Sam Atlman-backed Exowatt Wants To Power Ai Data Centers With Billions Of Hot Rocks

When Hannan Happi started reasoning astir really to lick nan AI powerfulness crisis, he kept 1 fig successful mind: 1 cent per kilowatt-hour.

“We went done each sorts of configurations and designs,” Happi, co-founder and CEO of Exowatt, told TechCrunch. “They each look different from each other. We tried to study from everyone 1 of them: How do I trim structural costs? How do I trim attraction costs? How do I optimize for this?”

After years of brainstorming and building, Exowatt’s first measurement toward that extremity is simply a elemental container nan size of a shipping instrumentality topped pinch a clear awning. Inside is likewise simple. If Exowatt tin present connected its committedness of delivering inexpensive star powerfulness that generates energy 24-7, it could upend nan information halfway marketplace and broader power world, delivering round-the-clock powerfulness astatine very debased cost.

To standard accumulation successful pursuit of nan one-cent per kWh target, Exowatt has raised an further $50 cardinal successful an hold to its $70 cardinal Series A information that closed successful April, TechCrunch has exclusively learned. 

The hold was led by MVP Ventures and 8090 Industries pinch information from Atomic, BAM, Bay Bridge Ventures, DeepWork Capital, Dragon Global, nan Florida Opportunity Fund, Massive VC, New Atlas Capital, Overmatch, Protagonist, and StepStone. Previous investors see Andreessen Horowitz and Sam Altman.

Happi said that Exowatt wasn’t looking to raise further superior aft nan April round, but “strong momentum we saw successful nan market” and “strong investor interest” encouraged him to return nan caller money astatine a higher valuation.

Exowatt’s backlog is presently astir 10 cardinal P3 units representing 90 gigawatt-hours of capacity, he said. “The extremity is to standard arsenic accelerated arsenic imaginable to nan millions and yet billions of units,” he said. The institution should deed its one-cent target erstwhile accumulation hits astir 1 cardinal units per year, Happi said.

Exowatt is fundamentally repackaging a exertion that’s been astir for decades. Known arsenic concentrated star powerfulness aliases thermal star power, it uses nan Sun’s power to power materials that are bully astatine storing aliases transporting thermal energy. In cases wherever that thermal power is stored for an extended play of time, those materials thin to beryllium derived from aliases lucifer rocks — hence nan technology’s nickname of “rocks successful a box.”

Each P3 instrumentality consists of a metallic container topped pinch lenses that attraction nan sunlight into a tight beam. That beam past heats a typical ceramic wrong nan shipping container. A instrumentality blows aerial complete nan ceramic to transportation nan power to different container which contains a Stirling motor (a piston-driven instrumentality that converts power to mechanical energy) and a generator. To shop much power, developers would instal much P3 boxes. “Everything is designed to beryllium highly simple,” Happi said.

Each thermal artillery tin clasp power for up to 5 days, ensuring continuous operation, and Exowatt will drawstring respective together to provender a azygous generator unit. How galore depends connected really quickly and really overmuch energy a customer wants to generate. The system’s ratio is connected par pinch photovoltaic star panels, and somewhat amended than PV paired pinch lithium-ion batteries, Happi said.

Other companies person built various approaches to nan aforesaid technology, though astir person failed to compete pinch photovoltaic star and lithium-ion batteries, some of which person amazed experts pinch really quickly they’ve travel down successful cost.

Happi based on that nan P3’s mini size and Exowatt’s iterative attack sets it apart. There are conscionable complete 100 star thermal aliases concentrated star powerfulness projects that are planned, built, aliases decommissioned astir nan world, he said. “If you comparison that to nan truth that we nutrient 1.5 cardinal star panels per year, you tin spot nan learning curve effects are very, very acold isolated from each other. 

“What Exowatt is astir is taking a modular strategy that we cognize successful rule works, and really scaling nan manufacturing of that and past applying nan learning curves of manufacturing.”

Exowatt is improbable to beryllium costs effective everywhere, and nan number of P3 units needed to powerfulness a information halfway could require monolithic amounts of land. Plus, it useful champion successful nan sunniest regions, which could limit its broader impact. 

But Happi counters that there’s a “high overlap” betwixt wherever Exowatt’s P3 excels and wherever caller information centers are being built. “We’re not moving short of immoderate projects to do,” he said.