Panasonic's Tecnair Unveils New CDUs for Data Centers
Panasonic’s Tecnair launches innovative CDUs for data centers, addressing the growing thermal demands of AI. Discover the future of cooling!
The servers powering today's most demanding AI workloads generate heat that would have seemed almost absurd a decade ago. A single GPU rack can now push 100kW or more — and that number keeps climbing. Air cooling, which served the industry well for thirty years, is hitting a wall. The question for data center operators isn't whether to move toward liquid cooling; it's how fast and with what equipment.
Tecnair, a Panasonic-owned Italian cooling firm, just gave that conversation a new reference point.
A Purpose-Built Answer to a Real Problem
Tecnair launched its new range of Cooling Distribution Units this month, available in 400kW and 800kW capacities. On the surface, that sounds like a product announcement. Look closer, and it's a clear signal of where the industry is heading — and how quickly thermal demands are outpacing legacy infrastructure.
A 400kW CDU isn't a niche product for edge cases anymore. It's a baseline requirement for any facility running dense AI compute. For context, a hyperscale data center deploying thousands of NVIDIA H100 or Blackwell GPUs needs cooling infrastructure that can handle rack densities conventional CRAC units simply weren't designed for. Tecnair's new units are built specifically to meet those demands.
Founded in 1994, Tecnair has decades of experience delivering air cooling solutions to data centers and healthcare environments. Its Techline range — which includes perimeter and in-row air cooling units, fan walls, and outdoor chillers — has long been respected in European markets. The acquisition by Panasonic in 2023 injected new resources and a broader strategic vision. This CDU launch is the first major hardware fruit of that partnership.
What the Hardware Actually Does
The technical spec sheet matters here, so it's worth being specific.
Both CDU models offer modular scalability, which is critical for operators who need to expand capacity without ripping out and replacing entire cooling systems. Modular design also means phased capital expenditure — a significant operational advantage when deploying at scale. BMS (Building Management System) connectivity comes standard, enabling real-time monitoring of coolant flow, temperature differentials, and system health. In a world where thermal runaway can cause millions in hardware damage in minutes, that visibility isn't optional.
The units are designed to integrate with Panasonic's ECOi-W free-cooling chillers, which use R1234ze — a low global warming potential refrigerant that's becoming the responsible default across European infrastructure projects. That's not just a sustainability checkbox; it's a hedge against incoming regulatory pressure on high-GWP refrigerants in the EU and beyond.
The headline efficiency figure is a PUE of 1.02 — meaning the cooling infrastructure consumes just 2% overhead energy relative to the IT load it serves. That's exceptional. Industry average PUE sits around 1.58 globally, and even best-in-class hyperscale facilities typically operate in the 1.1 to 1.2 range. Achieving 1.02 at the CDU level reflects how fundamentally more efficient liquid cooling is compared to air-based alternatives at high density. Air cooling circulates massive volumes of conditioned air through server rooms; liquid cooling delivers thermal transfer directly where heat is generated, dramatically cutting the energy wasted in between.
Maintenance Access as a Feature, Not an Afterthought
One detail worth highlighting: Tecnair specifically called out easy maintenance access in its product description. That's not marketing fluff. CDUs that are difficult to service create operational risk — delayed maintenance windows, reluctance to perform routine checks, and harder troubleshooting when something goes wrong. In mission-critical environments, accessibility is part of the reliability equation.
The AI Thermal Problem Is Only Getting Worse
Here's what most product announcements won't say plainly: the thermal density problem in AI data centers is accelerating faster than cooling infrastructure has historically evolved.
When hyperscalers started deploying early AI training clusters five years ago, 10-20kW per rack was considered high-density. Today, liquid-cooled racks running the latest accelerators operate at 50-100kW routinely, with some deployments pushing past 150kW. NVIDIA's next-generation architectures are expected to push those numbers further still.
Air cooling becomes economically and physically untenable above roughly 20-30kW per rack. Beyond that threshold, you need more floor space, more airflow infrastructure, more energy — and you still get worse thermal performance than a properly deployed liquid cooling system. Liquid cooling isn't a premium option for high-performance computing; at the densities AI workloads demand, it's the only option that makes the math work.
This is why Tecnair's CDU launch matters beyond the spec sheet. Operators who are still evaluating liquid cooling as a "future consideration" are already behind. The infrastructure decisions made in 2025 and 2026 will determine which facilities can accommodate next-generation AI hardware — and which will require costly retrofits or become functionally obsolete.
Panasonic's Bigger Bet
What elevates this launch from interesting to significant is what Panasonic revealed alongside it. The parent company is actively developing CDUs with capacities of 1.2MW and above, with order acceptance planned to begin in March 2026.
A 1.2MW CDU is a different category of infrastructure. That's not a unit serving a cluster of GPU racks — it's cooling for an entire AI compute pod or a substantial portion of a hyperscale AI training facility. The jump from 800kW to 1.2MW+ represents Panasonic signaling that it intends to compete for the largest, most demanding deployments in the market: hyperscale builds, national AI infrastructure projects, and the next generation of colocation campuses designed explicitly for AI tenants.
This matters for buyers, too. Historically, operators building at that scale have had limited vendor choice in the liquid cooling space. A well-resourced manufacturer like Panasonic entering aggressively — backed by Tecnair's engineering heritage and European manufacturing — adds real competitive pressure to a market that needs it.
From an insider perspective, the integration between Tecnair's CDUs and Panasonic's broader ECOi-W chiller ecosystem is the product strategy worth watching. Selling CDUs as standalone units is one business. Selling an integrated liquid cooling ecosystem — CDU, chiller, controls, refrigerant — is a much stickier, higher-margin position. That's clearly where Panasonic is aiming.
What Comes Next
The data center cooling market is consolidating around a simple reality: facilities that can't handle 50kW-plus rack densities won't win AI tenants. Operators who move early on liquid cooling infrastructure lock in a competitive advantage that compounds over time — higher revenue per square foot, the ability to attract premium workloads, and lower long-term energy costs.
Tecnair's new CDU range gives operators a credible, technically sound option for that transition at the 400kW and 800kW scale. The forthcoming 1.2MW unit will extend that reach to the largest deployments on the horizon. Neither product exists in isolation — both reflect Panasonic's broader bet that infrastructure-grade liquid cooling is a growth market measured in years, not quarters.
For operators still finalizing cooling strategies for facilities coming online in 2027 and beyond, the specification window is now. The equipment being selected today will determine thermal capacity for the next decade. Waiting for the perfect solution means missing the deployments that define the next generation of AI infrastructure.
Explore the InfraSale Marketplace for the latest cooling solutions!