Some will understandably take a glance at this rising data center behemoth and wonder if climate tech is really an accurate description of what Crusoe actually does these days. As the steady drumbeat of announcements and press surrounding Crusoe’s partnerships and power deals has built up, I certainly wondered whether the company had pivoted to simply churning out data centers as quickly as possible. But investors — and the company itself — told me that’s far from true.
Clay Dumas, a partner at Lowercarbon Capital, which invested in the company’s $128 million Series B and $350 million Series C rounds, told me that Crusoe remains as mission-focused as ever. “When it comes to power, Crusoe is the most aggressive innovator in the AI infrastructure space,” Dumas said via text message. “There is no better team to integrate new energy sources for compute workloads so we don’t turn the whole world into one giant fracking operation.”
Ben Kortlang, a partner at G2 Venture Partners, which led the company’s Series C round, agreed, telling me that Crusoe is best positioned to build out data centers in a way that doesn’t “plant the seeds for 50 or 100 years of environmental damage.”
Yet it’s hard to pin down exactly what the energy mix will end up looking like for the high-profile data centers in Crusoe’s pipeline, including the complex it’s currently building for OpenAI, which is part of the Stargate project in Abilene, Texas. The company announced on Tuesday that it had started construction on the second phase of the facility, which expands the total scope from around 200 megawatts of power across two facilities to include a total of eight buildings over 4 million square feet, using 1.2 gigawatts of power. Crusoe’s spokesperson, Andrew Schmitt, declined to comment on whether this additional capacity would serve Stargate.
What Schmitt did confirm via email is that while the project has a 1.2 gigawatt grid interconnection — enough to meet the entirety of its power needs — Crusoe will also rely on natural gas as “backup energy,” as well as behind-the-meter energy solutions such as solar and battery storage to “create a highly optimized and efficient power plan for the full site.”
The company also won’t speculate on how much energy will come from each particular source. To some degree, the exact grid energy mix and what additional energy resources will get built is unknowable, though Schmitt told me that Crusoe chose Abilene for the area’s abundant wind resources. There’s often too much of it for the grid to handle, meaning the excess energy is curtailed or sold at a negative price. But if a large load — say, a Crusoe data center — were added to the grid, less renewable energy would go to waste, thereby increasing the profitability of renewables projects and incentivizing more buildout overall.
This strategy, Schmitt told me, “reflects [Crusoe’s] guiding principle of bringing load to stranded and under-utilized energy” rather than bringing energy sources to the data center load itself, as the industry has traditionally done. G2, the venture capital firm, is all in on this premise. “By putting a big load center right there in a fantastic renewable resource environment, the thing that will naturally get built is renewables,” Kortlang told me. “Crusoe doesn’t need to mandate that, or control that, or be the one building the renewables. They’re creating the demand.”
But this approach is only net-positive for the climate if it increases the share of renewables in the mix overall, i.e. if new, large loads are leading to more solar and wind buildout than new natural gas buildout. And while a renewables-heavy buildout seems to be what Crusoe and its investors are assuming will happen, Crusoe can’t actually control what gets put on the grid or the economic or political factors that drive those decisions.
It appears to be inevitable that gas will play some role, even if it’s providing power directly to the data center itself and not to the grid overall. According to Business Insider, public filings with the Texas Commission on Environmental Quality show that so far, Crusoe plans to operate on-site natural gas turbines at the Abilene facility totaling 360 megawatts of power. That represents 30% of the data center’s total 1.2 gigawatts of announced capacity.
Although powering data centers with new solar or wind is usually the cheapest option — especially in places like Abilene — building natural gas can be quicker and more reliable, assuming you’re able to acquire the severely backlogged turbines. That’s something Kortlang readily acknowledged to me. “We will see a lot of buildout of natural gas over the last half of this decade, because it’s the easiest thing to controllably build that gets you large amounts of baseload power quickly,” he said.
Kortlang didn’t seem fazed by Crusoe’s announcement this Monday that it’s pursuing a joint venture with the investment firm Engine No. 1, giving the company access to a whopping 4.5 gigawatts of natural gas power. To put that in perspective, there’s only about 25 gigawatts of existing data center capacity in the U.S. today. Schmitt told me this latest announcement is unrelated to the Stargate Project.
Engine No. 1 has secured seven GE Vernova natural gas turbines through a partnership with Chevron announced in January. As Chevron puts it, this joint development will create “scalable, reliable power solutions for United States-based data centers running on U.S. natural gas.” But critically, as Crusoe emphasized, “plans for these data centers include the use of post-combustion carbon capture systems,” which are designed to capture the CO2 from power plants after the fossil fuels are burned, but before they’re released to the atmosphere.
Presumably, these plans will also incorporate either some way to utilize the CO2 in industry or to permanently sequester it underground, though the company hasn’t mentioned anything to this effect. This technology hasn’t been a part of the company’s strategy in the past, though Kortlang told me that Crusoe has been evaluating the viability of carbon capture and storage for as long as G2 has been involved.
Gas-fired power plants paired with carbon capture have never really caught on, simply because they’re pretty much bound to cost more than not building carbon capture. When I asked Kortlang if this meant Crusoe was banking on its data center customers being willing to pay more for greener power, he told me that was “to be determined.” Who exactly was going to design and build the carbon capture technology — Crusoe, Chevron, or another to-be-named project partner — was also “to be determined.” But there’s not actually all that much time to figure it out. In Chevron’s announcement, the company said it was planning to deliver power by the end of 2027.
So, is Crusoe still a climate tech company? The answer seems to be yes — or at least it’s definitely still trying to be.
No other developer has been as diligent about utilizing stranded assets to power data centers. And with its expansion into carbon capture, it certainly seems Crusoe is leaning into an all-of-the-above approach to data center decarbonization. As Dumas told me, “before too long” we’ll also see Crusoe powering its operations with “geothermal, bioenergy, and after that fusion technologies that keep them out ahead of the pack.”
But Crusoe’s business model — and its clean tech bonafides in general — have always relied upon ultimately unprovable counterfactuals. First it was: If this waste gas weren’t powering Bitcoin mining, it would be vented into the atmosphere. That seemed fairly certain, since flaring is common practice in many areas. Now the company is pitching a somewhat fuzzier hypothetical: If this Crusoe data center, powered by some combination of natural gas and stranded renewables, were instead built by another company, it would inevitably be dirtier. Whether or not Crusoe is a boon for the climate ultimately depends upon the degree to which that unquantifiable claim ends up being true.
Smog covers New York City in 1953.Library of Congress/Walter Albertin
More smog in 1970...Library of Congress/Bernard Gotfryd
...and again in 1973.The National Archives/Environmental Protection Agency/Wilbert Holman Blanche
