Ferveret won $150,000 at LEAP for rethinking how data centres stay cool
Category: Cloud, Infra & Data Centers
Published: 2026-09-07T07:48:00.000Z
Immersion-cooling startup Ferveret won LEAP 2026's $150,000 Scaling Star award, arriving just as Saudi Arabia's AI data-centre build-out makes cooling a strategic, not just technical, problem.
A cooling startup wins big at LEAP Ferveret, a startup working on immersion cooling technology for data centres, won LEAP 2026's Scaling Star award and its $150,000 prize, according to TechScoop's roundup of every 2026 Rocket Fuel winner . The Scaling Star award is given to the strongest growth company in the field, according to that roundup — a fitting label for a cooling technology that becomes more valuable the bigger a data centre's compute footprint gets. What immersion cooling solves Immersion cooling works by submerging server hardware directly in a thermally conductive but electrically non-conductive fluid, rather than relying purely on air or chilled water circulating around equipment. The approach can remove heat far more efficiently than traditional air-cooling, which matters enormously as the chips inside modern servers — particularly the GPUs used for AI training and inference — run hotter and draw more power than the processors data centres were originally designed around. As rack density rises, air alone increasingly struggles to keep pace, which is exactly the gap immersion cooling is built to close. Why cooling has become a strategic issue, not just a technical one Cooling used to be a background engineering detail in data-centre design. That has changed as AI infrastructure build-outs have accelerated globally, because the amount of power a facility can draw and the amount of heat it can safely remove now directly cap how much compute it can actually run. TechScoop has covered how that dynamic is playing out concretely in Saudi Arabia, where a wave of announced cloud and AI infrastructure projects is making power and cooling central questions for the country's AI infrastructure, in a companion piece on how Saudi Arabia's AI boom is becoming an electricity and cooling story . Seen against that backdrop, Ferveret's win looks less like a niche hardware-adjacent curiosity and more like a startup arriving at the right moment with the right technology category. Every additional megawatt of AI compute capacity that gets announced — and Saudi Arabia has announced several such projects — brings with it a proportional cooling requirement, and immersion cooling is one of the technologies operators are increasingly turning to as air-cooling limits are reached at higher rack densities. Where immersion cooling fits in a crowded cooling market Immersion cooling is not the only answer to the heat problem posed by dense AI compute — direct-to-chip liquid cooling, advanced air handling and rear-door heat exchangers are all part of the same broader shift away from traditional air-cooled server rooms. What sets immersion cooling apart is how far it can push density: because the entire server is bathed in coolant rather than relying on air moving across a heat sink, operators can pack more compute into a given footprint without hitting thermal limits as quickly. That is precisely the trade-off large-scale AI data-centre operators are increasingly willing to make as they try to fit more GPUs into the power and space they have available. It also explains why a startup working on this specific technology found itself winning an award at LEAP rather than at a narrower data-centre trade show. Saudi Arabia's AI infrastructure ambitions depend on being able to stand up large amounts of compute quickly, and every operator planning a new facility is weighing the same cooling questions Ferveret is built to answer. Cooling technology of this kind is one piece of that puzzle as the country's announced data-centre projects move toward construction. What the Scaling Star award signals For Ferveret specifically, winning the award for the strongest growth company, rather than an early-stage one, places the company further along than a concept-stage startup — the LEAP judging framework reserved its earlier-stage recognition, such as the Shooting Star award won by Imagine Devices , for startups earlier in their journey. That distinction matters for a hardware-adjacent infrastructure company, where the gap between a promising prototype and a bankable, scalable product is often the hardest part of the business to prove, and it is why Ferveret's $150,000 prize reads as a scale-stage vote of confidence rather than a seed-stage one. The commercial test ahead The company's real test now shifts from the LEAP stage to the data-centre floor. Immersion cooling adoption tends to move in step with new-build construction cycles, since retrofitting an existing air-cooled facility is a harder, costlier sell than specifying liquid cooling into a project from day one. With Saudi Arabia's AI data-centre pipeline still largely in the planning and early-construction phase rather than fully built out, Ferveret's win arrives at a moment when there is still time for its technology to be designed into new facilities rather than bolted on afterward — arguably the best possible timing a cooling startup could ask for. Sources TechScoop (existing coverage