OpenAI burned through $34 billion last year

The $34 billion figure is striking, but the more consequential detail is what it implies about the floor: any serious competitor now faces a capital requirement that effectively excludes all but sovereign-backed or mega-cap players from the frontier tier.

This story lands alongside Wired's recent reporting on enterprise token consumption ('Pretty Crazy' Token Usage Is Testing Bosses' Bet on AI), which documented how inference costs are already straining the unit economics of AI deployment at the customer level. Read together, the two pieces describe pressure from both ends of the value chain: OpenAI is burning capital at the infrastructure layer while its enterprise customers are simultaneously discovering that token costs threaten their own ROI models. That double squeeze matters for how the industry prices inference going forward. If OpenAI needs to raise prices to close the gap toward profitability, it collides directly with enterprise buyers who are already renegotiating usage contracts to control spend.

Watch whether OpenAI announces a revised enterprise pricing tier or consumption-based rate change within the next two quarters. If it does, that confirms the burn rate is already feeding back into commercial terms rather than being absorbed by continued fundraising.