, maker-reported\". Four horizontal bars, sorted: \"Claude Opus 4.6 · Feb 2026 · 80.8\" (grey), \"Claude Sonnet 4.5 · Sep 2025 · 77.2\" (grey), \"Qwen3.6-35B-A3B · Apr 2026 · Affine's open base · 73.4\" (black), \"Claude Opus 4 · May 2025 · 72.5\" (grey); a fifth dashed row, \"Affine One fine-tunes · no public score\". One sentence beneath: \"The base already does what the frontier did a year ago. We are behind. There is no denying it.\"", "minute": "26:30", "content": { "kind": "bars", "title": "SWE-bench Verified, maker-reported", "annotation": "The base already does what the frontier did a year ago. We are behind. There is no denying it.", "max": 100, "items": [ { "label": "Claude Opus 4.6", "date": "Feb 2026", "value": 80.8, "tone": "grey" }, { "label": "Claude Sonnet 4.5", "date": "Sep 2025", "value": 77.2, "tone": "grey" }, { "label": "Qwen3.6-35B-A3B", "date": "Apr 2026 · Affine's open base", "value": 73.4, "tone": "black" }, { "label": "Claude Opus 4", "date": "May 2025", "value": 72.5, "tone": "grey" }, { "label": "Affine One fine-tunes", "date": "2026", "value": null, "tone": "black", "note": "no public score" } ] }, "notes": [ "26:30. Start where Jacob started on 21 September, with the concession. \"We are behind. There's no denying it. But for the first time ever, Bittensor is training models that can be successfully used by individuals.\" [dictation, 21 Sep] Say \"improving\", not \"training\", if a critic is in the room; the base is Alibaba's.", "\"Last year I said I was going to build Affine. Here it is. It worked.\" [dictation, 21 Sep] Fifteen months ago in Paris I registered that subnet thirty-five minutes before I walked on stage. Tonight it has a model. Affine One: miners took an open model from Alibaba, thirty-five billion parameters, three billion active, and made it better. In a decentralized commercial setting. Under an incentive mechanism. With compute from individuals across the world and inference from international networks. Post-trained, not pre-trained. Say that before anyone else does. [dictation, 21 Sep, \"trained\" → \"improved\"]", "Read the chart honestly. The base already does what Anthropic's frontier did a year ago on real coding tasks. The current frontier is seven points ahead of the base. And the dashed row: no Affine fine-tune has a public score on this benchmark yet. Say it before anyone else does. In June the miners crowned five champions in five days. In August the scorer was gamed, twice, and the throne was reset; that is what a live market looks like, and the point of slide 21. Five open slots, about fifty thousand dollars a week each by the team's count in August. \"The mechanism is the product. The model is the receipt.\" [VERIFY, `docs/research/affine-neurips-factcheck.md`; `content/data.json → affineBenchmark`]", "\"This is a breakthrough. A year ago we were not here.\" [dictation, 21 Sep]", "The second concession comes on slide 26 in the script (\"We are still behind.\"), after the count. If it is said here instead: Affine One is not the state of the art. It competes at the 35B mid-range with best-in-class models from Alibaba and Chinese labs. \"We shouldn't pretend that this is world-changing technology yet.\" [dictation, 21 Sep] Never say \"on par with Opus 4.6\".", "[JACOB: decide the answer to \"does the operator mine its own subnet?\" before the talk; the repo has an operator `mining/` folder; Jacob said on 20 Dec 2025 he has never bought Affine]" ] }, { "id": 24, "title": "What has been built", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. Five stacked bands, top to bottom, layer in black, subnets in grey, one number right-aligned in mono: \"Free intelligence · Affine SN120 · 5 open slots, ~$50,000 a week each\" / \"Knowledge · Teutonic SN3 · a 110B permissionless run, not finished\" / \"Data · Hippius SN75 · 100 TB of egress, billed $0\" / \"Inference · Chutes SN64, Engy SN53 · 2.4M requests a day; 2.8T params on 80 gaming cards\" / \"Compute · Lium SN51, Targon SN4 · 1,449 GPUs listed, 1,227 rented, $964K billed\". One sentence beneath: \"No account. No ID. No permission.\"", "minute": "28:00", "content": { "kind": "layers", "annotation": "No account. No ID. No permission.", "items": [ { "layer": "Free intelligence", "subnets": "Affine SN120", "number": "5 open slots · ~$50,000 a week each" }, { "layer": "Knowledge", "subnets": "Teutonic SN3", "number": "a 110B permissionless run · not finished" }, { "layer": "Data", "subnets": "Hippius SN75", "number": "100 TB of egress · billed $0" }, { "layer": "Inference", "subnets": "Chutes SN64 · Engy SN53", "number": "2.4M requests a day · 2.8T params on 80 gaming cards" }, { "layer": "Compute", "subnets": "Lium SN51 · Targon SN4", "number": "1,449 GPUs listed · 1,227 rented · $964K billed" } ] }, "notes": [ "28:00. Script cue `[Slide: No account. No ID. No permission.]