The competition to lead artificial intelligence may be decided as much by financing as by superior chips, according to Fortune's CEO Daily newsletter. This week brought fresh evidence that the rivalry among chipmakers extends well beyond technology into debt deals, equity incentives and customer-financing arrangements.

Broadcom has reportedly held talks to raise around $30 billion in debt to help OpenAI buy chips the two companies are developing together. The discussions follow a similar $35 billion deal involving Anthropic, and another $60 billion financing package is reportedly in the works for Anthropic and others.

Nvidia recently partnered with six finance firms to mobilize more than $500 billion in third-party capital—money committed by outside investors rather than drawn from Nvidia's own balance sheet—to fund customers' AI infrastructure, and it holds direct stakes in AI companies valued at almost $100 billion. AMD, meanwhile, has offered OpenAI and Meta warrants for up to 320 million shares at a penny each—warrants being instruments that convey the right to buy stock at a set price—giving those customers a stake in its success.

The newsletter notes that these playbooks are not new: manufacturers such as General Motors and General Electric have long helped customers buy their products through financing and by investing in infrastructure. What may be different this time is the scale of the commitments and the uncertainty surrounding the core product. As Nvidia, Broadcom and AMD compete through financing and equity incentives, questions about competition also arise—are customers paying the right price, and can they shop around?

Before building their businesses around an AI provider, leaders should understand the ecosystem behind it, the newsletter argues: what capital is firmly committed, who bears the losses, and whether investment or purchasing agreements constrain the provider's choices—and ultimately their own.

Much of this financing rests on the AI labs' revenue growth, which is why the underlying numbers matter. As OpenAI and Anthropic move toward initial public offerings, a clearer picture should emerge. The Financial Times reported that OpenAI shared updated financials showing annualized revenue approaching $50 billion as of the end of September—about $20 billion less than the figure circulated to investors the previous month. Annualized revenue simply extrapolates recent sales into a full-year figure, so how each lab books its revenue can materially shape the headline number.

The two labs also calculate revenue differently, with Anthropic including revenue booked by its cloud partners. Strip that out, and Anthropic's $60 billion annualized revenue through the end of July is probably about a third lower. Those are large swings on very large numbers—all the more reason, the newsletter argues, to understand what is being committed, on what terms, and what happens if things go wrong.

It draws a historical parallel: there was a time when GE Capital was reported as a single line item despite making almost half of General Electric's earnings. The financial crisis and demands for transparency forced greater disclosure, highlighting vulnerabilities investors did not like. These are different times, the piece concludes, and going public may create higher demands for clarity on what is really fueling the AI boom.

This story was originally featured on Fortune.com.