- Anthropic's IPO week arrives: pricing Wednesday, trading Thursday, demand already heavy
- Nvidia reports its first quarter without Jensen Huang tomorrow - the Street expects a beat
- The New York Times: Lutnick pushed to soften chip rules that overlap his family's business
- Alibaba launches Qwen4-Max: a 2.4T-parameter flagship it says beats GPT-6.1-Sol and Claude Opus 5.5 in coding
- A day after the Times story, Commerce moves to close the remote-access chip loophole
- The Fed's Michael Barr: AI could transmit stress through the financial system
- OpenAI expands free ChatGPT access to up to 200,000 researchers
AI briefing: this week
The current week of AI briefings, Monday to Sunday, in plain English. The weekly summary is on top; each day's full report is linked below. Updated every morning and again mid-afternoon on weekdays; a new week starts each Monday.

This week at a glance
The week opens with the biggest listing ever attempted. Anthropic priced its IPO on Wednesday at $83 a share - above its $70 to $78 range - upsized to a record roughly $81 billion at a $1.5 trillion valuation, and ANT is now two days into trading. One day earlier, Nvidia reported its first quarter since Jensen Huang's resignation - a blowout, with the numbers in Wednesday's entry below.
Off the exchanges, the New York Times tied Commerce Secretary Lutnick's push for softer chip rules to his family's business - clients of his former firm Cantor Fitzgerald run Southeast Asian data centers that sell Nvidia computing to Chinese customers, legally skirting the direct-sales ban. And Alibaba released Qwen4-Max, a 2.4-trillion-parameter flagship with a million-token context that it claims beats GPT-6.1-Sol and Claude Opus 5.5 in coding at about 60% lower cost - its own benchmarks, not independent tests.
By mid-afternoon the story had already moved: the Commerce Department issued a rule requiring export licenses for advanced American AI chips hosted in third-country data centers - the loophole at the center of the Times report - Federal Reserve Governor Michael Barr warned that AI could transmit financial stress through buildout financing, synchronized trading models and customer-facing agents, and OpenAI opened free ChatGPT access to up to 200,000 researchers, with 50,000 new grants awarded from November 23.
Tuesday was verdict day: Nvidia's first post-Huang results landed after the close and beat - consensus was around $109 billion, and the fourth-quarter guide was the real test for interim CEO Colette Kress (Wednesday's entry has the blowout numbers). Ahead of it, Trump and Xi took AI safety talks to the APEC summit on November 18-19 - the September crisis hotline becoming a working channel, with Monday's chip-export rule hanging over the table - and the administration is preparing to designate more than two dozen AI companies as national champions, a status carrying expedited chip export licenses, overseas government support and federal financing.
Also on Tuesday: Tencent launched Hy4 Flash, a budget sibling of its flagship model priced at $0.07 per million input tokens that it claims cost one-twentieth as much per task in its own tests, and Amazon disclosed that cybercrime groups used AI agents to attack AI applications themselves, hunting API keys inside exposed gateways like LiteLLM and Dify through its MadPot decoy network.
By mid-afternoon: Microsoft rebuilt Copilot around long-running agents at its Ignite conference - shared human-agent Workspaces, an unprompted Autopilot agent, Agent 365 analytics and new in-house MAI models - CNBC reported Anthropic's IPO was likely to price above its $70 to $78 range as order books closed at 4pm - it priced at $83 the next day, and Trump pushed AI cooperation with Quad leaders in Manila one day before his AI safety talks with Xi.
Wednesday brought the verdicts. Nvidia's first post-Huang quarter is a blowout - $115.8 billion of revenue, up 103%, a $130 billion fourth-quarter guide that assumes no China data-center compute, and a raised fiscal 2028 growth outlook, with Vera Rubin racks already shipping. Hours earlier, the board said it does not intend to bring Huang back, hiring Heidrick & Struggles to find a permanent CEO and standing up a new national-security oversight committee.
Also on Wednesday: Anthropic guided its IPO price to the top of its range or above ahead of that day's pricing - it priced at $83 - the Trump-Xi AI safety talks slipped a day to Thursday, and Waymo suspended driverless freeway rides across the US after California's DMV banned its vehicles from state freeways over an "imminent hazard to public safety".
By mid-afternoon: the Trump-Xi bilateral - and a trilateral with Putin - turned out to be Thursday's affair, with Xi using day one to call for global AI cooperation while Trump stayed silent at the welcome banquet; Anthropic upsized its IPO after books closed with roughly $500 billion of demand against at most $64.35 billion of stock - the final price came in at $83; and Colette Kress held Nvidia's first all-hands since Huang's exit - "a new era", with pay raises per WIRED - while some employees push for his return and the stock faded from its earnings pop.
Thursday is debut day: Anthropic priced at $83 - above the range - and upsized to a record roughly $81 billion, with ANT's first trade due this morning. In Shenzhen, the Trump-Xi-Putin trilateral ended in a four-point peace declaration - nuclear restraint, UN reform, oil supply, the ICC - with the AI safety agenda eclipsed. Meanwhile 15 national-security and AI figures urged the DOJ to probe what Nvidia knew about chip smuggling to China, and OpenAI shipped GPT-6.2 Astra and Sol as Microsoft priced its new MAI models to undercut. In the afternoon, ANT opened 34% above its IPO price and topped a $2 trillion market value, Trump told APEC the US opposes global AI regulation while Xi called for cooperation, and a short seller's China claim knocked AMD down about 8%.
