What does David Sacks say about Anthropic and recursive self-improvement?

David Sacks argued that Anthropic is pushing harder toward recursive self-improvement than any other company, using its own models to find the efficiency gains that let it cut prices. He cited an August Anthropic blog showing 26% of model-development tasks already used this approach. He predicted a split between Anthropic and the AI-safety groups that fear a runaway superintelligence. He also said Anthropic needs these gains to stay ahead of cheaper open models.

Sacks called Dario Amodei's earlier pacing letter an advance alibi and an exercise in virtue signaling.
Sacks pointed to a Vox article quoting safety-focused nonprofits that questioned Anthropic's actions.
Sacks said Anthropic has built a large government affairs operation and warned against letting it win regulatory capture.

▸In their words

“There's no single company in the world that is doing more to race towards full RSI than Anthropic.” David Sacks · at 43:51 — open in the transcript →

Source: our own transcription of the episode. The full conversation is on All-In with Chamath, Jason, Sacks & Friedberg (2026-10-30). Read the whole episode summary →

▸More from this episode

Q1What did the All-In podcast say about Anthropic's S-1 and IPO terms?Q2Why does David Sacks say Anthropic's $80 billion loss is not a real loss?Q3What valuation does Chamath Palihapitiya expect for the Anthropic IPO?Q4What did David Friedberg say about Anthropic's customer growth and the AI value chain?Q5Why does Chamath Palihapitiya say AI companies are competing with US Treasuries?Q7Why did Google lose $500 billion in market value, according to the All-In hosts?Q8What did the All-In hosts predict for the 2026 midterm elections?

Transcribe your links. Paste a YouTube, Google Drive or podcast link and WhipScribe gives you the transcript. Transcribe a link →