LG joins Microsoft’s AI infrastructure push as Nvidia certifies cooling and battery systems

LG is expanding its AI infrastructure partnership with Microsoft while Nvidia approves its cooling and battery systems for AI data centres. The important shift is that the AI build-out is reaching the less glamorous machinery that keeps powerful servers running: cooling, electricity and factory data.

Watch Desk analysis

What happened

LG and Microsoft agreed to work together on data-centre infrastructure, AI transformation and physical AI, according to the Korea JoongAng Daily. LG plans to supply cooling, power and information-technology infrastructure as Microsoft expands its data centres, with longer-term plans for prefabricated facilities and wider use of Microsoft’s cloud and Copilot tools across LG’s white-collar workforce.

The companies also plan to build a “data flywheel” linking factory-floor information to AI models used in manufacturing, logistics and mobility. That is a concrete industrial-AI angle, not merely two companies standing beside a banner containing the word “ecosystem”.

LG Electronics said Nvidia had certified its 2.5-megawatt coolant distribution unit for the Nvidia DSX AI factory platform. LG Energy Solution also received certification for its JF2 AC Link battery energy-storage system, which is designed to smooth sudden AI loads and maintain power during grid-voltage fluctuations. The report says LG had previously received certification for smaller cooling systems.

Why it matters

AI data centres increasingly need infrastructure designed around dense, power-hungry accelerator clusters. A 2.5-megawatt cooling unit can serve roughly 20 racks rated at about 120 kilowatts each, while battery storage can help absorb rapid changes in demand and protect systems from unstable grid supply.

For readers, the practical point is simple: the AI race is also a race to build reliable factories for computation. Chips may grab the headlines, but a server that overheats or loses clean power is an expensive sculpture.

Our read

This is a meaningful infrastructure development, although the Nvidia certifications are approvals for compatibility and deployment rather than proof of superior performance. The Korea JoongAng Daily report does not provide independent testing of LG’s cooling efficiency, battery economics or the Microsoft partnership’s commercial value.

Still, the combination of Microsoft’s data-centre expansion, LG’s industrial systems and Nvidia’s AI-factory platform shows how the market is broadening beyond model developers. The next competitive advantage may belong to whoever can make the entire stack dependable, from silicon to substation.

What to watch

• Deployment:
  whether Microsoft uses LG systems in named data-centre projects.
• Performance:
  whether LG publishes efficiency, reliability and lifecycle data from operating sites.
• Factory AI:
  whether the proposed data flywheel produces measurable manufacturing gains.
• Power demand:
  how quickly storage and liquid cooling become standard in new AI facilities.

Discussion spark: Should AI infrastructure suppliers be judged mainly by chip performance, or by whether they can deliver reliable cooling and power at industrial scale?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.