OpenAI Targets $600 Billion in Compute Spend by 2030 as It Eyes a $1 Trillion IPO

OpenAI has told investors it expects to spend roughly $600 billion on compute through 2030 as the company quietly lays the groundwork for what could become the largest technology IPO in history --- a public offering that would value the ChatGPT maker at up to $1 trillion. The numbers, first reported by CNBC and confirmed by multiple outlets, paint a picture of a company betting everything on an AI future that will require infrastructure spending at a scale the tech industry has never seen.

The Numbers Behind the Ambition

OpenAI's 2025 financial performance was, on the surface, impressive. The company generated $13 billion in revenue, comfortably beating its own $10 billion internal forecast. Spending came in at $8 billion, under the $9 billion target.

But the details tell a more complicated story. Inference costs --- the expense of actually running AI models --- quadrupled in 2025 as demand for ChatGPT and API services outstripped capacity. OpenAI was forced to buy expensive computing capacity on short notice at premium rates. Adjusted gross margins fell to 33% from 40% year-over-year --- roughly half what a healthy software company typically runs.

The company projects more than $280 billion in cumulative revenue by 2030, split nearly evenly between consumer products and enterprise services. But it expects margins between 52% and 67%, missing its previous goal of 70% by 2029. Training costs alone are projected to hit $440 billion through 2030, with $32 billion earmarked for model training in 2026 and around $65 billion in 2027.

The Funding Machine

To feed this appetite, OpenAI is assembling the largest private funding round in corporate history. NVIDIA is nearing a deal to invest $30 billion as part of a broader fundraising effort that could exceed $100 billion, which would value the company near $830 billion --- already approaching the trillion-dollar IPO target.

The infrastructure commitments alone are staggering. OpenAI has approximately $1 trillion in deals lined up this year, including:

• A $300 billion partnership with Oracle for data center capacity
• The $500 billion Stargate initiative with Oracle and SoftBank, targeting 10 gigawatts of computing power
• Federal loan guarantees being sought for additional infrastructure financing

The company is also seeking government backing to help finance the build-out --- a remarkable ask for a private company that hasn't yet demonstrated sustainable profitability.

Stargate: Progress and Problems

The Stargate project, announced with fanfare alongside then-President Trump, is OpenAI's cornerstone infrastructure bet. The joint venture with Oracle and SoftBank aims to build a network of AI data centers across the United States, with sites in Abilene, Texas; Shackelford County, Texas; Doña Ana County, New Mexico; Lordstown, Ohio, and additional locations in the Midwest.

The combined planned capacity from these sites brings Stargate to nearly 7 gigawatts and over $400 billion in investment over the next three years. OpenAI says the project is ahead of schedule.

But not everyone agrees. The Information reported that the consortium has been plagued by disagreements between OpenAI, Oracle, and SoftBank over responsibilities, organizational structure, and who ultimately controls the planned data centers. More than a year after the announcement, some aspects of the project have yet to materialize as promised.

The Trillion-Dollar Question

The IPO timing remains unclear, but the pieces are falling into place. At an $830 billion pre-IPO valuation and a trajectory toward $280 billion in 2030 revenue, OpenAI would need to demonstrate it can eventually turn compute spending into sustainable margins --- something it has not yet done.

The core tension is structural: OpenAI's costs are scaling faster than its revenue. Every new user, every API call, every reasoning chain costs real money in GPU-hours. The company's own projections show margins missing earlier targets, and the gap between spending and revenue keeps widening before it's supposed to converge.

For context, the Kobeissi Letter flagged 2026 as a historic year for IPO launches --- with SpaceX expected at $1.5 trillion, OpenAI at $1+ trillion, and Anthropic at $500 billion. Whether OpenAI can justify a trillion-dollar valuation while burning through $600 billion in compute will be the defining question of the AI industry's next chapter.

What It Means for AI

If OpenAI pulls this off --- raising $100 billion, spending $600 billion, building 10 gigawatts of compute infrastructure, and going public at a trillion dollars --- it would represent the largest private-to-public capital deployment in technology history. It would also validate the thesis that AI is not just another tech cycle but a generational platform shift worth betting national-scale resources on.

If the margins never materialize, it could become the most expensive bet in corporate history. Either way, OpenAI has made its intentions unmistakable: the company is going all-in, and it's asking the entire financial system to go all-in with it.

Sources:

• CNBC: OpenAI Resets Spend Expectations, Targets ~$600B by 2030
• Tech Startups: OpenAI Expects $600B Compute Spend by 2030 as It Eyes $1T IPO
• U.S. News: OpenAI Expects Compute Spend of ~$600B by 2030
• The Decoder: OpenAI Adds $111B to Cash Burn Forecast
• OpenAI: Five New Stargate Sites
• The Decoder: Stargate $500B Project Stalls Over Disputes
• Business Standard: OpenAI Expects $600B in Compute Spending