Anthropic Files for an IPO Days After OpenAI, a Hot Jobs Report Triggers a $1 Trillion Chip Selloff, and Berkshire Bets on Housing

The rivalry between the two leaders of generative AI now has a finish line in view. Anthropic filed confidentially for an IPO on Monday, just ten days after OpenAI did the same, turning a contest over models and talent into a sprint to the public market. The surprise is who is in front. Anthropic’s latest private round, which raised $65 billion, valued the Claude maker at $965 billion, comfortably above the $852 billion OpenAI carried into its own filing. The company long cast as the cautious, safety-first underdog is, by that measure, now the more valuable of the two.

The gap underneath the valuations should worry OpenAI more than the valuations themselves. Anthropic has told investors its revenue run rate has reached roughly $47 billion, up from about $10 billion a year ago, and that it is closing in on profitability, a claim no other frontier lab can credibly make. After two years in which the market funded AI almost entirely on faith, a lab that can point to real revenue and a path to making money is a fundamentally different proposition from one still measured purely in promise.

The catch is the one shadowing the whole industry: that revenue is bought with enormous compute, and Anthropic is among the largest buyers of AI computing power on earth. Its profitability case rests on revenue outrunning a cost base that is still exploding. When the full S-1 eventually goes public, it will be the first hard look at whether that arithmetic holds, and not just for Anthropic. It will be the closest read yet on whether the economics of the entire AI build-out actually work.

For the banks, this is the richest fee event in years arriving twice in two weeks. Goldman Sachs, JPMorgan and Morgan Stanley are all circling the lead roles on two near-trillion-dollar listings filed barely ten days apart. Whichever lab opens its books to public investors first will set the template, and the valuation anchor, for every AI offering that follows it to market.

For two years the question hanging over Berkshire Hathaway was whether Greg Abel, who took the chief executive seat in January with Warren Buffett stepping back to chairman, would actually spend. On Monday he answered it, agreeing to buy homebuilder Taylor Morrison for $72.50 a share in cash, an equity value of about $6.8 billion and close to $8.5 billion including debt, a 24% premium to Friday’s close. Chief executive Sheryl Palmer stays on to run it.

The more revealing part is what Berkshire already owned. It makes the paint (Benjamin Moore), the flooring, the insulation and the structural components; it builds manufactured homes (Clayton Homes); and it runs one of the largest residential brokerages in the country (HomeServices of America). The one major piece it lacked was a production homebuilder. Taylor Morrison, a top-five builder, drops straight into that gap and turns a scattered set of housing bets into something close to a vertically integrated machine, from the lot to the paint to the mortgage at closing.

It is also a contrarian call. Most of Wall Street has spent the year treating high mortgage rates as a millstone on housing; Berkshire is paying a premium to own more of it, wagering that chronically tight supply matters more than the rate cycle. That is classic Buffett logic, now executed by his successor, and it carries a second message for the market: the Abel era will be acquisitive. Every large, dull, cash-generative business just edged a little further into play.

The 172,000 jobs the economy added in May, reported Friday, were not just double the 80,000 forecast. They were a problem for one person in particular: Kevin Warsh. The new Federal Reserve chair was widely expected to usher in rate cuts, and markets had bet he would. Then the data refused to cooperate. Unemployment held at 4.3%, wages rose 3.4% over the year, and revisions added 93,000 jobs to March and April. A labor market this strong does not need loosening.

So a chair many penciled in as a dove walks into his first meeting on June 17 with the hawks holding the better hand. By Friday’s close, futures priced roughly a 43% chance of a rate *hike* by December, a stunning reversal for a central bank investors spent last year expecting to ease. Equities did not wait for the meeting to react, as the next item shows. And for anyone financing a deal, the message is colder than the headline number: the rate cuts that half the market was quietly underwriting may simply never arrive.

Monday felt like a coronation. The S&P 500 closed above 7,600 for the first time, led by Nvidia and its new chip for PCs, and the exuberance was everywhere: Marvell Technology leapt 25% after Nvidia’s Jensen Huang mused it could be the next trillion-dollar company, and Hewlett Packard Enterprise matched that 25% jump on strong results. When chief executives start anointing the next trillion-dollar stock, the market is usually closer to a top than a bottom.

It was. Friday’s jobs report turned the tape inside out. The Nasdaq sank about 4%, its worst day since April 2025; the S&P 500 lost 2.6%; the Dow dropped 695 points. The pain pooled in the very chips that had led the climb. The Philadelphia Semiconductor Index posted its steepest one-day fall since March 2020, more than $1 trillion in value gone, with Nvidia off 6.2% and Intel, Micron, AMD and Broadcom down 8% to 13%. The VIX spiked 34%.

The lesson sits beneath the wreckage. The AI rally has quietly become a leveraged bet on interest rates: priced for cheap money and an easing Fed, it has the most to lose the instant that assumption cracks. One hotter-than-expected jobs print was enough to crack it, and the priciest names fell the hardest. A rally years in the making handed back a record week in a single afternoon, which is what happens when an entire market is leaning on one assumption and a single number knocks it loose.

OPEC+ is in the odd position of repeatedly promising more oil it cannot deliver. Meeting Sunday, the cartel’s core producers, led by Saudi Arabia and Russia, agreed to raise output targets by another 188,000 barrels a day from July, the fourth such increase since the Strait of Hormuz was effectively shut. On paper, the group has added almost 600,000 barrels a day of quota since April. In reality, its production has gone the other way.

With Hormuz blocked and Gulf exporters holding tankers back, actual OPEC+ output has collapsed to roughly 33.2 million barrels a day in April from 42.8 million in February, a fall of more than a fifth. So why keep lifting the targets into a closed chokepoint? Because quotas are about market share, not this month’s barrels. By raising them now, the group is staking its claim to pump the instant the strait reopens, rather than ceding ground to US shale and other rivals once supply normalizes.

The market itself is being pulled two ways at once. Brent sat near $93 and US crude near $91 by Friday, still up more than 6% on the week after fresh US-Iran clashes dimmed hopes of a deal, the supply-fear trade. Yet Chinese crude imports just slid to a ten-year low, the demand-worry trade. Oil caught between a shut chokepoint and a softening China is exactly the kind of two-sided risk that makes hedging, deal timing and energy financing so fraught right now, and it feeds straight into the inflation read the Fed is watching after Friday.

Two companies at opposite ends of the risk spectrum tested the IPO market this week, and investors mobbed both. Quantinuum, the quantum computing outfit spun out of Honeywell and Cambridge Quantum in 2021, priced above its range at $60 a share to raise about $1.68 billion, with a book more than 20 times oversubscribed; the stock opened at $68, touched $71.35, and settled roughly flat for a $15.7 billion valuation. The same day, Applied Aerospace & Defense, a contractor that private equity firm Greenbriar Equity Group built by bolting two suppliers together, priced at $20 to raise $650 million at a $3.4 billion valuation, its book around 10 times oversubscribed.

That a speculative quantum bet and a buttoned-down defense roll-up both cleared with that kind of demand is the real signal, more telling than any single megadeal. The IPO market spent two years thin and hostage to a handful of giant names. Breadth like this, mid-sized deals pricing above range and holding up in the aftermarket, is what a genuinely open window looks like, and it is a better omen for the SpaceX-and-Anthropic pipeline lining up behind it than another blockbuster headline would be.