Crusoe Triples to $30 Billion on the AI Data-Centre Boom

Crusoe, the company that builds and runs the data centres behind much of the current AI boom, has raised more than $3 billion in a Series F round at a valuation of roughly $30 billion. The figure, reported on September 3, is about triple what the company was worth less than a year ago, and it captures how hard capital is chasing the physical layer of artificial intelligence.

The round was co-led by Atreides Management and Valor Equity Partners, with Mubadala Capital among the investors. It lifts a company that most consumers have never heard of into the same valuation neighbourhood as the model labs its infrastructure serves. Crusoe supplies cloud computing and data-centre capacity to customers including OpenAI, Microsoft, and Meta.

From flared gas to frontier compute

Crusoe's origin story is an unusual one. It started in 2018 capturing flared natural gas at oil fields and using it to power cryptocurrency mining, a way of turning wasted energy into computation. When demand for AI training capacity exploded, the company pivoted its expertise in cheap power and dense computing toward GPUs. That bet has paid off at a pace few investors expected. The Series E, worth $1.375 billion, closed only in October 2025 at a valuation above $10 billion.

The funding is not the only recent sign of momentum. Bloomberg reported that Crusoe recently signed a $13 billion, five-year deal to supply the trading firm Jane Street with GPUs and other AI infrastructure through its cloud. A contract of that size, from a customer outside the usual roster of AI labs, hints at how far demand for raw compute now reaches beyond the companies building the models.

Capital keeps flowing to the plumbing

Crusoe's leap fits a pattern that has defined AI investment this year. The largest cheques are landing not on the applications people use but on the power, land, and silicon underneath them, from Anthropic's multibillion-dollar data-centre commitments in Texas to venture funds raised specifically to build AI's hardware base. As TechCrunch noted, that appetite shows little sign of cooling. The open question is the one that shadows every infrastructure boom: whether the revenue from AI products eventually justifies the mountain of capital now being poured into the ground to support them.