Friday closes the loop. ANT ended its first trading day at $97.17, up 17% on the IPO price - a volatile fade from the 34% opening pop - and the record offering itself closed Friday. By mid-afternoon, day two went the other way: ANT fell as much as 12% intraday toward its IPO price as a broad AI sell-off knocked the Nasdaq down more than 4% and Nvidia more than 11%. On Capitol Hill, House Democrats opened a probe into whether Nvidia swayed the Commerce Department's Megaspeed chip-smuggling inquiry, with records and written answers due December 1. And the model race kept moving: Google put its Genie 4 world model in the Gemini API while OpenAI shipped GPT-Live-1.1, a voice model that listens while it speaks.
Saturday brought a different kind of headline: an Anthropic mathematician claimed a Claude-assisted proof of the Riemann hypothesis - 17 million lines of machine-checkable Lean code, announced a day after the IPO, with independent verification under way and the Clay Mathematics Institute still listing the problem as unsolved. The morning also caught up on three items from the week: Microsoft brought Nous Research's Hermes Agent into Teams, the Store and Azure (announced Wednesday), Gemini's Mac app learned to use your computer for you (Thursday), and California's attorney general said he will move quickly on an AI emergency shutdown power (Thursday).
Key stories of the week
- Anthropic priced the largest IPO ever at $83 a share - above its $70 to $78 range - and upsized to 980 million shares, raising roughly $81 billion at a $1.5 trillion valuation; ANT opened Thursday at $111.50, topped a $2 trillion market value intraday, and closed its first day at $97.17, up 17% - then fell as much as 12% intraday on Friday, back toward the IPO price, as a broad AI sell-off knocked the Nasdaq down more than 4%.
- Nvidia's first quarter since Jensen Huang's resignation is a blowout: revenue of $115.8 billion, up 103% from a year ago, a fourth-quarter guide of $130 billion - roughly $5 billion above consensus and assuming no data-center compute from China - and a raised fiscal 2028 growth outlook, with Vera Rubin NVL72 racks already shipping in its fastest product ramp ever.
- An Anthropic mathematician claims Claude helped him prove the Riemann hypothesis - the 1859 prime-distribution conjecture and a $1 million Millennium Prize Problem - releasing 17 million lines of machine-checkable Lean code a day after Anthropic's record IPO; the Clay Mathematics Institute still lists it as unsolved and independent verification is under way.
- The Trump-Xi-Putin trilateral in Shenzhen ended with a four-point 'Declaration Toward Peace' - nuclear restraint, UN reform, petroleum supply and the ICC - while the AI safety hotline agenda was eclipsed; Nvidia pressure continued as 15 security and AI figures urged the DOJ to probe what the company knew about chip smuggling to China.
- House Democrats on the Foreign Affairs Committee are probing whether Nvidia influenced the Commerce Department's chip-smuggling inquiry into Megaspeed - asking whether BIS head Jeffrey Kessler authorized Nvidia's general counsel to contact investigators, and why Commerce paused the probe and left the Justice Department to lead; records and written answers are due December 1, the day Jensen Huang briefs three House committees.
This week's briefings
- Nvidia reports after the close: the numbers matter less than the call
- Trump and Xi take AI safety talks to APEC, with Monday's chip rule hanging over the table
- The administration plans to designate two dozen AI 'national champions'
- Tencent's Hy4 Flash undercuts the market at seven cents per million tokens
- Amazon: AI agents ran cybercrime campaigns that hunt other AI apps for their keys
- Microsoft rebuilds Copilot around long-running agents at Ignite
- Anthropic's IPO is signaling an above-range pricing as order books close today
- Trump pushes AI cooperation with Quad leaders ahead of the Xi talks
- Nvidia's first post-Huang quarter is a blowout: $115.8 billion, guidance of $130 billion
- Same night, different verdict: Nvidia's board says Huang is not coming back
- Anthropic prices the biggest IPO ever today, guided to the top of its range or above
- Trump and Xi hold AI safety talks at APEC in Shenzhen today
- Waymo pulls driverless taxis off every US freeway after California's safety ban
- The Shenzhen wait: the Trump-Xi talks are Thursday, and day one belonged to Xi
- Anthropic weighs upsizing the biggest IPO ever: about $500 billion of demand for $64 billion of stock
- 'A new era': Kress faces the troops - and a workforce split over Huang
- Anthropic prices at $83 - above the range, upsized, and the biggest IPO ever
- The Shenzhen trilateral delivered a four-point peace declaration - AI was eclipsed
- Fifteen security and AI figures urge the DOJ to probe what Nvidia knew about chip smuggling
- OpenAI answers Ignite: GPT-6.2 Astra and Sol, as Microsoft prices its MAI line to undercut
- ANT opens at $111.50 - 34% above the IPO price - and tops a $2 trillion market value
- Trump tells APEC the US opposes global AI regulation; Xi calls for cooperation
- AMD slides about 8% on a short seller's claim its firmware reached a blacklisted Chinese company
- ANT's first day closes 17% above the IPO price - a volatile fade from a 34% opening pop
- House Democrats probe whether Nvidia swayed the Commerce Department's Megaspeed chip-smuggling inquiry
- Google opens Genie 4, its world model, to developers through the Gemini API
- OpenAI ships GPT-Live-1.1: a full-duplex voice model that listens while it speaks
- Day two: ANT gives back nearly all of its day-one gain as the AI sell-off deepens
- An Anthropic mathematician claims a Claude-assisted proof of the Riemann hypothesis - a claim, now under review
- Microsoft brings Nous Research's Hermes Agent into Teams, the Store and Azure (catch-up from Wednesday)
- Gemini's Mac app can now use your computer for you (catch-up from Thursday)
- California's AG says he will move quickly on an AI emergency shutdown power (catch-up from this week)