AGI Timeline — Situational Awareness (Aschenbrenner) + the Deployment Brake
conceptAI & Compute · The Blackmail Network
The famous 'AGI by 2027' bet — and the study that says slow down.
Who they are
Leopold Aschenbrenner (the Situational Awareness essay), checked against McKinsey and METR.
What they do
He argues machines hit human-level intelligence around 2027 and automate most work soon after. The money side is real and even bigger than he said ($5-8 trillion in data centers by 2030). But a controlled study found developers were actually 19% SLOWER using AI while feeling 20% faster — so 'all jobs gone in five years' is hype; more like 2029-2032.
How it works
Follow the trend lines in computing power and efficiency for the 'yes'; watch real-world deployment (not test scores) for the reality check.
Why it matters
It's the timeline under the whole automation story — and it feeds the 'life is already a rigged lottery' reading: when work disappears, the sorting machine underneath stops being hidden. Noted 13 Sep 2026: the engine now also holds a separate entry on METR as an organisation. That doesn't change anything here — the 19%-slower study stands or falls on its method and whether others can repeat it, not on who funds the people who ran it.
What's new (Aug–Oct 2026)
Aug 17, 2026 — Nvidia will invest $1.5bn in SB Energy and provide up to $105bn in credit for a data center growing from 4.25GW to 8GW. (techcrunch.com)
Aug 26, 2026 — Nvidia reported revenue of $96.2bn for the quarter ended July 26, 2026, up 18% from the prior quarter and 106% from a year earlier (the Guardian said $92bn was expected); data-center revenue was $89bn and operating income $63.7bn (ynet). (nvidianews.nvidia.com ; ynetnews.com ; theguardian.com)
Sept 28, 2026 — Nvidia said Anthropic has signed up for $180bn worth of computing infrastructure, and Nvidia sped up its share buyback by $150bn to a total of $235bn. TechTimes put Nvidia's Vera Rubin chips at "$40 billion per gigawatt". (techtimes.com ; gurufocus.com)
The engine's record — word for word
**In plain terms:** the dated AGI-forecast spine the engine had no node for. Leopold Aschenbrenner's Situational Awareness (2024) argues AGI by 2027 is strikingly plausible, extrapolating compute (~0.5 orders-of-magnitude/year) plus algorithmic efficiency plus 'unhobbling' toward a drop-in remote worker, then an intelligence explosion (2028-2030) and a national-security 'Project'
[1]. SUBSTRATE CHECK (the money is real and exceeded): McKinsey projects roughly $6.7 trillion of worldwide data-center capex by 2030 — about $5.2 trillion for AI workloads, up to $7.9 trillion in the accelerated scenario
[2]. THE BRAKE (capability is not deployment): METR's 2025 randomized controlled trial found experienced open-source developers were about 19% SLOWER using AI tools, while believing they were ~20% faster — an illusion-vs-reality gap
[3]; agentic organizational absorption lags the benchmark curve. Held forceful dated lean: the DIRECTION (labor automation) is credible; 'all jobs in five years' is forecast-inflated, adjusted median ~2029-2032. Feeds the sorting-state/lottery reading (see maxwells_demon_sorting_state) — automation strips the meritocracy camouflage over the sort. Name no holder. [2026-09-13] The METR citation this node rests on is now accompanied by an institutional record for the same body (metr_frontier_risk_report_2026). The engine's use of METR's 19%-slower RCT as its deployment brake is unchanged by that addition: the study stands or falls on its methodology and replication status, not on its author's funding posture. Recorded so the dependency is visible rather than implicit.
[Live pass Oct 6 2026] 2026-08-17 — Nvidia to invest $1.5bn in SB Energy and provide up to $105bn in credit for a data center scaling from 4.25GW to 8GW. (1 source) (
techcrunch.com) | 2026-08-26 — Nvidia Q2 FY2027 (quarter ended July 26, 2026): revenue $96.2bn, up 18% q/q and 106% y/y (Guardian: vs $92bn expected); data-center revenue $89bn; operating income $63.7bn (ynet). (
nvidianews.nvidia.com ;
ynetnews.com ;
theguardian.com) | 2026-09-28 — Nvidia: Anthropic has contracted $180bn in infrastructure value; Nvidia accelerated its buyback by $150bn to a total authorization of $235bn. TechTimes gives Vera Rubin at "$40 billion per gigawatt". (
techtimes.com ;
gurufocus.com)
Follow the trail
Walk this on the live map →