The generative AI industry must reach over $2 trillion in annual revenue by 2030 to justify its infrastructure buildout, according to an analysis published by Where's Your Ed At. The piece argues that combined compute commitments from OpenAI and Anthropic now exceed $1.1 trillion, creating debt pressure across hardware and cloud providers that only a dramatic surge in enterprise spending could offset.
The Cost of Planned Data Centers
Citing data from Sightline Climate and NVIDIA CEO Jensen Huang, the analysis estimates roughly 190 gigawatts of data centers are planned globally. At $80 billion to $100 billion per gigawatt, that implies capital expenditure of $9.5 trillion to $15 trillion. NVIDIA is separately projected to bring in $1 trillion in total revenue through the end of 2027, with 54% of that revenue coming from just three clients, per the same analysis.
OpenAI and Anthropic Drive Compute Demand
The analysis puts OpenAI and Anthropic's combined share of AI compute demand at 70-90%. It cites OpenAI's projected cash burn at $852 billion through 2030, including more than $770 billion in compute commitments across Microsoft, Amazon, CoreWeave, Cerebras, and Oracle, with $50 billion in compute spending expected in 2026 alone. It also cites $250 billion in new OpenAI funding needed by year-end.
For Anthropic, the analysis cites $330 billion in chip and compute commitments with Google, Amazon, and Microsoft, plus $30 billion with CoreWeave and $15 billion with SpaceX, and says the company needs $174 billion in annual revenue by 2029 to meet these obligations, along with $200 billion in additional capital next year.
The analysis further claims the two companies account for 89% of all AI startup revenue, making the broader compute ecosystem dependent on two currently unprofitable firms.
Enterprise Demand and Trade-Offs
Combined 2026 revenue projections for OpenAI and Anthropic sit near $60 billion, according to the analysis, which calculates that reaching their infrastructure targets would require roughly 496% growth by the end of 2029. The piece also notes Oracle's data center commitments tied to OpenAI, estimated between $340 billion and $700 billion for 7.1 gigawatts of capacity, as an example of infrastructure partners' exposure if enterprise demand falls short.
Caveats
The analysis assumes infrastructure commitments and data center buildouts proceed at full scale without renegotiation. If hyperscalers or model developers slow spending or adjust obligations, the projected revenue gap could shrink, though likely alongside declines in hardware valuations.
Why It Matters
If enterprise adoption doesn't keep pace with these compute commitments, the financial strain could extend beyond AI labs to cloud providers, chipmakers, and debt markets — raising questions about whether current AI infrastructure spending reflects sustainable economics or a temporarily subsidized market.