Ecolab ties Siemens power and Dassault design tools to its liquid cooling system
Ecolab announced an open partner ecosystem for liquid cooling at SC26 on November 17, 2026. Each partner connects into its first smart thermal management system for direct liquid cooling. Siemens adds power distribution, Alfa Laval works with Ecolab on next-generation CoolIT CHx 3000 coolant distribution units, and Dassault Systèmes provides a virtual twin that measures trade-offs in water per token. Ecolab released no pricing, delivery schedules or measured results from installations.
This story was produced through MarketScale. See how Energy teams put it to work with Customer Stories & Case Studies.
Key takeaways
Ecolab said Siemens power distribution, Alfa Laval heat exchange technology and a Dassault Systèmes virtual twin connect into its first smart thermal management system for direct liquid cooling. For operators, the announcements suggest an Ecolab cooling decision could pull power and modeling choices into the same integrated package—depending on their existing vendors and tools.
Water per token turns a cooling design choice into a number the people paying for the compute can read, but it becomes a market benchmark only once an operator publishes the figure from a running hall.
The Siemens power integration matters most to teams already planning on Siemens electrical gear; teams on another power vendor should ask what an 'open' ecosystem means in the contract.
Free workspace
Turn your Energy expertise into content.
Record interviews, organize footage, and write with AI on a free trial of the MarketScale platform for qualifying companies. No demo required, no credit card.
A 3-megawatt coolant distribution unit sits at the center of Ecolab's pitch to AI data center builders, and as of this week it has company. At SC26 on November 17, Ecolab announced an open ecosystem of partners, each assigned a different stage of the data center lifecycle: Siemens on power, Alfa Laval on heat exchange and Dassault Systèmes on virtual twin modeling. All of them connect into what Ecolab calls its first smart thermal management system for direct liquid cooling.
IT Brief Asia's Joseph Gabriel Lagonsin reported that Ecolab has launched an integrated cooling platform. It pairs Ecolab's fluid management technology, software and service network with direct liquid cooling equipment from CoolIT. The core of the system is the CoolIT CHx 3000, a coolant distribution unit rated at 3 megawatts, with Ecolab's 3D TRASAR technology built in. Together, these pieces aim to let operators see how their cooling systems are performing in real time.
The stated payoff is threefold: better reliability, tighter control of water and energy use, and higher computing density as AI workloads grow. Ecolab unveiled that platform on November 16, the day before the partner news, and it completed its acquisition of CoolIT Systems on July 2, 2026, which makes the unit an Ecolab product line.
The order of those two announcements matters to whoever is writing the mechanical and electrical specs for the next AI hall. Ecolab chief executive Christophe Beck said the problem is that cooling hardware, fluid management and digital technology have operated separately for too long. Combining them, he said, is meant to give customers more computing power more quickly, with lower water and energy use. "And we're just getting started," he added.
The partner list pushes that argument outward to power and design. So the real question for a buyer is how much of the build should arrive already connected.
Direct liquid cooling uses a coolant loop to remove heat from processors, as Kalkine describes it. IT Brief Asia notes it can pull heat away more efficiently than air-based systems in high-density environments, which is why it draws more attention as AI systems pack equipment more densely and carry higher power loads.
Three partners, one coolant unit
Ecolab said each partnership addresses a separate stage of the data center lifecycle. Alfa Laval's work ties most directly to the coolant distribution unit. According to Ecolab, Alfa Laval was brought into the initiative behind its next generation of CoolIT CHx 3000 units because of the company's expertise in heat exchangers and flow components. The pairing combines Alfa Laval's know-how in heat exchangers, pumps and valves with Ecolab's own thermal loop work, which runs from facility cooling water treatment down to direct-to-chip cooling.
Alfa Laval chief executive Tom Erixon tied the collaboration to heat-transfer reliability at scale, to performance and uptime across the data center's lifecycle, and to heat re-use. That last item is worth flagging for any operator with a heat recovery plan in the design.
Siemens covers the electrical side. The two companies describe a single, deployable power and cooling package for high-density AI workloads. It pairs Siemens power distribution, automation, software and digital twin tools with Ecolab's liquid cooling and CDU technology. They say the package bridges the physical and digital layers of the data center and aims to cut deployment complexity, which they bill as IT/OT convergence at scale: the information technology running compute meeting the operational technology running the plant.
Peter Koerte, who leads Siemens Smart Infrastructure and sits on the Siemens managing board, argued that the heat and energy loads of AI have made piecemeal data center designs obsolete. He named building high-performance AI infrastructure fast, efficiently and sustainably as the industry's biggest bottleneck, and pitched the joint offer as a scalable blueprint that cuts water and energy use while shortening time to token. That phrase, like Beck's "design to first token," refers to the stretch from a facility's design to the moment its compute starts producing output.
Read together, the three deals suggest Ecolab wants the CDU decision to bring the power and modeling decisions along with it. For a team already planning around Siemens switchgear and automation, that could cut integration work. For a team standardized on another power vendor, the Siemens piece matters less, and the useful question becomes what an "open" ecosystem means once a contract is on the table.
