How much money does OpenAI need to raise by 2028?
About $60 billion to $105 billion of new outside money by the end of 2028, even if everything goes to OpenAI's own plan. If growth comes in at half the plan, the need reaches about $320 billion by 2030. If revenue stalls near today's level, it reaches about $470 billion.
This is our estimate. OpenAI has not published this figure, and it has no public financial statements. Data as of 2026-11-05.
New money needed, by scenario
| Scenario | By end of 2028 | By end of 2030 |
|---|---|---|
| OpenAI's own plan | $60B | $111B (peak of $131B in 2029) |
| Plan, with spending earlier and Nvidia's $30B not fully paid | $106B | $141B |
| Growth at half the plan | $59B | $320B |
| Revenue stalls at about $50B a year | $84B | $467B |
Our model, in US dollars. New money needed means the shortfall after spending about $167B of available funds. It does not count the round now being raised.
How we get to the number
OpenAI's July 2026 projections show $840 billion of revenue, $856 billion of compute and infrastructure spending, and about $278 billion of negative free cash flow from 2026 to 2030. The Star / Reuters, 2026-09-19 The Next Web, 2026-09-19
We set that against about $167 billion of available funds: about $40 billion of cash at the end of 2025, the $122 billion round closed in March 2026, and a $4.7 billion credit line. SiliconANGLE, 2026-03-31 The FT reported that OpenAI is on track to use up the March round by 2028. The Star / Reuters, 2026-09-19 On the plan, the money runs out in 2028.
The range reflects two unknowns. The company has not published the timing of its spending year by year, and it has not said how much of Nvidia's $30 billion has been paid. OpenAI, 2026-02-27 Spending earlier and Nvidia paying later moves the 2028 figure from about $60 billion to about $106 billion.
Why the need grows so fast if revenue disappoints
About $688 billion of OpenAI's planned compute is already under contract: about $300 billion with Oracle, $250 billion with Microsoft and $138 billion with Amazon. DatacenterDynamics, 2025-09 The Register, 2026-01-29 GeekWire, 2026-02-27 We treat that share of the plan, about 80 percent, as spending OpenAI cannot cut. We also assume half of its other costs are fixed. Those are our assumptions, and they drive the downside cases.
Where things stand now
On October 30, 2026, the FT reported that OpenAI cut its current round from about $30 billion to about $20 billion and its target valuation from $1.4 trillion to $1.25 trillion. Bloomberg Law, 2026-10-30 Reuters via KFGO, 2026-10-30 If it closes at that size, the round would cover about a third of the on-plan shortfall.
Reports of its current revenue conflict. The FT put annualized revenue at about $50 billion in early October, while Reuters reported it nearing $70 billion in late September. Part of the gap comes from how sales through cloud partners are counted. Bloomberg News, 2026-10-08 DealStreetAsia / Reuters, 2026-09-30
Our read
This section is our judgment, kept separate from the sourced record above.
Needing the money is not the same as failing to get it. OpenAI's rounds through March 2026 kept getting bigger, which is why the October cut matters. Investors reducing their commitments is the first sign of that pattern weakening, though one round is not a trend.
The 2028 number is the less alarming one. The figure to watch is how quickly the need grows if revenue misses, because most of the costs are already signed.
What would change this answer
- The final size of the current round
- Whether OpenAI reaches its $70 billion year-end revenue target
- Any public financial statements from an IPO filing