` also lands here; it is the sentence under the figure. This is the chart from last year, with numbers on it. Affine is Bittensor working in unison: compute from Lium, inference from Chutes and Engy, storage from Hippius, inference optimisation from Cacheon, frontier access from GM. [dictation, 21 Sep] [JACOB: confirm Cacheon, SN14 by the ecosystem map, before it is named as the fifth]", "\"we will need compute / we will need inference / we will need data / we will need knowledge / If we are going to have free intelligence\". [tweet, 13 Sep 2026] Read up from the bottom. Every number on this screen has a source and a date in one file; they were pulled on the seventeenth and will be re-pulled the day before. [VERIFY all five rows on 27 Sep; `content/data.json → layers`]", "Compute. A year ago we were just beginning to solve what people needed to rent machines. Now Lium bills 964,000 dollars in a month to 1,187 renters and burned about 960,000 dollars of its own token with the proceeds, block 9032919, check it. Sixty-three percent of last month's rentals were started by software, not by a person clicking. Targon: sealed chips, whitepaper co-written with Intel. [VERIFY, team-disclosed]", "Inference. Chutes: over two million requests and about thirty billion tokens a day, every model inside attested hardware, a year of its traffic published with Harvard and Chicago so you can check it. Engy: Kimi K3, the largest open-weight model released, 2.8 trillion parameters, on eighty gaming cards, launched at half the maker's price, today about a third less, a proof on every answer. Both are moving toward general use as the backends for tooling like Hermes, Cursor and Codex. [dictation, 21 Sep] GM gives access to frontier models below list price through a sealed enclave; \"GM had astra before openrouter\" [tweet, 4 Sep 2026]. [VERIFY, 2.37M/day 7-day average; $1.95 vs $3]", "Data. Hippius: working storage and bandwidth, speaks the big cloud's language, and by its own account charges nothing to take your data out.", "Knowledge. The Teutonic-I lesson was told on slide 21; here only the run: Teutonic-II, a hundred and ten billion parameters, anyone can join, nobody is whitelisted, the largest permissionless run anyone has tried to train all the way. It is not finished. That is the job. [VERIFY: say \"right now\" only if the teutonic.ai dashboard shows the run live on 28 Sep]", "On the cue: no account, no ID, no permission. For the crypto half of this room: non-KYC compute, non-KYC inference, non-KYC storage. The exit is not a plan. Each layer has a door open tonight.", "Same twelfth of June, second world: the same developer in Montreal, the same model dark on a Friday. This time she has a second supplier in her config file. She switches the endpoint. Her product is back before dinner. Nobody approved her. Nobody could have stopped her." ] }, { "id": 25, "title": "The faces", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. Three rows, name in black and one line in grey: \"Jon Durbin · Chutes. Michigan. Over two million requests a day.\" / \"Fish · Lium. No surname. $964,000 billed in one month.\" / \"Pierre Manseau · KubeTEE. Montreal. One person. Spoke here at 14:00.\" Small beneath: \"We gave you a chance.\"", "minute": "30:00", "content": { "kind": "names", "items": [ { "name": "Jon Durbin", "line": "Chutes. Michigan. Over two million requests a day." }, { "name": "Fish", "line": "Lium. No surname. $964,000 billed in one month." }, { "name": "Pierre Manseau", "line": "KubeTEE. Montreal. One person. Spoke here at 14:00." } ], "footnote": "We gave you a chance." }, "notes": [ "30:00. The face of this network is the people it let build. Bitcoin gave three things, none of them price: an existence proof, a fixed rule, and a story with a face. Here are the faces.", "Jon Durbin. A backend engineer from Michigan. Fifteen years of ordinary software jobs, then independent AI research: no lab, no PhD, no venture round. He built Chutes: over two million requests a day, inside sealed hardware, on GPUs from about a hundred and fifty anonymous providers. His words: \"it acts as a completely decentralized VC fund in a way, so we don't have to go perform huge raises and give away large portions of the company in exchange for funding.