ZutaCore Raises 100M to Cut AI Data Center Cooling Costs

The most valuable real estate in an AI data center is no longer the floor space or even the chips. It is the few millimeters directly above each processor where heat has to go somewhere. On June 2, 2026, ZutaCore raised a $100 million Series C to own that sliver of space, and the names writing the checks reveal where the industrial world thinks the next bottleneck in AI lives. This was not a venture round. It was a strategic land grab.

What Actually Happened

ZutaCore, an Israeli cooling startup, closed a $100 million Series C round on June 2, 2026, at an estimated $600 million valuation. The financing was led by a trio of global industrial heavyweights: Mitsubishi Electric, Carrier, and Samsung Electronics through its investment arm Samsung Ventures. That composition matters. These are not financial investors chasing a markup; they are the companies that build air conditioners, heat pumps, and electronics at planetary scale, and they just decided that direct-to-chip liquid cooling for AI is strategic enough to fund directly.

The round brings ZutaCore's total funding since founding to roughly $200 million, meaning this single Series C doubled the capital the company had raised across its entire prior history. The money is earmarked for global commercialization, scaling manufacturing to meet what the company describes as rapidly growing bookings and deployments, and advancing research to handle the thermal demands of next-generation chip architectures. In plain terms, ZutaCore has more demand than it can currently serve, and the cash is meant to close that gap before a competitor does.

The technology at the center of the round is called HyperCool, a waterless two-phase direct-to-chip system. Instead of pushing water through cold plates, HyperCool circulates a proprietary dielectric fluid that is non-conductive and non-flammable directly across the processor. As the fluid absorbs heat, it boils and turns to vapor, carrying thermal energy away through the phase change rather than through simple temperature rise. ZutaCore claims the approach eliminates water risk entirely and cuts cooling energy use by up to half, two pitches that land hard in a year when data center water and power consumption have become political liabilities.

Why This Matters More Than People Think

Cooling has quietly become the gating factor on AI scale. Modern accelerators from Nvidia, AMD, and the custom silicon teams at the hyperscalers now dissipate so much heat per chip that traditional air cooling physically cannot keep up. A single rack of next-generation GPUs can draw more than 100 kilowatts, a density at which air handling fails and liquid becomes mandatory rather than optional. ZutaCore is selling into a market where the question is no longer whether to adopt liquid cooling, but which liquid cooling architecture becomes the standard for the next decade of buildout.

The water angle is where this gets politically charged. Data centers have drawn fierce local opposition over their consumption of municipal water for evaporative cooling, with communities from Arizona to Spain pushing back on projects that drain aquifers during droughts. A waterless system removes that flashpoint entirely. For a hyperscaler trying to win planning permission for a multi-gigawatt campus, a cooling architecture that uses zero water is not just an efficiency story, it is a permitting strategy. That reframes ZutaCore from a components vendor into a partner that can unblock projects regulators would otherwise stall.

The energy math compounds the appeal. Cooling typically consumes a large share of a data center's total power draw, and any system that cuts that overhead by up to half directly improves the facility's power usage effectiveness, the industry's core efficiency metric. In an environment where the binding constraint on AI growth is increasingly the availability of electricity itself, freeing up cooling power means more of every scarce megawatt goes to actual computation. ZutaCore is effectively selling additional compute capacity disguised as a cooling product, and the buyers understand the trade.

Step back and the round fits a pattern that has defined every prior computing era: the constraint moves, and capital chases the new bottleneck. In the cloud era, the constraint was server utilization, and virtualization companies captured the value. In the mobile era, it was battery and radio efficiency. In the AI era of 2026, the constraint has collapsed onto thermals and power, the two physical limits no amount of software cleverness can route around. ZutaCore sits precisely at that pinch point, which is why a cooling company can suddenly command a nine-figure round from investors who normally fund chips and models.

The Competitive Landscape

ZutaCore is not alone in the liquid cooling gold rush. The dominant approach today is single-phase direct-to-chip, championed by players like CoolIT Systems and Boyd, which circulate water or treated coolant through cold plates without a phase change. Immersion cooling, where entire servers are submerged in dielectric fluid, is pushed by companies like Submer and GRC. ZutaCore's two-phase, waterless, direct-to-chip approach stakes out a middle position: more efficient than single-phase water plates, but less disruptive to existing server design than full immersion.

The strategic investors tilt the board. Carrier brings a global HVAC and thermal management footprint, Mitsubishi Electric brings power electronics and industrial reach, and Samsung connects to both chipmaking and its own data center ambitions. Compare this to the cleaner-energy build-outs of the past, where component startups lived or died on whether a major industrial partner adopted their standard. ZutaCore has effectively pre-recruited three of the partners most capable of pushing its architecture into mass deployment, a structural advantage that pure-play rivals funded only by venture capital will struggle to match.

The historical parallel is the format war that decides every infrastructure transition. Just as the industry once converged on standard rack units and specific power distribution schemes, the AI buildout is now choosing its cooling standard, and the winner captures a recurring position in every future facility. The danger for ZutaCore is the same one that sank many technically superior formats: the market can standardize on a good-enough incumbent for reasons of supply chain familiarity and switching cost, leaving the better mousetrap stranded. Two-phase cooling has to prove not just that it works, but that it deploys at scale without operational headaches.

