Anthropic: AI brought up AMD GPUs in 3 days
Claude Mythos Preview · IPO roadshow · vs. 6 to 12 months
Anthropic's share sale closes, and its price support ends. Nvidia's board faces a White House call to bring back Jensen Huang. Wall Street slides into a correction on AI doubts. And the Lean community certifies a Riemann hypothesis proof built with Claude.

ON AIR
ACN · AI Coding News, the podcast. New every weekday.
Sponsor
Anthropic's share sale closes, and its price support ends. Nvidia's board faces a White House call to bring back Jensen Huang. Wall Street slides into a correction on AI doubts. And the Lean community certifies a Riemann hypothesis proof built with Claude. From AI Coding News, this is your Weekend Report for the week of November 16th.
We start with Anthropic. Its record IPO is now closed, and a source of support for the stock is gone. Anthropic began trading on Nasdaq on Thursday, November 19th, at an IPO price of eighty three dollars a share. At that price, the company is valued at about one and a half trillion dollars, fully diluted. On Friday, November 20th, Anthropic closed the offering. Its underwriters bought every extra share their option allowed. That option, called a greenshoe, is how banks help steady a new stock. It lifted the total raised to ninety three and a half billion dollars. Reuters reports the timing has a cost. Morgan Stanley, the bank steadying the stock, can no longer buy shares in the market if the price falls below eighty three dollars. Morgan Stanley declined to comment. For Berenberg's Carsten Hesse, investors are losing an important source of buying support. During that day's trading, the shares fell to within a few dollars of the IPO price. Anthropic's filings list hundreds of billions of dollars in spending commitments, mostly for computing power. Not everyone cheered the listing. Protesters outside Nasdaq on its first day of trading called for a freeze on AI development. And other big listings show the risk. SpaceX, which went public in June, closed the same day below its own IPO price. Why it matters: teams building on Claude can now follow Anthropic's revenue, losses and spending in its public quarterly filings.
Next, Nvidia. The White House and, reportedly, regulators turned up the pressure on Nvidia this week. Nvidia's board ruled out reinstating Jensen Huang as CEO on Tuesday, November 17th. On Friday, November 20th, President Trump said he would like to see Huang make a big apology and go back. Backers in tech want him back too. The same day, Reuters reported that enforcement staff at the Securities and Exchange Commission, the SEC, are examining potential accounting issues at Nvidia. The SEC declined to comment, and Nvidia did not respond. An examination is not a finding of wrongdoing. Meanwhile, interim CEO Colette Kress is changing course. Nvidia will stop selling its H200 AI chips in China. A new board committee now oversees national security and export controls. Kress also plans to return spare cash to shareholders. One plan holds. Kress has recommitted Nvidia to open AI models. On the earnings call, she did not directly address Nvidia's twelve point nine billion dollar deal to buy Hugging Face, a hub where developers share open models. The board is looking at internal and external candidates for a permanent CEO. Why it matters: developers who get open models from Hugging Face now depend on Nvidia's interim leaders to see the Hugging Face deal through.
When we come right back: the AI selloff that pushed Wall Street into a correction. Your coding agents say they're done. Are they? Levr is the tracker and control plane that verifies and gates your coding agents' work against the plan. Levr holds the plans, issues, tests and gates for every agent doing the work. And work only moves forward when the gates are verified, not when the agent claims it's done. Levr. Try it free at levr.one. Welcome back.
Next, the market. Anthropic's record listing did not stop the AI selloff. By the close on Friday, November 20th, the selloff had pulled in the broader market. That day, Anthropic's shares gave back much of the gain from their first day. The S&P 500 and the Dow fell into a correction, more than ten percent below their recent highs. The Nasdaq erased its gains for the year. Chip stocks fell hardest. The Philadelphia Semiconductor Index dropped nearly eleven percent in one day, led by Nvidia, Broadcom and Micron. Strategists point to growing unease over whether the AI investment boom can sustain itself. Treasury Partners' Richard Saperstein calls it an AI repricing, as data center buildouts slow. Baird's Michael Antonelli questions whether AI can keep spending unlimited amounts of money and still support the stock market. Even so, Antonelli doesn't think the bull market is over. That bull market began in October 2022. For one wealth manager, the Anthropic listing is the show me moment for the AI story. The selloff spread overseas, and into bitcoin and gold. Why it matters: teams relying on AI coding tools buy from vendors that now raise money from investors who doubt AI spending.
And finally, the proof. The Lean check of a Riemann hypothesis proof, built with help from Claude, now carries its community's certification. On Friday, November 20th, Anthropic mathematician Levent Alpöge posted what he believes is a proof of the problem, found working with Claude. The problem dates to eighteen fifty nine. Alpöge posted the proof in Lean, a programming language whose checker verifies every step. On Saturday, November 21st, the Lean and Mathlib communities issued a joint statement. They certify that both the statement and the proof are correct. The Lean Focused Research Organization and the Mathlib Initiative endorsed it. MIT mathematician Andrew Sutherland had already compiled the proof on his own. What the check does not settle is understanding. Lean confirms every step follows. It does not explain the new ideas. Sutherland says there is much work to be done to understand the mathematical details. Alpöge has committed to an explanation written for human readers, which is not yet available. A rival claim adds a twist. Stanford's Ravi Vakil posted a proof by postdoc Alessandro Masullo that Vakil describes as all human. That proof is not in Lean, and it is too early to say whether it holds. Why it matters: teams giving AI agents long, hard tasks can see automated checks, like Lean's proof checker, give a proof built with Claude its first formal approval.
The bottom line: Anthropic's record listing did not calm the AI selloff, and strategists point to growing unease over whether the AI investment boom can last. That's your Weekend Report for the week of November 16th. Sources are linked in the description. ACN is brought to you by Levr, the tracker and control plane that holds your coding agents' work to the plan, at levr.one.