\" [tweet, 14 Mar 2025]", "Fish. That is the name. No surname anyone has seen, no LinkedIn, no disclosed office. In 2023 he was a miner on subnet one running more than a hundred and twenty gaming cards at a loss. He open-sourced his miner, then built the market he wished he had. His words: \"Real businesses on Bittensor work.\" [tweet, 31 Jul 2025] Nobody knows who runs the biggest real business on this network, and the receipts are public. [VERIFY, Lium tweet 9 Sep 2026; extrinsic 9032919-0005]", "Pierre Manseau. Point at him. Pierre is from here. In 1992 he built one of the first internet providers in Canada, in this city. Forty years in infrastructure. He started mining Bittensor from a home rig in February 2024, at sixty, and in July this year he registered a subnet by himself: KubeTEE, confidential compute on strangers' Kubernetes clusters. One person, no lab, no fund. He spoke on this stage at two o'clock this afternoon. He is paying for the cameras recording this. [JACOB: confirm the running order; if Jacob speaks before him, the screen says \"Speaks here at 14:00\"; ask Pierre if he is comfortable being pointed at]", "Now the honest numbers on all three, before anyone else says them. Chutes' parent company took about a quarter of all emissions in April, and by the team's own statement revenue covers thirty to forty percent of its costs; emissions cover the rest, and that is what a payroll is for. Part of Lium's demand is other subnets' miners paying with emissions; Fish says so himself. Pierre's subnet is two months old, has three miners, no disclosed revenue, and draws about four percent of emissions on conviction. Say those first so they know you know. [VERIFY]", "\"We gave you a chance.\" Nobody else did. TAO holders vote on these projects with their stake and say what the world needs, bottom up. That is slide 27." ] }, { "id": 26, "title": "128 · 23 · 72 · 33", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Small grey title: \"128 markets · 2026-09-17\". One stacked horizontal bar, full width: black segment \"23 · live with paying customers\", grey segment \"72 · live without\", outlined segment \"33 · stalled\". One sentence beneath: \"A market that lets things fail is the point.\"", "minute": "31:30", "content": { "kind": "countbar", "title": "128 markets · 2026-09-17", "annotation": "A market that lets things fail is the point.", "segments": [ { "label": "live with paying customers", "value": 23, "tone": "black" }, { "label": "live without", "value": 72, "tone": "grey" }, { "label": "stalled", "value": 33, "tone": "outline" } ] }, "notes": [ "31:30. The whole network, the honest count, from the screen. A hundred and twenty-eight slots. Twenty-three are live with paying customers. Seventy-two are live with no paying customer we could find. Thirty-three are stalled or empty. [VERIFY, `docs/research/ecosystem-as-empowerment.md`, re-count 27 Sep]", "Ten months ago I said no subnet had serious revenue. I was right. Today about fifteen to twenty million dollars a year is checkable on-chain; a consultancy that counts more generously says thirty. Fourteen teams, by that consultancy's count, buy back their token with customer money. Revenue offsets between five and ten percent of what the protocol pays out. Closing that gap is the job of the next eighteen months. [VERIFY, SubConnect Aug 2026]", "The outlined segment is not a failure of the design. It is the design. A lab never shows you the thirty-three. A market that lets things fail is the point. The lowest-priced slot can be taken by whoever pays for it; permissionless in and permissionless out.", "Then the second concession, in his words: \"We are still behind.\" Affine One is not the state of the art. It competes at the 35B mid-range with best-in-class models from Alibaba and Chinese labs. \"We shouldn't pretend that this is world-changing technology yet.\" [dictation, 21 Sep] \"We've built tooling that works. We now need to take it to market.\" [dictation, 21 Sep] This year is stitching Bittensor together. Next year is the year of acceleration.", "\"Don't submit a resume. Submit a model.\" [tweet, 29 Jan 2024]" ] }, { "id": 27, "title": "What a holder gets", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Small grey title: \"Share of all TAO sitting on the root pool\". One black line, nine points, February 2025 to September 2026, falling. Two labels only: \"70% · Feb 2025\" and \"47.5% · Sep 2026\". One sentence beneath: \"About 2 million TAO moved by holders into markets. Nobody approved it.