Airsys Wins Inaugural Ad Astra: AI Infrastructure Competition

Per aspera ad astra: through hardship, to the stars. It’s a fitting concept for this industry, where the work that gets us somewhere extraordinary can look ordinary. Cooling systems. Fabrics. Data layers. Control planes. These critical systems often don’t get the attention they deserve given they determine how far AI can actually go. When we launched Ad Astra: AI Infrastructure Competition as an official media partner of the 2026 AI Infra Summit, we wanted to aim a light at that work.

We asked for 200 words. Make the case for why your technology belongs among this year’s most significant advances in AI infrastructure. Our judging panel composed of TechArena Advisors evaluated every entry against three criteria:

• alignment to one of the summit’s focus areas

• real-world impact backed by measurable outcomes

• clarity that a senior technical or business decision-maker could absorb without decoding jargon

Today I’m pleased to announce the results.

Winner: Airsys

Airsys took the top spot with LiquidRack, a liquid cooling system built around a straightforward observation: AI is driving rack densities that existing cooling infrastructure was never designed to handle.

Rather than routing cooling through centralized facility infrastructure, LiquidRack integrates it directly into the rack. Airsys makes the case that this changes the deployment math for operators, who can add AI capacity incrementally, expand over time, and extract more compute from the power they already have rather than overhauling an entire facility to get there.

The company backs the approach with specifics. Airsys reports that its patented spray cooling technology transfers heat up to three times more efficiently than conventional approaches, freeing more power for compute. It cites a PUE below 1.02, no water consumption, and roughly 80% less dielectric fluid than immersion cooling. The company also notes that LiquidRack has been validated in customer deployments across multiple industries and geographies, which moves the conversation past the lab.

Our TechArena Advisor Laura St. John gave the following reasoning for choosing Airsys as the winner:

“Airsys’s LiquidRack paired a clear innovation with the strongest real-world evidence in the pool, backed by validated deployments across multiple industries, not just projections. In a field tackling one of AI infrastructure's biggest bottlenecks, cooling density, LiquidRack stood out as ready to deploy today.”

Honorable Mentions

Three additional companies earned recognition from our Advisors.

VAST Data argued that AI has transformed computing while much of the underlying data infrastructure still reflects an earlier era of enterprise IT, with separate storage systems, databases, event streams, and vector stores endlessly copying the same information. The VAST AI OS collapses those layers into a single shared platform. The company says it now powers more than 3 million GPUs worldwide, including AI factories where one system serves over a billion CUDA cores.

Cornelis Networks made a data movement case that emphasizes the network’s critical role in the AI era. The company points to production AI clusters running at 30 to 50 percent model FLOPs utilization, with the gap lost to congestion and synchronization stalls. Its CN6000 SuperNIC carries Omni-Path, RoCEv2, and Ultra Ethernet on a single 800 Gbps adapter. Cornelis modeling indicates AllReduce collectives complete roughly 24 percent faster than standard Ethernet, which the company translates to about 13% less training time for a 250-billion-parameter model on a simulated 10,000 GPU cluster.

Rafay addressed the operational sprawl of running an AI data center, unifying Kubernetes, virtual machines, SLURM, bare metal, GPUs, and AI services under one control plane, with Token Factory extending that model into inference by turning GPU capacity into metered, standardized services. In production environments, Rafay reports improving GPU cluster efficiency by 25% to 40% and supporting three to five times more tenants without adding headcount.

The TechArena Take

The binding constraint on AI infrastructure has moved off the chip and into everything surrounding it, and the companies competing hardest right now are the ones working that perimeter. Airsys addressed thermal and power headroom, VAST Data addressed the data layer, Cornelis Networks addressed movement between GPUs, and Rafay addressed the operational model that holds the whole thing together. Congratulations to Airsys, VAST Data, Cornelis Networks, and Rafay. We will see you in Santa Clara this September.

Read more about our winners here.

Free Tickets and Discounts from TechArena

If you haven’t yet registered for the AI Infra Summit, now is the time to act! As part of our partnership, we’re pleased to offer benefits to our TechArena followers who want to join the conversation in Santa Clara this fall:

• Free expo tickets (valued at $447) are available to qualified individuals working within the AI infrastructure field. Apply for free expo tickets.

• A registration discount is available for full access and VIP tickets. Quote TECHARENA15 on the registration page to save 15%.