Crusoe began in 2018 by mining bitcoin with natural gas that would otherwise have been flared into the air. On September 17, the company that grew out of that trick announced a $3.9 billion funding round that values it at $30.9 billion, and a new product line that is meant to solve one of the hardest problems in the AI build-out: getting electricity to the chips.
The Series F round was led by Atreides Management, Mubadala Capital and Valor Equity Partners, with Nvidia, Founders Fund, GIC, Qatar Investment Authority and TPG among the participants. The valuation is a sharp climb from the $10 billion the company was worth as recently as last October, a rise that tracks the broader repricing of anyone who can deliver compute.
The money is headed toward two places. Some of it will fund projects already underway. The rest will back a modular data center called Spark, which the company builds in its own factories and then trucks to wherever the power is, plugging in and running once it arrives. The pitch is that Spark sidesteps two of the industry’s chronic bottlenecks at once: the shortage of construction labor, and the local opposition that greets giant campuses.
That opposition has become a real constraint on the AI build-out. Communities that once welcomed data centers have begun to push back against the water, the noise and the footprint, and several large projects have been delayed or scaled down. A factory-built unit on a truck is an answer to that problem, the company argues, because it can be placed where the electricity already is instead of where the politics allows.
The shift from flared-gas mining to data centers is the story of the company’s whole decade. Crusoe found a way to make money from energy that would have been wasted, then found that the same instinct, putting compute where the power is stranded, applied to a much larger market. Its customers now include Meta, Microsoft and Oracle, names that did not exist in its original business plan.
The contract book has grown to match the valuation. Last month Crusoe signed a five-year deal worth $13 billion with Jane Street, the trading firm, a commitment that demonstrates how far the company has moved beyond its crypto origins and into the ranks of serious compute suppliers.
The next question is the public market. Crusoe has been in contact with Goldman Sachs and Morgan Stanley about an initial public offering, according to reports, a move that would convert a decade of private growth into a test of what public investors are willing to pay. The $30.9 billion valuation is a number that will either be validated or revised by that market.
The compute market the company is chasing is changing under it. Demand has begun to tilt from training, which needs thousands of chips in one place, toward inference, which can be spread across smaller sites. That shift favors the kind of modular, distributed capacity that Spark represents, and it is one of the reasons the new product exists.
Competition is not standing still. The large cloud providers are building their own capacity, and a wave of funded startups is doing the same. Crusoe’s answer has been speed and specialization: it builds one kind of facility and builds it quickly, and it places it where the power problem is already solved.
The capital raise is a bet on that answer. Every dollar of the $3.9 billion is a wager that the demand for rented compute keeps growing, that the contracts hold their value, and that a company built on stranded energy can keep finding new sources of it. The modular data center is the physical expression of that bet, and the truck is the delivery mechanism.
Crusoe’s history gives the strategy a certain logic. The company has never built where it was easiest to build; it has built where the energy was cheapest and most available, and let the compute follow the power. Spark is the same idea, industrialized. The question now is whether the market that rewarded the instinct at $30.9 billion will keep rewarding it as the numbers get bigger and the scrutiny gets sharper.
The investor list tells its own story. Mubadala and the Qatar Investment Authority are sovereign funds with long horizons, the kind of capital that enters a company when the ambition runs beyond a single cycle. Their presence alongside Nvidia, which has become a quiet backer of the entire AI infrastructure trade, suggests that Crusoe’s round was less about distress than about scale, and less about survival than about expansion.
The path from flared gas to sovereign capital has taken eight years, and it has required the company to reinvent itself at each step. The next step, if the IPO talks turn into a filing, will be the first time the company’s full financial picture is laid out for public investors, who will have their own view of whether a company built on stranded energy deserves a $30.9 billion price.