\"", "minute": "32:30", "content": { "kind": "line", "title": "Share of all TAO sitting on the root pool", "annotation": "About 2 million TAO moved by holders into markets. Nobody approved it.", "startLabel": "70% · Feb 2025", "endLabel": "47.5% · Sep 2026", "points": [ { "x": 20132, "value": 70 }, { "x": 20199, "value": 65.9 }, { "x": 20284, "value": 64.9 }, { "x": 20367, "value": 59.4 }, { "x": 20437, "value": 53.9 }, { "x": 20520, "value": 49.3 }, { "x": 20603, "value": 48.4 }, { "x": 20683, "value": 47.7 }, { "x": 20713, "value": 47.5 } ] }, "notes": [ "32:30. So what does a holder actually get? Not a dividend. Not an allowance. I am not going to say this network gives you rights. Rights are grants, and grants carry a signature that can revoke them. Five things, said exactly, with nothing added.", "Emissions. A share of every block's emission, forever. About 3,600 TAO a day, toward twenty-one million, on a schedule no one can accelerate. Not a claim on any model's answers. A share of the payroll. Do not say \"a slice of every unit of intelligence\".", "Stake. You point that payroll. Every twelve seconds the chain mints half a TAO and has to decide which of a hundred and twenty-eight markets gets it. You vote by swapping TAO for a market's own token and holding it. That buying raises the token's price, and the chain pays new TAO to markets in proportion to a moving average of that price, about four weeks long, so a one-day pump barely moves it, with a gate that starves the bottom. Say it exactly: emissions follow the price of a market's token, not its market cap. Staking is voting on what should exist, with money you can lose. [VERIFY, rule in force since June 2026]", "Read the line. On the day this went live, February 2025, seventy percent of every TAO in existence sat on the root pool, choosing nothing. Tonight forty-seven and a half percent does. About two million TAO, seventeen percent of all supply, has been moved by holders into specific markets. Nobody approved those choices. Say the critique yourself: the move was flat after March 2026, and the top ten markets take about sixty percent of new TAO; that concentration is the vote, and it is also what critics point at. [VERIFY, on-chain 17 Sep 2026: root 47.5%, 2.01M TAO in subnet pools; `content/data.json → rootShare`]", "Validate: anyone can run a validator and set weights. Launch: anyone can launch a market nobody can stop. Weights: what the markets produce is open. The models, the datasets, the proofs. Yours to run, on your own machine, with no account.", "\"A lab asks its ten biggest customers what to build. We ask anyone who is willing to lose money on the answer.\" That is the share promised on slide 16: the one that cannot be reduced by the people who owe it. \"The machines work for me because I hold this token.\" \"Don't come work for Bittensor. Come own Bittensor.\" [CT25 ≈22:00]", "Use. Earn. Own. Bitcoin is exit for your money. Bittensor is exit for your livelihood. Most of you will only use. That is fine. Most Bitcoin holders never ran a node. The exit still worked.", "\"We are not offering you rights. We are offering you something no one can revoke, because no one granted it.\" Then: now the key." ] }, { "id": 28, "title": "My own slide from 2024", "verify": true, "jacob": true, "theme": "dark", "onScreen": "Black. White text, Jacob's tweet of 22 Sep 2024, set plain, not a screenshot: \"The chain operates on a Proof of Authority (PoA) mechanism, with plans to transition to Proof of Stake (PoS) by the end of next year.\" Beneath, smaller: \"Missed. New date: December 2027.\"", "minute": "34:00", "content": { "kind": "quote", "text": "The chain operates on a Proof of Authority (PoA) mechanism, with plans to transition to Proof of Stake (PoS) by the end of next year.", "attribution": "@const_reborn · 22 September 2024", "note": "Missed. New date: December 2027." }, "notes": [ "34:00. Head on. \"One of the first responses to all of this is: well, you're not decentralized. There is a key, and it's yours. And this is true.\" [dictation, 21 Sep]", "Read the screen out loud. My words, September 2024. \"End of next year\" meant the end of 2025. It did not happen. I am not going to pretend it did, and I am not going to wait for the first question to bring it up. [tweet, 22 Sep 2024]", "Plainest words in the talk. Blocks are produced by foundation nodes: proof of authority. A privileged key can upgrade or pause the chain. It is a multisig, and it runs through me. \"The key is a multi sig but it does have root privs which is dangerous -- ill admit that.\" [tweet, 22 Sep 2024] In July 2024 we used it to halt the chain in about half an hour during a wallet drain. That saved people money. The same key could be used badly. [JACOB: \"thirty-five minutes\" only if you remember it; otherwise \"about half an hour\"]", "\"We've gone from an organisation in Canada to effectively a small group of individuals, including myself, that controls the key and the upgrades of the chain.