Geography sharpens the picture too. ZutaCore is another entry in the cluster of Israeli infrastructure companies, from Vast Data to DriveNets, that have positioned themselves at the physical layer of the AI economy rather than the application layer. That layer is where demand is least hostage to which model or which lab wins, because every data center, regardless of whose chips or whose software it runs, has to move heat. By selling the most fundamental and unavoidable function in the building, ZutaCore has chosen a market with the widest possible set of buyers and the least exposure to the volatility of the model wars happening one layer up.

Hidden Insight: The Buyers Are the Real Story

The non-obvious signal in this round is not the dollar amount. It is the identity of the lead investors and what their participation telegraphs about the industrialization of AI infrastructure. When Mitsubishi Electric, Carrier, and Samsung co-lead a cooling round, they are not making a bet on a startup. They are placing a marker on which thermal architecture they intend to manufacture and distribute through their own channels. Strategic money of this kind comes with an implicit roadmap: these companies want HyperCool inside the products they already sell to data center operators worldwide.

This reflects a deeper shift in how the AI supply chain is consolidating. For the first two years of the boom, the attention and capital concentrated on chips and models, the glamorous layers. What is happening in 2026 is the quiet industrialization of everything underneath: cooling, power distribution, networking, and the physical plant. The smart money has noticed that while any single model can be commoditized in a quarter, the company that supplies the cooling for ten thousand racks earns a durable, recurring position regardless of which lab wins the model race. ZutaCore sells to all sides of a war whose outcome it does not need to predict.

The bear case, however, deserves equal billing. Two-phase cooling has been the perpetual bridesmaid of thermal management for years, repeatedly described as the superior physics that never quite wins the deployment war. Skeptics point out that boiling a dielectric fluid at scale introduces operational complexity, fluid management, and serviceability challenges that single-phase water systems avoid. The risk is that hyperscalers, who prize operational simplicity above marginal efficiency, default to the boring, well-understood cold-plate approach and leave ZutaCore selling a better technology that the largest buyers consider too clever for production.

There is also a concentration risk hiding inside the strength of the investor syndicate. When three industrial giants anchor a round, the startup's destiny becomes tied to their strategic priorities, which can shift with a single reorganization or a change in their own data center plans. If Samsung, Carrier, or Mitsubishi reprioritize, ZutaCore could find its distribution promise evaporating faster than a financial investor would ever allow. Strategic capital cuts both ways: it opens doors that venture money cannot, but it also makes the company a passenger in someone else's larger plan.

There is a final wrinkle that makes the timing of this round shrewd. The dielectric fluids at the heart of two-phase cooling have themselves come under regulatory scrutiny, with some earlier coolants facing restrictions over environmental persistence. ZutaCore emphasizing a non-conductive, non-flammable proprietary fluid is partly a technical claim and partly a regulatory hedge, an attempt to get ahead of the same kind of policy pressure now reshaping refrigerants across the HVAC industry. With Carrier and Mitsubishi, two companies that have navigated exactly those refrigerant transitions, sitting on the cap table, ZutaCore is buying not just distribution but regulatory know-how for the fluid chemistry that underpins its entire approach.

What to Watch Next

In the next 30 days, watch for named deployment announcements. A $600 million valuation on a cooling company only holds if the bookings ZutaCore alludes to materialize as disclosed contracts with named hyperscalers or colocation providers. The single most important leading indicator is whether one of the major cloud operators publicly commits to HyperCool for a flagship campus, which would validate two-phase cooling as a production standard rather than a pilot curiosity.

On a 90-day horizon, track whether the strategic investors begin embedding ZutaCore technology into their own product lines. A Carrier or Mitsubishi data center cooling product that ships with HyperCool inside would convert the investment thesis into revenue and signal that the format war is tilting ZutaCore's way. Watch also for competitive responses from CoolIT, Boyd, and the immersion players, who will not cede the high-density segment without aggressive counter-pricing and their own efficiency claims.

Looking out 180 days, the deciding question is whether next-generation chips force the issue. As accelerators push past current thermal limits, the rack densities that make two-phase cooling compelling become the norm rather than the exception. If the roadmaps from Nvidia and the custom silicon teams demand cooling that single-phase water cannot deliver, ZutaCore's moment arrives by physics rather than persuasion. The company that has spent years waiting for the heat to catch up to its technology may finally be selling into a market with no alternative.

It helps to put the energy claim in concrete terms. A large AI campus can draw a gigawatt or more of power, and cooling has historically consumed as much as 30 to 40 percent of that draw before any of it reaches the chips. If a waterless two-phase system can cut that cooling overhead by up to half, the recovered power translates into thousands of additional accelerators running on the same grid connection. In a year when utilities across the United States are warning that they cannot bring new capacity online fast enough for AI demand, that recovered headroom is worth more than the cooling hardware itself. ZutaCore is, in effect, selling grid capacity that already exists but is currently being wasted on moving heat.

That framing, cooling as recovered compute, is ultimately why three industrial giants moved at a $600 million valuation rather than waiting for a cheaper entry. They are not buying a thermal product. They are buying a position in the single most binding constraint on AI growth in 2026, and they would rather own it early than rent it late.

The AI race will be won in the few millimeters above each chip, and the companies funding ZutaCore just decided that waterless boiling is where the heat goes next.

---