Big Oil Meets Big Tech: Shell and Airsys Tackle AI's Thermal Crisis

WOODRUFF, S.C. – October 05, 2026 — The physical reality of artificial intelligence is surprisingly industrial. Beneath the seamless, cloud-based interfaces of modern generative AI lies a foundation of heavy metal, silicon, and, increasingly, an overwhelming amount of heat. As the digital economy races to build denser and more powerful compute clusters, the infrastructure that houses these systems is hitting a literal boiling point. In a move that perfectly encapsulates the blurring lines between legacy energy giants and modern technology infrastructure, Airsys and Shell Lubricants have announced a joint partnership to demonstrate a next-generation spray liquid cooling system at the upcoming 2026 OCP Global Summit in San Jose, California.

This collaboration is more than just a technical showcase; it is a bellwether for a massive shift in how the world’s most critical infrastructure is designed, operated, and maintained. The partnership pairs Airsys, a South Carolina-based leader in mission-critical thermal management with over three decades of experience, with Shell, the world’s leading supplier of finished lubricants. Together, they are introducing a patented rack-level spray liquid cooling platform that utilizes specialized dielectric fluid. The system is engineered to simplify deployment for mid-density AI, telecom, and edge environments, offering a critical bridge for operators caught between the limits of traditional air cooling and the daunting logistics of full-scale immersion systems.

Bridging the Thermal Divide: The Retrofit Bottleneck

To understand the significance of this development, one must look at the math driving modern data centers. For years, enterprise facilities operated comfortably with rack power densities hovering around 5 to 10 kilowatts (kW). Today, driven by the voracious power demands of advanced GPUs—which can individually draw upwards of 1,200 watts—AI workloads are pushing rack densities to 60kW, with some hyperscale deployments exceeding 100kW.

At these extremes, traditional air cooling is not just inefficient; it is physically incapable of removing the heat fast enough to prevent catastrophic thermal throttling. The industry standard, as outlined by the ASHRAE TC 9.9 thermal management guidelines, has increasingly pointed toward liquid cooling, specifically the H1 class for high-density liquid-cooled racks. However, the transition is fraught with friction.

Many operators face a massive retrofit bottleneck. Existing "brownfield" data centers were simply not built to handle the structural weight or plumbing requirements of full immersion cooling, where entire servers are submerged in massive tanks of dielectric fluid. Tearing down and rebuilding these facilities is a multi-billion-dollar proposition that most enterprises cannot afford.

This is where the cassette-based architecture from Airsys enters the equation. By delivering targeted heat removal directly to CPUs and GPUs via a spray mechanism, the system provides the thermal efficiency of liquid cooling without requiring a large-scale facility redesign. It is an incremental, scalable approach that allows facility managers to upgrade specific high-density AI rows while leaving the rest of their air-cooled infrastructure intact.

The Fluid Dilemma: Less is More

Introducing liquids into environments filled with millions of dollars of sensitive electronic equipment has historically been a hard sell. Facility engineers and site reliability managers have long viewed water and synthetic fluids with deep suspicion, fearing catastrophic leaks and complicated maintenance routines.

One of the most compelling claims made by the partnership is that their spray-based approach uses up to 80 percent less dielectric fluid than traditional immersion systems. In the world of data center operations, this fluid reduction is a game-changer.

Managing a full immersion tank involves handling hundreds of gallons of synthetic coolant per rack. While these fluids are non-conductive and safe for electronics, they are heavy, requiring reinforced raised floors. They also complicate routine maintenance; swapping out a failed server component in an immersion tank is a messy, time-consuming process that requires specialized hoists and fluid handling protocols.

By contrast, a spray system contained within a sealed cassette drastically minimizes operational complexity. "The hesitation isn't about whether liquid cooling works; the physics are undeniable," noted one veteran data center facility manager evaluating the upcoming OCP demonstrations. "The hesitation is about the Tuesday morning reality of a technician trying to replace a faulty memory module without spilling synthetic oil all over the hot aisle." By targeting the fluid only where it is needed—at the hottest components—the technology mitigates the risk and the mess, making liquid cooling a more palatable operational reality for skeptical IT teams.

Big Oil's Pivot to the AI Heat Sink

Perhaps the most fascinating angle of this collaboration is the aggressive push by energy majors into the data center space. For over a century, Shell's core business has been fueling internal combustion engines and industrial machinery. But as global energy transitions accelerate and the internal combustion engine sunsets, oil and gas giants are desperately seeking new growth vectors.

The AI thermal management market—projected by industry analysts to reach up to $27 billion by the early 2030s—presents a massive, untapped opportunity. Thermal management is rapidly replacing compute power as the primary bottleneck of data center expansion. Recognizing this shift, Shell executed a strategic pivot in 2025 to move its AI strategy from an internal optimization tool to an offensive commercial strategy.

The fluid powering the upcoming demonstration is a product of this pivot. Developed using proprietary Gas-to-Liquids (GTL) technology, these synthetic dielectric fluids are engineered for extreme thermal stability and long-term corrosion protection. The company has already secured official certifications from major chipmakers like Intel, proving that their fluids can reliably cool next-generation processors over extended lifecycles.

They are not alone in this gold rush. Competitors like ExxonMobil and TotalEnergies are also pouring resources into developing advanced dielectric coolants and biosourced fluids. The message is clear: the future of Big Tech is inextricably linked to the chemical engineering prowess of Big Oil. As data centers consume more power and generate more heat, the companies that once lubricated the industrial revolution are now positioning themselves to cool the artificial intelligence revolution.

A Glimpse into the Future at OCP

The industry will get a firsthand look at this convergence during the OCP Global Summit's Liquid & Immersion Community Outreach event, aptly titled "Keeping it Liquid: Forever Fluids," on October 14. Through a specially designed clear cassette, attendees will witness a live, running server being cooled by spray technology and synthetic fluid in real-time.

This demonstration represents more than just a clever engineering trick. It is a vital proof of concept for an industry that is rapidly running out of thermal headroom. The ability to retrofit existing facilities with scalable, low-fluid liquid cooling architectures could be the key to democratizing AI infrastructure, allowing enterprise and mid-tier data centers to compete with hyperscalers.

As the boundaries between energy suppliers, cooling manufacturers, and digital infrastructure providers continue to dissolve, partnerships that bridge these sectors will become the new normal. The AI boom has created a thermal crisis, and solving it will require a synthesis of chemical innovation and mechanical engineering that transcends traditional industry silos.