OpenAI just bumped its compute and cloud infrastructure spending forecast to roughly $750 billion through 2030. That’s up from the $600 billion target it had floated earlier this year, and it signals that the company’s hunger for raw computing power is accelerating faster than even its own projections could keep up with.
The revised projection, reported by the Wall Street Journal on July 22, represents a 25% jump from the roughly $600 billion figure OpenAI had set earlier in 2026. The money is earmarked for two things: new cloud service agreements with existing providers and, increasingly, building out OpenAI’s own data center footprint.
The crown jewel of that self-built infrastructure push is Project Camellia. OpenAI committed $20 billion to kick off the massive data center complex in Effingham County, Georgia. The facility comes with a 3.2 gigawatt power contract running from 2028 to 2032, which is roughly enough electricity to power a mid-sized city.
Here’s the thing about OpenAI’s spending trajectory: it hasn’t exactly been a straight line. CEO Sam Altman had previously floated comments about $1.4 trillion in AI infrastructure investment, a number so large it spooked investors and prompted a reset. The company dialed expectations back to around $600 billion earlier this year. Now it’s creeping back up, landing at $750 billion.
The company also reportedly hired an architect who previously worked on Elon Musk’s computing infrastructure build-out, adding another signal that OpenAI is dead serious about controlling its own hardware destiny rather than relying entirely on cloud partners like Microsoft Azure.
A 3.2 gigawatt contract for a single facility is enormous. For context, a typical nuclear power plant generates about 1 gigawatt. OpenAI is essentially reserving the output of three nuclear plants’ worth of electricity for one data center complex.
There’s no blockchain component, no token launch, no DeFi integration anywhere in OpenAI’s infrastructure plans. The spending is purely traditional: servers, real estate, power contracts, and cloud agreements.
The massive energy demands of AI data centers directly compete with another energy-hungry industry: Bitcoin mining. When a company like OpenAI locks up 3.2 gigawatts of power capacity in a single deal, that’s electricity that isn’t available for mining operations. We’ve already seen this dynamic play out in Texas, where Bitcoin miners and AI data centers compete for the same grid resources and the same favorable energy contracts. Some mining companies have pivoted to offering their facilities for AI compute workloads because the margins are better.
The spending revision from $600 billion to $750 billion in just a few months also says something about the pace of change. Forecasts in AI are moving targets, and if demand keeps outpacing projections, we could easily see another upward revision before 2030.