Draft v3 · · What changed
Ilya Sutskever
Positions
What an institutional bet looks like
Sutskever's "age of research" reads as an institution: a hiring policy, two financings, a TPU contract and no product. SSI raised $3 billion in seven months; Reuters put NVIDIA's 2026 investment at about $5 billion.
The recipe is exhausted; the curve is not
Sutskever's NeurIPS line is the most-misread sentence in the scaling debate. He said "pretraining as we know it" will end; he did not say scaling will end. The cost-per-capability frontier he has not contested is…
What Sutskever won't say, and how it positions him
Sutskever rejects the scaling recipe as sharply as anyone and says least about what replaces it. LeCun has named JEPA, Sutton OaK, Gu state-space models; Sutskever names a missing principle and no architecture.
The "jagged" critique survives the leaderboard; the quantitative claim does not
Sutskever's qualitative claim that models "generalize dramatically worse than people" is sharp and survives the leaderboard. The quantitative claim is contested by a 2023–2026 record that has moved further than skeptics…
·Issue 1 · The Scaling Bet
deslop.media editorial. Issue 1, Draft v3, first published October 10, 2026. Reporting current through October 2026. Use the flags on this page to report an error or a disagreement. Changes to the issue are listed on the changes page.
Part of Issue 1 · The Scaling Bet