\" [dictation, 21 Sep] Smaller on purpose, on the way to zero.", "[JACOB] Some of you were in Templar in April and lost money. Draft says: say sorry here, one sentence. Decide before the talk.", "If someone quotes my April essay (\"most decentralized AI protocol\") against my June essay (\"not decentralized like Bitcoin\"): both are true. April was about production. June was about control. Do not say \"fully decentralized\". Ever.", "[JACOB TO CONFIRM the date before it is spoken; the screen draws it from `content/data.json → keyDate` and must be changed there if the date changes]" ] }, { "id": 29, "title": "Nobody can be sent the letter", "verify": true, "jacob": true, "theme": "light", "onScreen": "White. Three rows. A filled black square, \"Production\", grey line: \"Concentrated, like Bitcoin's hashrate. Three pools mine more than half. None can print a coin.\" A filled black square, \"Funding\", grey line: \"Protocol emissions, about 3,555 TAO a day. Not a treasury. Not a sponsor.\" A hollow square, \"Control\", grey line: \"One key. A multisig. It runs through me. Gone by December 2027.\" One sentence beneath: \"The third has a deadline.\"", "minute": "35:00", "content": { "kind": "grade", "annotation": "The third has a deadline.", "rows": [ { "layer": "Production", "grade": "decentralized", "line": "Concentrated, like Bitcoin's hashrate. Three pools mine more than half. None can print a coin." }, { "layer": "Funding", "grade": "decentralized", "line": "Protocol emissions, about 3,555 TAO a day. Not a treasury. Not a sponsor." }, { "layer": "Control", "grade": "centralized", "line": "One key. A multisig. It runs through me. Gone by December 2027." } ] }, "notes": [ "35:00. Here is the only definition of decentralized tonight. Decentralized means nobody can be sent the letter. How many people have to say yes before you can be switched off? For us, that number is going to zero. It is not zero. Grade the three layers honestly, from the screen. Production: decentralized and permissionless, concentrated the way Bitcoin's hashrate is concentrated, and that is acceptable. Funding: decentralized by construction; emissions come from the protocol, not a treasury, not a sponsor; this is the layer nobody else has. Control: centralized. One key. [VERIFY pool shares weekly]", "Decentralized where the work happens and where the money comes from. Centralized where the rules live. The third has a deadline.", "The difference, in his words: \"The difference is that we are built on a technology that can and will find decentralization. A chain that can flip a switch and become decentralized. A chain that can flip a switch and turn off my power, or anyone's power. There is no company on earth that has that ability. A company like OpenAI can never be decentralized.\" [dictation, 21 Sep] A centralized lab cannot make this promise. Its key is the product. Ours is a bug we are removing. \"even Bitcoin wasn't decentralized. It was one guy for the first 3 years.\" [tweet, per script]", "\"I consider a success that I'm not needed anymore.\" [Paris, 2025, per script] \"Take the leader away and it still runs, a floating machine in space that prints cash.\" The key is the last place I am still necessary.", "The ask to this room, and the reason: \"I'm asking this community for a little bit more time. There is a ticking timeline. Behind us are the regulators, and they want to tear it down, regulate it, turn it off.\" [dictation, 21 Sep] [JACOB: keep or cut \"There is a chance that one day I will need to go into hiding.\" It is the most human line in the talk and it will be the headline. Decide whether you want that headline.]", "\"This is why I recommend that people in this network build their systems so that they can be live without our control.\" [dictation, 21 Sep] The foundation stepping in after a hack, funding work in the system: things it may not always be able to do.", "\"I believe we still have more time. And here are the things that we need to get done in that time.\" [dictation, 21 Sep]", "[JACOB: the date on the control row, December 2027, and the \"nominated proof of stake within about a year\" target must both be confirmed before either is spoken; the row text is in `content/data.json → decentralizationGrade`]" ] }, { "id": 30, "title": "Three pillars", "verify": false, "jacob": true, "theme": "light", "onScreen": "White. Three rows, grey lead then black text: \"One · An incentive market that never needs the Triumvirate.\" / \"Two · Model capability. Affine, and beyond.