The Morning Download: Anthropic and OpenAI Diverge on the Road to Profit

What's up: The AI Cold War; Meg Whitman on the 'AI gold rush;' CoreWeave's revenue doubles; Softbank sells Nvidia stake; politicians link rising electricity prices to data centers.

Good morning. Anthropic expects to break even in 2028, with rival OpenAI projected not to turn a profit until 2030, according to financial documents reviewed by the Journal's Berber Jin.

Questions about whether AI spending will generate enough revenue to cover massive infrastructure costs have intensified in recent weeks, as investors punish cloud giants for heavy capital outlays and business customers press for clarity on their vendors’ path to financial stability.

Now we know a bit more.

The documents, shared with investors this summer, reveal two distinct road maps with Anthropic taking a more cautious path, focused on growing sales among corporate customers, who account for about 80% of its revenue.

OpenAI, by contrast, plans far heavier spending on chips, data centers, and talent as it pursues broader consumer and enterprise markets — and expands into other costly ventures such as video and image generation.

The ChatGPT-maker expects to burn through roughly 14 times as much cash as Anthropic before turning a profit in 2030.

“Every dollar we invest in AI infrastructure goes to serving the hundreds of millions of consumers, businesses and developers who rely on ChatGPT to get more done,” an OpenAI spokesperson tells the Journal.

News Corp, owner of The Wall Street Journal, has a content-licensing partnership with OpenAI.

More--much more-- on the companies, individuals and nation-states driving AI infrastructure spending below. 

CoreWeave reports doubling of revenue from AI boom

Cloud provider CoreWeave, the patron saint of borrowing massive sums to pay for AI infrastructure, saw revenue more than double to $1.36 billion in the third quarter, WSJ reports. The results reflect an “unprecedented demand for AI,” the company said Monday.

Michael Intrator, CoreWeave’s co-founder and chief executive, last week played down fears of a data-center bubble, while he played up the power of debt.

I mean, that's the nature of the business, right? Like we are building infrastructure, delivering infrastructure. The way that we built our business and we're able to scale our business to such enormous scale so quickly is that we used debt, because debt is the correct way to do this.

Speaking of debt. JPMorgan Chase in a Monday report said that the AI boom is so capital-intensive that it will reshape global finance with every major source of money, from bonds to private credit, addressing the gap, Bloomberg reports. The bank estimates that the tab could be between $5 trillion and $7 trillion.

But is it all just too, too much? As a CEO of HP and eBay during the 1990s and 2000s, Meg Whitman has seen her share of booms and busts. On stage last night at WSJ Leadership Institute's Board of Directors Council Summit in Palm Beach, Fla., Whitman, who currently sits on CoreWeave's board, offered her take on the AI boom.

From where I sit in this industry now, the demand is insatiable, like the demand for compute, the demand for energy, the demand for engineers is unlike anything I've seen since 1998... And you can go back to the mainframe, to the PC, to the internet, to e-commerce, to the iPhone, to the cloud. These are all ginormous changes. This is the biggest I've ever seen, and it is the fastest moving I've ever seen... I think if I was running one of these big energy companies or or infrastructure companies, I'd be doing exactly what they're doing.