Ad
Skip to content

OpenAI adds $111 billion to its cash burn forecast as AI costs spiral beyond projections

Image description
Sora prompted by THE DECODER

Key Points

  • OpenAI has once again raised its spending projections and now expects a cumulative cash burn of $665 billion by 2030 to train and run its AI models, roughly $111 billion more than previously forecast.
  • The company does not plan to become cash-flow positive until 2030, while rival Anthropic is aiming for break-even as early as 2028.
  • Inference costs are set to quadruple by 2025, adjusted gross margins have dropped to 33% instead of the targeted 46%, and training expenses alone are projected to reach nearly $440 billion by the end of the decade.

OpenAI is raising its revenue forecasts while warning investors about a dramatic increase in cash outflow. The cost to train and run AI models is growing faster than revenue.

Citing internal financial documents, The Information reports that OpenAI has revised its projections again. The company now expects cumulative cash burn to jump by roughly $111 billion more than previously estimated through 2030. In total, OpenAI expects to spend $665 billion training and operating AI models. While revenue is climbing, it cannot keep pace.

Under the new projections, OpenAI forecasts a cash burn of $25 billion in 2026 and $57 billion in 2027, roughly $30 billion more than previous estimates combined. The company doesn't expect to become cash-flow positive until 2030, when it projects a positive cash flow of $39 billion. This puts OpenAI behind competitor Anthropic, which is targeting breakeven as early as 2028.

Year Forecast Q1 2025 Forecast Q3 2025 Forecast Q1 2026
2024 approx. -2 billion -2 billion -2 billion
2025 approx. -7 bn -9 billion -8 billion
2026 approx. -8 bn -17 billion -25 bn
2027 approx. -20 bn -35 billion -57 billion
2028 approx. -11 bn -47 billion -85 billion
2029 approx. +12 bn -8 billion -51 bn
2030 approx. +41 bn +38 bn +39 billion

Shift in cash burn forecasts across three projection dates. Figures in US dollars. Source: The Information.

Ad
DEC_D_Incontent-1

OpenAI already revised its spending forecast sharply upward last fall, raising expected cumulative cash burn by roughly $80 billion to $115 billion through 2029. The latest correction shows this trend accelerating.

Consumer Subscriptions Fuel Revenue While Hardware Takes a Back Seat

OpenAI more than tripled its revenue to $13.1 billion in 2025, beating its own forecast by $100 million. The company expects $30 billion in 2026 and around $62 billion in 2027. Overall, revenue projections through 2030 are roughly 27 percent higher than previous estimates.

The consumer business remains the biggest revenue driver and is expected to generate around $150 billion by 2030. OpenAI now boasts 910 million weekly active users—a new high, but reportedly below the one billion mark the company wanted to hit by the end of 2025. GPT-5 is allegedly to blame for the slowdown, though the "warmer" successor models 5.1 and 5.2 reportedly got growth back on track. OpenAI CEO Sam Altman recently wrote in an internal memo that ChatGPT is growing again at 10 percent month over month. The company is targeting 2.75 billion weekly active users by 2030.

In the enterprise segment, OpenAI wants to push ChatGPT Enterprise and B2B revenue to $70 billion by 2030, a massive jump from just $2 billion in 2025. The API segment is expected to hit $47.5 billion, while hardware and other new products are projected at just $15 billion in 2030. OpenAI expects its first revenue from hardware and "new products" in 2026, starting at $100 million and growing to $1.5 billion the following year. The first device could ship in spring 2027; OpenAI has several products in development, including a smart speaker.

Ad
DEC_D_Incontent-2

Soaring inference costs squeeze profit margins

A major driver of this cost explosion is inference. The day-to-day expense of running AI models quadrupled in 2025, according to the report. OpenAI had to buy more expensive computing capacity on short notice when demand for ChatGPT and API services exceeded expectations.

The margin numbers are particularly rough. OpenAI's adjusted gross margin dropped to 33 percent in 2025, down from 40 percent the previous year and well below its 46 percent target. Moving forward, the company expects margins between 52 and 67 percent, missing its previous goal of 70 percent by 2029. For context, successful software companies typically maintain gross margins above 70 percent.

Training costs alone are projected to hit nearly $440 billion through 2030. OpenAI plans to spend $32 billion on model training in 2026 and around $65 billion in 2027. The main beneficiaries of that spending—assuming OpenAI can pay the bills—are Microsoft, Amazon, and Oracle, which all signed long-term contracts providing computing capacity and hardware.

To fund these massive expenses, OpenAI is currently negotiating a funding round of more than $100 billion at a valuation of around $750 billion. SoftBank, Amazon, Nvidia, and Microsoft are reportedly on board as investors. OpenAI ended 2025 with roughly $40 billion in cash and cash equivalents.

AI News Without the Hype – Curated by Humans

As a THE DECODER subscriber, you get ad-free reading, our weekly AI newsletter, the exclusive "AI Radar" Frontier Report 6× per year, access to comments, and our complete archive.

Source: The Information