Public markets pulled back from AI hardware this week while the labs kept shipping and private money kept flowing. The Philadelphia semiconductor index fell 11.4%, its worst week since March 2025, after a Chinese memory maker announced a denser chip design and hot inflation data raised the odds of rate hikes. Anthropic told prospective IPO investors its third-quarter revenue reached $16.8 billion, Google and Meta released new frontier models, and Replit switched its default model from Anthropic to OpenAI. Eight US states sued OpenAI over ChatGPT's safety, Modal and Ramp raised large private rounds, Ottawa added $500 million to its regional AI program, and Kroger said it will cut about 3,000 corporate jobs.
1. Chip stocks had their worst week since March 2025
The Philadelphia semiconductor index fell 11.4% in the week to October 16, its biggest weekly decline since March 2025, after eight straight losing sessions that erased about $2 trillion in market value, with South Korea's chip-heavy benchmark falling nearly 8% in a single day on October 15. Reuters attributed the selling to hot US inflation data that raised the odds of Federal Reserve rate hikes and to doubts about whether heavy AI spending will pay off, and chip stocks recovered some ground on Monday morning.
2. A Chinese memory chip design helped set off the selloff
ChangXin Memory Technologies (CXMT), China's leading memory chipmaker, said over the October 10-11 weekend that it had made progress on a denser memory design and plans to launch a server memory module built on it before the end of 2026, and on October 13 SK Hynix fell more than 10% and Samsung Electronics 7.4% in Seoul while Micron fell 5% in New York. The design targets the server memory market where AI demand has pushed prices to records, but it does not cover the high-bandwidth memory that sits inside AI chips, where SK Hynix and Samsung remain well ahead.
3. Anthropic told investors its quarterly revenue hit $16.8 billion
At a meeting with prospective IPO investors on October 14, Anthropic said third-quarter revenue was $16.8 billion, up from about $11.6 billion in the second quarter, and that it was profitable on an adjusted basis, with the Financial Times reporting $1.70 in revenue for every dollar spent on computing power. Reuters reported that some investors were disappointed that sequential growth slowed to 45% after revenue more than doubled the quarter before, ahead of a listing reportedly planned for November.
4. Google and Meta both shipped new frontier models
Google made Gemini 4 Argon broadly available to developers and businesses on October 13 at an introductory $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 on December 1, and the model ranks first on the Vals Index of finance, legal, coding and tax tasks. A day earlier Meta released Muse Wonder and Muse Wonder Full, its most capable models, built for multi-step agent tasks with a one-million-token context window and available through the Meta Model API, which gives buyers more frontier options from outside the two largest labs.
5. Replit made OpenAI its default model in place of Anthropic
Replit chief executive Amjad Masad said at TechCrunch Disrupt on October 16 that OpenAI is now the preferred partner powering Replit Agent, calling it "a decision based on both cost, partnership, and performance," after Replit had been one of Anthropic's top five customers until recently. Claude models remain available to Replit customers, and a person familiar with the matter told TechCrunch the Anthropic partnership is still in place, which shows large customers are now switching default models on price and performance.
6. Eight US states sued OpenAI over ChatGPT's safety
Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia each filed suit against OpenAI in state court on October 14, alleging under consumer protection laws that ChatGPT harmed children and that the company misled parents about its safety, with Iowa seeking up to $40,000 per violation and a halt to testing of advanced models until safeguards are in place. OpenAI said it has built age-specific protections, parental controls and break reminders into ChatGPT, and the suits follow Florida's in June and open investigations in Alabama and California.
7. Private investors kept funding AI infrastructure
AI cloud provider Modal raised a $750 million Series D on October 14 at a $15.75 billion valuation, more than triple its value after its May round, and its chief executive told Axios annualized revenue has passed $1 billion. Expense management company Ramp raised $1.85 billion at a $60 billion valuation on October 15 as it expands its AI products, so private rounds held their size in the same week public chip stocks fell.
8. Ottawa added $500 million to its regional AI program
AI Minister Evan Solomon announced on October 13 a $500 million top-up over five years for the Regional Artificial Intelligence Initiative, bringing the program to $700 million, delivered through Canada's seven regional development agencies to support AI adoption and commercialization in sectors including health, energy, agriculture and manufacturing. The same day Canada and the UK renewed their AI security partnership with a focus on sharing information about frontier AI risks, and BetaKit notes business AI adoption in Canada has risen from 6% in 2024 to over 19% in 2026.
9. Kroger plans to cut about 3,000 corporate jobs
Kroger will cut about 3,000 corporate jobs over the next three years, chief executive Greg Foran wrote in an October 12 memo, saying the grocer has "more layers, more handoffs and more overhead than we need" and will use technology to reduce duplicative work. Kroger recently appointed a chief data and AI officer, and the cuts came the same week Netflix said it would eliminate about 1,000 jobs, roughly 6% of its workforce.
