OpenAI raises less than half its target as AI funding cools

Anthropic’s IPO filing details voting power for seven founders

A Wall Street Journal technology newsletter by Erin Woo reports that lenders and investors have become more cautious about financing the global AI build-out, just as Google reported that its own AI offerings are bleeding growth from its cash-cow search ads. Confidence that hyperscalers such as Google and Meta could fund much of the build-out out of their profits took a blow Wednesday, with monetization of AI-style queries still a work in progress.

Google’s market capitalization shed $500 billion in the stock’s worst day ever, and indexes fell further as Intel and Amazon reported results that concerned investors.

Lenders and investors, meanwhile, are getting choosier about where to place their bets and are demanding tougher terms, according to Wall Street Journal reporters Peter Rudegeair, Robbie Whelan and Jack Pitcher. Eagerness to crowd into the AI trade in any way possible has cooled as banks have raised concerns over exposure and data-center delays, while climbing interest rates have made debt offerings less attractive.

Neo-clouds, the newer AI cloud providers, have been especially affected by the pullback, forcing some companies on the IPO track to delay or downsize their offerings.

OpenAI has also had to pare its ambitions in a private funding round ahead of an expected IPO next year, according to Berber Jin reporting for the Wall Street Journal. The company had been seeking $30 billion at a $1.4 trillion valuation but has raised less than half of that at a $1.25 trillion valuation.

Anthropic has filed public paperwork ahead of what may be the largest IPO ever, and its prospectus detailed the voting power given to the company’s seven founders at the expense of would-be investors. In the filing, the AI company did not explicitly list a “total addressable market” figure but suggested it is staring down a $40 trillion-a-year opportunity if it captures the entirety of demand.

Microsoft offered voluntary buyouts to 60,000 employees, about one quarter of its staff, facing pressure to shrink head count as it invests in AI.

In a separate item, the newsletter reported that innovators are using techniques refined by the pharmaceutical industry to create “bioidentical” cocoa-bean cells, leading to the first bars made from experimental, lab-grown chocolate. Other companies, it said, are scaling up processes for roasting or fermenting agricultural waste products to make them good-enough substitutes for the primary ingredients in chocolate.

On the hardware side, the newsletter said Apple “for years bullied flash-storage makers into steep discounts,” until AI hyperscalers bid up prices and gobbled up supply. A joint venture between Anduril and Qualcomm aims to make rugged laptops for military and government use, with materials sourced from the U.S. and allies.

Thrive Holdings, a Kushner-led firm that acquires older businesses and infuses them with AI, agreed to acquire ADT in a $7.6 billion all-cash deal with an enterprise value of $15.3 billion. The newsletter described the transaction as the latest move in his aggressive push to back businesses far removed from the heated AI race.

The roundup also noted that Google, Meta and other tech giants are ramping up resistance to ICE as midterms near, in what the newsletter described as a bid to avoid Democratic scrutiny.

None of the developments settles who will pay for the build-out. As the newsletter put it, the question is “a trickier question than it seemed just a few days ago.”