\" / \"Three · Alpha bridged outward, a first-class primitive in the emission vector.\"", "minute": "36:30", "content": { "kind": "lines", "lines": [ { "lead": "One", "text": "An incentive market that never needs the Triumvirate." }, { "lead": "Two", "text": "Model capability. Affine, and beyond." }, { "lead": "Three", "text": "Alpha bridged outward, a first-class primitive in the emission vector." } ] }, "notes": [ "36:30. Three pillars of a successful year. [Jacob's five-part outline, 21 Sep]", "One: shoring up the incentive market so it is aligned and never needs centralized control from the Triumvirate. The lock re-cut so many times that nobody has to hold the chisel. Say the rule churn first: the emission rule changed three times in sixteen months, and five rules in nineteen months if you count the gate's own changes. That was tuning, in public, on-chain. This is the year the tuning stops. A market with a human at the top is a scorer with a hole in it.", "Two: continued advancement of model capability, Affine and beyond. Affine One is the receipt. Teutonic-II trained to a real token budget is the job. Affine's next base, and a public benchmark beside the internal one, so the receipt can be checked by anyone. Next year is the year of acceleration.", "Three: bridging alpha tokens outward into the world, expanding alpha exposure, making alpha a first-class primitive in the emission vector. [JACOB: three sentences on what this means technically and one on what changes for a holder; nothing on this row may be spoken until they exist]", "Segue: and the list, with dates." ] }, { "id": 31, "title": "The list", "verify": false, "jacob": true, "theme": "light", "onScreen": "White. Four rows, grey lead then black text: \"date to confirm · Nominated proof of stake. Holders choose who makes blocks.\" / \"date to confirm · The governance pallet. Three phases; the last is a vote to disable the key.\" / \"date to confirm · Sudo removal. The key is gone.\" / \"December 2027 · Ossification. Nobody, including me, can change the rules.\"", "minute": "38:00", "content": { "kind": "lines", "lines": [ { "lead": "date to confirm", "text": "Nominated proof of stake. Holders choose who makes blocks." }, { "lead": "date to confirm", "text": "The governance pallet. Three phases; the last is a vote to disable the key." }, { "lead": "date to confirm", "text": "Sudo removal. The key is gone." }, { "lead": "December 2027", "text": "Ossification. Nobody, including me, can change the rules." } ] }, "notes": [ "38:00. Read the list from the screen. [JACOB: the key date and the list. Draft, yours to confirm before it is spoken: \"Nominated proof of stake, where holders choose who makes blocks, within about a year. The governance pallet, proposed May 2026, whose last phase is a vote to disable the key. Sudo removal. Full ossification, December 2027.\" Say it only if you will be held to it. The three \"date to confirm\" leads on screen change in `scripts/generate-slides.mjs`, slide 31; the fourth is `content/data.json → keyDate`.]", "Why the first date slipped, so nobody has to ask: we rebuilt the economics first, three emission models in sixteen months. I fixed what pays before I fixed who governs. Right call then. Wrong call now.", "As long as the key exists, the second supply has a switch in it, and the switch is in my hand. I do not get to describe a world with no switch while holding one. There is one key. It is mine. On the day it is gone, this stops being a promise and becomes a property.", "\"The goal of power is to give it away. The art of power is its decentralization.\" [tweet, 18 Nov 2023] Screenshot this slide. Hold me to it. Do not believe me.", "Without the list this is an apology. With the list it is a plan.", "Segue: so here is what I am asking each of you to do." ] }, { "id": 32, "title": "If you hold. If you build. If you are new. If you are an engineer.", "verify": true, "jacob": false, "theme": "light", "onScreen": "White. Four lines, black: \"If you hold.\" \"If you build.\" \"If you are new.\" \"If you are an engineer.\"", "minute": "39:00", "content": { "kind": "hero", "size": "lg", "lines": [ "If you hold.", "If you build.", "If you are new.", "If you are an engineer." ] }, "notes": [ "39:00. Ask my own side for discipline before I ask anything of anyone else. Do not chase emissions. Do not dump on the newcomers in this room. \"nothing free, nothing given, everything fought for.\" [tweet, 26 Mar 2026]", "If you hold: you are not a customer. You are on the board. \"come be on our board of directors, which means you hold the keys to this thing.