OpenAI is still in business despite not doing good business

For a company on the verge of running out of cash, OpenAI is having a pretty good quarter.

According to the data gathered by Ramp, 46.6% of American companies were paying for AI products and services by December 2025, this marked the fastest month-over-month jump. A huge chunk of that surge flowed directly to OpenAI, whose enterprise adoption has realized a record 36.8%, up 2 percentage points in a single month.

This momentum has enabled OpenAI to maintain a comfortable lead as far as paid AI adoption is concerned. By March 2026, 35.2% of US businesses were paying for OpenAI’s models, followed by Anthropic, which surged to 30.6%, and Google’s AI tools sitting below 5%. Still, dominance hasn’t been translated into durability.

The numbers being brought in are remarkable, but perhaps AI is spending more than it earns. In 2025, OpenAI reportedly burned through more than $8 billion, even as it became one of the fastest-growing enterprise software platforms ever. Financial analyst Sebastian Mallaby had cautionary words for the company predicting that “over the next 18 months,” it could run out of money. The argument set forth is that rivals like Google and Meta can subsidize AI losses using profits from their legacy businesses which is not really up the alley for OpenAI.

That asymmetry matters because OpenAI’s ambitions are enormous, and they sure know how to splurge, and that has translated into a commitment to spend well over $1 trillion before the decade comes to end. AI remains one of Silicon Valley’s most commercially adopted software companies and simultaneously a bottomless pit of cash dumps, yet businesses keep swiping their corporate card.

Tellingly, AI adoption has shifted from pilot phase to core operational budgets corroborated by OpenAI pencilling in 1 million business customers, and 4 million Codex active users product attained by April 2026. In other words, the AI boom isn’t contingent on whether companies use AI; that part seems settled. The bigger question is whether the companies selling it can survive long enough to profit from it.