One detail deserves a closer look. Siemens brings digital twin capabilities of its own, and Dassault supplies the cooling twin. Two modeling partners inside one ecosystem could be a strength if the models share inputs, and extra work for a design team if they have to be kept in step by hand.
Scoring cooling in water per token
Dassault Systèmes works upstream, before anything is bought, and its contribution is the most novel. Ecolab bills it as the industry's first comprehensive cooling loop. The virtual twin Ecolab brought to Supercomputing 2026 links two models, the technology cooling system and the facility water system, and simulates performance from rack design inputs through the CDU and coolant selection. Ecolab describes it as physics- and chemistry-based, built to identify the right CDU architecture, the right coolant and the safest facility water temperature.
Ecolab built its virtual twin with Dassault Systèmes. The twin connects models of the facility water system (FWS) and the technology cooling system (TCS), and it follows performance starting with rack design inputs, then the coolant distribution unit (CDU), and on to coolant selection. The model is designed to help identify the optimal CDU architecture and the safest facility water temperature. That suggests the CDU is where decisions about both loops come together, which could explain why Alfa Laval's components and Dassault's model both center on the same unit.
The twin's measurements are the interesting part. It lays out trade-offs among water use, coolant performance, power efficiency and CDU redundancy, then expresses them as water per token, token efficiency and time to token capacity. Those units tie the cooling plant to what the hall actually produces. Water per token turns a cooling design choice into a number the people paying for the compute can read.
That framing lands in a live argument. According to IT Brief Asia, the industry is in a wider debate over AI's electricity and water footprint, and operators and suppliers face pressure to prove they can meet growing demand for computing while using resources more efficiently. A metric that divides water by output, rather than reporting gallons alone, suggests a way to defend a larger facility on efficiency grounds even as its total draw grows.
Elisa Prisner, Dassault Systèmes' executive vice president for strategy, industry, marketing and transformation, made the timing argument: decisions for these facilities have to happen earlier and with the full picture, and a twin that joins chemistry, thermal engineering and water management in one model lets these "AI factories" be designed to be more efficient from the start. The practical implication is that coolant chemistry and facility water temperature become concept-stage inputs, sitting next to rack density.
Ecolab published no reference values for water per token, so for now the metric compares design options within one model, and a figure from a running site is what would make it comparable across sites.
The size of Ecolab's high-tech bet
For operators weighing a platform that spans hardware, fluids and monitoring, where the supplier is headed could matter as much as the product itself. IT Brief Asia reported figures that show how far Ecolab is leaning into this business:
Ecolab's high-tech unit in numbers
IT Brief Asia
The $4 billion goal is not only a press interpretation. Ecolab put it in the headline of its own July 2 release announcing the CoolIT close, which described the global high-tech business as targeting $4 billion by 2030. For scale, Ecolab's corporate profile reports $16 billion in annual sales, 48,000 employees and customers in more than 170 countries and 40 industries. Set against that base, a unit approaching $1.5 billion is still a minority of the company, and IT Brief Asia reads the targets as a push beyond Ecolab's established markets in water, hygiene and infection prevention.
Acquisitions also play a clear part in how Ecolab is building this business. The direct liquid cooling hardware came with CoolIT, while Ovivo Electronics added ultrapure water systems for semiconductor manufacturing. Ecolab also sells water and process technologies used in power generation, and says it helps protect a quarter of the power generated. It describes its role in AI as supporting the full water needs of advanced computing, from ultrapure water for chip manufacturing to direct liquid cooling that it says reduces environmental impact through circular water use.
That power-side footprint helps explain the Siemens pairing: Ecolab already sells into the electricity behind AI, and the new package puts it next to the switchgear inside the hall. For an operator, the 2030 target is a useful signal of where Ecolab plans to spend engineering and service effort over the next few years.
Still, this week's news is a set of partnerships and a design tool announced at a conference. As Kalkine's Sonal Goyal noted, the release includes no pricing, no delivery schedules and no measured results from installations, so the water and energy claims remain goals until operators report numbers from running halls.
What to ask before the next cooling spec
For design and procurement teams evaluating direct liquid cooling for high-density racks, the announcement sharpens the questions to put to Ecolab or any competing supplier:
- Can the virtual twin run on our own rack design inputs, and what water per token and time to token figures does it return for two or three CDU architectures side by side?
- What does 3D TRASAR monitoring on the CHx 3000 report in real time, and how does that data reach our building management and power systems?
- Which CHx 3000 capabilities ship today, and which depend on the next-generation units being developed with Alfa Laval?
- If our electrical design uses a vendor other than Siemens, which parts of the integrated package still apply?
- Do Siemens' digital twin tools and the Dassault-built cooling twin share one set of inputs, or will our team maintain two models?
The thing to watch for is the first operator that publishes a water per token figure from a live hall built on this stack. That one number would give the rest of the market something real to measure its own designs against.
Sources
Your experts belong here
Every story in MarketScale Energy starts with a company putting its field engineers, operations leads, and project developers on the record. Buyers are already reading this topic. The only question is whose experts they find.
Developers and operators shortlist on credibility, and your engineers give your sales team something real to send.
About the author
The MarketScale Newsroom reports on the companies, technologies, and trends shaping 16 B2B industries. It turns primary sources and expert commentary into clear, useful coverage for the people doing the work.