Anthropic's S-1: three investors, three reasons to read it

Sarah Guo shared the SEC filing with the line "read for yourself (not investment advice)." She wants to see whether a company with a strong mission changes how it operates under quarterly public scrutiny, and whether public opinion shifts once retail investors own part of the upside. @saranormous Her first observations were an inflection in operating profit, a strong emphasis on novel discovery and bio, a mismatch between consumption-based revenue and long-term compute obligations, and limited shareholder influence. @saranormous She singled out the "20 pages of chosen case studies starting 220." @saranormous

Jason Lemkin said: "Read the filing, not the headlines." He linked his SaaStr write-up, 5 Learnings from Anthropic's S-1, which highlights 900% NRR, $6.6M in revenue per employee and "$41M to $16.3B in half a year." @jasonlk The write-up is his own material.

Aaron Levie pointed to Ben Bajarin's post on the filing's Fortune 500 data. According to Bajarin, Anthropic's revenue from those customers averages less than 0.1% of their expenses. @BenBajarin Levie called this "the key stat" and sees no reason it can't grow 10–50X. He argues the big jump will come as enterprises rewire their processes for agents. @levie His takeaway: the S-1 shows "the strength of the frontier lab business model" and "how early we are in the build out of AI." @levie

Patrick O'Shaughnessy called it a "good day to rewatch" his conversation with Anthropic CFO Krishna Rao. @patrick_oshag It was Rao's first podcast appearance. Topics include how he allocates compute across Trainium, TPUs and GPUs, what investors misunderstand about model companies, and platform vs. application strategy. @patrick_oshag

Amjad Masad: Marc Andreessen on chaotic technology adoption

Masad called Andreessen's Replit and the dawn of the self-driving company "a fantastic piece placing this shift in the history of computing." @amasad The essay argues that a chaotic entrance is one of the best signs a technology will change the world. Andreessen cites Andy Grove's question of who will "let chaos reign, and then rein in the chaos." He compares agents to PCs, which people brought to work whether IT allowed it or not. @pmarca Masad has a stake here: the essay praises Replit's own "Agent Habitat." @pmarca

OpenAI's claimed refutation of ETH

Ryan Williams posted a direct link to OpenAI's preprint. It claims 3SAT on n variables can be solved in about exp(n/(log n)^{1/2+o(1)}) time. @rrwilliams Tim Sweeney shared it with "Wow!" @TimSweeneyEpic Martin Casado called ETH being false "a stunning result." @martin_casado It is still a claim, not a peer-reviewed result.

Shorter mentions

• Clem Delangue shared Liza Dwoskin's Washington Post story. It reports that ex-DOGE cyber researchers used AI agents to get into Anthropic's systems in under 12 hours, including Slack channels where its most sensitive models are discussed. @lizzadwoskin His takeaway: anyone who thought "concentrating AI behind closed doors is safe" should "update your priors." @ClementDelangue
• Ryan Hoover framed YC's interview with Farza as a modern-day version of Bezos's "working backwards" press release: start from the launch and ask what the Instagram caption says. @rrhoover In the interview, Farza tells YC's Aaron Epstein why founders should launch constantly and how he scripts videos that still feel off the cuff. @ycombinator
• Balaji Srinivasan: on a Farcaster stream, the host read out Balaji's prize brief for a 59-second video on civility, "starting with Norbert Elias's book, The Civilizing Process." The brief's argument is that people who are polite to each other can cooperate at larger scale and build more valuable things. GM Farcaster