OpenAI seeks a $1.2T-plus valuation while projecting $278B cash burn
Its five-year plan pairs $840B in revenue with $856B in compute spending, leaving investors to finance the gap.
By Ryan Merket · Published
Primary source: Financial Times
Why it matters
OpenAI's plan makes private capital a core production input. Its burn forecast also transfers risk to the chip, cloud and data-center companies counting on its contracts.
OpenAI expects to generate negative free cash flow of $278 billion from 2026 through 2030 as its infrastructure bill outpaces even its aggressive revenue targets, the Financial Times reported on September 18th, citing a recent presentation prepared for investors.
Sam Altman, OpenAI's co-founder and CEO, is pairing that forecast with early talks for another private funding round. Investors have discussed a valuation of at least $1.2 trillion, according to the FT, and OpenAI is seeking a higher figure. A $1.2 trillion valuation would represent a 41% increase from the $852 billion post-money valuation OpenAI secured less than six months ago.
The presentation lays out the cost of Altman's central bet: OpenAI can spend its way to enough computing capacity, model capability and distribution to outrun declining AI prices. OpenAI forecasts about $856 billion in spending on computing power and infrastructure through 2030. That figure is $16 billion higher than the $840 billion in total revenue OpenAI expects to record over the same period.
Revenue growth does not close the gap
OpenAI projects annual revenue will rise from $36 billion in 2026 to $350 billion in 2030, nearly a tenfold increase. The forecast still produces cumulative negative free cash flow equal to roughly 33 cents for every dollar of revenue over the five-year period.
The projections have improved since May, when OpenAI expected $305 billion in cumulative negative free cash flow. Model releases helped lift annualized revenue by about 20% in July, according to the FT. The revised plan trims expected burn by $27 billion, while leaving OpenAI dependent on capital markets for years.
OpenAI's March financing provides the immediate runway. In a company announcement on March 31st, OpenAI said it had closed $122 billion in committed capital at an $852 billion post-money valuation. Amazon, Nvidia and SoftBank anchored the round, with Microsoft continuing to participate. SoftBank co-led alongside Andreessen Horowitz, D. E. Shaw Ventures, MGX, TPG and accounts advised by T. Rowe Price.
The new investor presentation suggests OpenAI will exhaust that capital in 2028, according to the FT. That makes another private round part of the operating plan rather than optional balance-sheet insurance.
Altman said on September 12th that OpenAI would not pursue an IPO in 2026, arguing that remaining private would give OpenAI room to focus on AI safety. OpenAI had confidentially filed IPO paperwork in June. Deferring the listing leaves private investors as the near-term bridge between the March round and the hundreds of billions of dollars OpenAI expects to spend.
Compute has become the financing strategy
OpenAI has consistently presented access to computing capacity as its main strategic advantage. In April, OpenAI said its Stargate infrastructure program had already exceeded an original target of securing 10 gigawatts of US AI capacity by 2029. OpenAI's infrastructure plan relies on a network of chipmakers, cloud providers, data-center developers, utilities and financial partners rather than a single supplier.
That network spreads the construction and financing burden, while binding some of the technology industry's largest companies to OpenAI's growth assumptions. Nvidia supplies the GPUs at the center of OpenAI's training and inference systems. Oracle provides cloud infrastructure and data-center capacity. SoftBank is both an investor and an infrastructure partner. The FT reported that future revenue expected by Nvidia, Oracle and SoftBank's data-center operation depends heavily on OpenAI contracts.
OpenAI's revenue forecast also faces pricing pressure from Anthropic and lower-cost open-weight models. OpenAI cut the price of its GPT-5.6 Luna model by about 80% in July, an example of how quickly model providers are trading margin for usage and market share.
The investor pitch assumes scale will eventually reverse that pressure: larger infrastructure supports stronger models, stronger models attract more users, and greater utilization lowers unit costs. OpenAI described that cycle as its core economic engine when it raised the March round.
The latest projections put a price on making that cycle work. OpenAI is asking investors to value the business above $1.2 trillion while accepting that $122 billion of recently committed capital may last only until 2028. Revenue growth is already built into the forecast. The financing requirement remains.