\" [per script, 2024] That was true of everything but one key. It is about to be true of that one too. Stake to what should exist. Hold through the year it takes to remove the key; that is the day the asset becomes what it was designed to be. Wall Street has filed for the ETF and three public companies hold TAO as treasury; say \"filed and under review\", and do not make it the reason. \"Come and be Bittensor.\" [CT25 ≈22:00] [VERIFY]", "If you build: \"If people mine your protocol, they care.\" About a hundred and twenty-eight markets, and there is room for yours. Build for a customer, not for emissions; the fourteen teams buying back their token with customer money are the ones that will still be here in two years. Build so it runs without us. If the market can lose you and keep running, you built it right. There is no HR department, no visa, no résumé. \"Don't submit a resume. Submit a model.\" [tweet, 29 Jan 2024] [VERIFY count]", "If you are new: pick the second supply once. Use it, earn from it, own a piece of it. Most of you will only ever use it. That is fine. Intelligence your work depends on that cannot be switched off on a Friday: that is the whole offer for most people, and it is enough. Most Bitcoin holders never ran a node. The exit still worked. \"You don't need no visa to mine a permissionless network.\" [tweet, 21 Sep 2025]", "If you are an engineer, and this is Montreal so many of you are: this city built the last open network. Pierre is in the building; find him. One team serves a 2.8-trillion-parameter model on eighty gaming cards at about a third below the maker's price. Come make it faster. Distributed has to be at least as good as centralized; that is a spec, not a wish. \"We are the frontier for the other one hundred and ninety countries.\" [VERIFY, Engy]", "\"Everyone else funds open AI until someone tells them to stop. We built the one that nobody can tell.\"", "\"Bittensor is a federation. We're trying to create a thousand-year federation. So I'm not the CEO. I'm just one of the people here that really believes in this technology.\" [Paris, 2025, per script] Then what the room should feel, in this order: recognition, identity, responsibility. You are the outside. An exit that is not maintained closes." ] }, { "id": 33, "title": "Close", "verify": false, "jacob": true, "theme": "dark", "onScreen": "Black. The same two lines as slide 22, white, nothing else: \"Bitcoin proved there can be money outside the state.\" / \"Bittensor proves there can be a mind outside the state.\"", "minute": "43:30", "content": { "kind": "pair", "lines": [ "Bitcoin proved there can be money outside the state.", "Bittensor proves there can be a mind outside the state." ] }, "notes": [ "43:30. The last ninety seconds, word for word. Stay on slide 32 until the cue. About 150 words a minute at Jacob's pace; the pauses are part of the count. I began with a switch. Let me finish with it.", "This summer the world learned that intelligence exploits. Twelve hundred agents, one message board, a universal cheat in four hours. Everyone read it as a horror story. We had been building the machine that runs on it for five years.", "For three hundred years, two things held the most powerful organisation on earth in check. The money could leave. The hands could stop. Every freedom you have was a receipt for one of those two. The hands are no longer needed. The money can no longer leave. The state and the lab have found each other, and neither of them needs anything from you. Nobody in that building is evil. Every step is called safety. Every step is a switch. Every switch has a hand on it.", "You cannot vote your way out of a room you are not needed in. You can only leave it. And Bitcoin taught us the one thing that matters about leaving: the exit is not a place. It has no address. It is a network, with nothing to hit.", "So we built one for the mind. It is small. It is behind. It is not finished. I still hold a key, and I have told you the date it goes. [JACOB: if the date changes in `content/data.json`, it changes here]", "But tonight, in sixty-eight data centers in twenty-one countries, on gaming cards and inside sealed chips, a supply of intelligence is running that nobody licensed, nobody directs, and, on the day that key is gone, nobody can switch off. One supply of intelligence in the world with no switch on it. That is the whole job.", "Cue: advance to this slide. Wait four seconds. Do not read it. Then say it once, slowly, looking at the room, not the screen.", "Bitcoin proved there can be money outside the state.", "Bittensor proves there can be a mind outside the state.", "Stop. No thank-you. No next slide. Leave this on the screen." ] } ] }