Now, according to Bloomberg, OpenAI is close to finalizing the first phase of a $100 billion-plus round at a valuation that could surpass $850 billion. This first phase will come from Microsoft, Amazon (up to $50 billion), SoftBank (up to $30 billion), and Nvidia, which alone could invest as much as $30 billion, separate from the $100 billion infrastructure commitment the pair made in September.

These unprecedented figures suggest that OpenAI’s investors truly believe it is on a path to achieve technological breakthroughs that alter the trajectory of capitalism itself, ushering an era in which A.I. replaces a significant percentage of global economic output. But it also represents a terrifyingly risky bet on a single company that has yet to prove it will ever achieve profitability.

---

Turn & Burn

OpenAI is a private company, but what we know about its P&L is indeed extraordinary. In 2024, The Information reported that it expected to burn a total of $44 billion between 2023 and 2028 before turning a $14 billion profit in 2029. Less than a year later, the company adjusted its cash-burn forecasts, according to The Information, and was telling investors that it expected to burn$115 billion through 2029 before becoming profitable in 2030. And these projections continue to move in the wrong direction. Just last week, The Information reported that OpenAI now expects to burn nearly $230 billion through 2029, before turning (hugely) cashflow positive in 2030.

Perhaps the most striking thing about the latest report is that these expected losses come despite the company increasing its revenue forecasts for the next five years by 27 percent. At the same time, OpenAI’s gross margins fell to 33 percent in 2025, from 40 percent in 2024—well off its earlier target of 46 percent. Overall, the company made $13 billion in revenue last year, but had negative cashflow of $8 billion. (OpenAI did not respond to a request for comment.)

For this cycle of capital infusions to become sustainable, OpenAI would require significant margin improvement, a drastic jump in model efficiency, a major reduction in inference costs, and stabilized revenue, according to one well-wired V.C. source. One path may be for the company to create more enterprise products, catering to S.M.B.s and mega-caps alike, to help businesses increase productivity and reduce labor costs. But Altman, at least so far, seems intent on OpenAI being primarily consumer-facing. (ChatGPT began testing ads alongside its subscription tiers earlier this month.)

There’s a clear economic logic there: Google and Meta made about half a trillion dollars from advertising, combined, in 2025. But advertising is a tough tightrope to walk, as evidenced by Perplexity’s recent retreat. And it remains to be seen whether generative A.I., at the enterprise level, will replace existing digital infrastructure or become an addition to it. If it’s the latter, the enterprise opportunity becomes significantly more limited, making it increasingly difficult to justify expenditures at this scale. (This is the great power of Google and even Meta: Their advertising cashflows can subsidize loss-making A.I. investments to their heart’s content.) If those structural advantages don’t materialize, Silicon Valley investors warn, it would likely test the sustainability of this relentless fundraising cycle.

---

Will A.I. Kill V.C.?

Of course, OpenAI isn’t the only frontier A.I. company burning through money to compete with Google. In 2024, Anthropic raised a total of around $6 billion. In 2025, it raised $20 billion. Earlier this month, the company completed a $30 billion round, bringing Anthropic’s lifetime funding to $63 billion… and C.E.O. Dario Amodei, himself an early OpenAI employee, says they’re just getting started. The effects on the larger Silicon Valley ecosystem have been profound.

For one, these higher table stakes are potentially crowding out smaller venture investors. “The underwriting now looks less like traditional venture SaaS and more like long-dated industrial financing, particularly if you have a need to enter into multiyear compute commitments or lock in G.P.U. supply,” said Natalie Hwang, founding managing partner of Apeira Capital. “Just in terms of the increasing size of these rounds, it reflects the cost of staying at the frontier. To a certain degree, once you enter that race, undercapitalization can materially impair competitiveness.”

These dynamics tend to benefit larger firms, which are better positioned to raise oversubscribed funds and deploy massive amounts of capital. Indeed, one recent report by Silicon Valley Bank found that one-third of all U.S. venture investment came from the six largest funds, primarily driven by A.I. deals. They’re also better positioned to withstand the time horizon that may be necessary to provide L.P.s with their target returns. “Historically, when markets believe they are underwriting a new paradigm, they can tolerate unusual burn profiles in the early innings,” Hwang said. “The longer-term question is whether the economic rent ultimately justifies the cumulative capital stack—and answering that requires taking at least a five-to-10-year lens.”

The expected payoff is that A.I. becomes a foundational layer of the entire global economy, embedded in payments, services, logistics, and everything else. Investors are “betting on OpenAI becoming infrastructure that powers an ecosystem, similar to how cloud providers are valued on the total economic activity they enable rather than just their direct revenue,” said Vanessa Larco, a longtime V.C. “Investors at this stage have different risk profiles and are taking calculated positions on transformative technology with full understanding of the capital intensity required.”

Larco herself boasts on X that she’s an investor in “companies that OpenAI won’t kill.” But there are real fears throughout Silicon Valley that the hyperscaler era might extinguish the original venture model of scarce capital, price discipline, and close partnerships with founders. One of my sources forwarded me an essay by venture capitalist Benjamin Black declaring that V.C. “is dead” as a result of winner-take-all power laws. “A group of mega-funds now command an unprecedented share of deployable capital,” Black wrote. And the same mega-funds keep getting more funding: Thrive recently secured a $10 billion fund; Lightspeed closed $9 billion; Insight Partners closed $12.5 billion; and a16z just raised over $15 billion.

Even then, the mega V.C.s alone don’t have enough capital to service the needs of OpenAI, Anthropic, or the dozens of blitzscaling companies that haven’t even been created yet. “So maybe they’re participating, but they’re doing a much smaller piece of it,” said Gené Teare, the senior data editor at Crunchbase News. Instead, Altman and Amodei keep turning to the likes of Microsoft, Nvidia, Google, and Amazon—not to mention sovereign wealth funds and private equity. Eventually, Teare predicted, the frontier labs will turn to the public markets for the ultimate liquidity event. She expects that we’ll soon see “one of the biggest I.P.O.s of all time for venture-backed companies”—probably this year.