There is no publicly announced formal partnership agreement between Nvidia and Mitsubishi Heavy Industries for AI data centers, but MHI exhibited at Nvidia GTC 2026 and established a Dallas base to collaborate with U.S. MHI has deployed Japan's first GPU servers using two phase direct to chip cooling with EXEO Group...
Research answer

Create a landscape editorial hero image for this Studio Global article: Search & fact-check with cited sources for What potential partnership is being discussed between Nvidia and Mitsubishi Heavy Industries (MHI. Article summary: ## Nvidia–MHI Partnership Discussion. Topic tags: general, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers, clickbait thumbnails, icons, and tiny thumbnail layouts. Make it useful as an illustrative visual, not as factual evidence.
Mitsubishi Heavy Industries (MHI) is aggressively positioning itself to supply cooling and power infrastructure for Nvidia-powered AI data centers, even though a formal partnership between the two companies has not been announced. MHI's strategy includes exhibiting at Nvidia's premier AI conference, establishing a U.S. base in Dallas, and deploying advanced two-phase liquid cooling technology for next-generation GPUs.
There is no publicly announced formal partnership agreement between Nvidia and MHI for AI data center infrastructure. However, MHI exhibited at Nvidia's GTC 2026 conference in San Jose (March 16–19, 2026), showcasing its data center cooling and power solutions to the Nvidia ecosystem M. MHI's Dallas base was established specifically to "reinforce collaborations with leading companies and advanced technology firms in the United States," which implicitly targets AI leaders like Nvidia N. MHI also offers a one-stop solution portfolio for AI data centers covering power, cooling, and digital management A.
MHI's two-phase direct-to-chip liquid cooling (DLC) uses refrigerant evaporation in cold plates attached directly to processors. It achieves a partial Power Usage Effectiveness (pPUE) of 1.05 and handles thermal design power (TDP) up to 1,000W per processor — critical for next-generation Nvidia GPUs JM. The closed-loop system significantly reduces both power and water consumption compared to traditional air cooling M. MHI's Spectra explainer notes two-phase DLC removes "significantly more heat" than single-phase by evaporating refrigerant in the cold plate S.
MHI also offers 10MW-class centrifugal chillers designed for gigawatt-scale AI computing, specifically targeting the North American market A. In partnership with KDDI, MHI commercialized single-phase liquid immersion cooling that cuts server cooling energy use by 94% and achieved a PUE of 1.05 in trials, now being deployed at industrial scale in Osaka AJ. MHI's integrated solution also includes power, control, and electrification engineering N.
In May 2025, MHI established a new strategic and business base in Dallas, Texas, under Mitsubishi Heavy Industries America (MHIA) to drive its data center business in the U.S. The base provides power, cooling, control, and electrification products and services, targeting collaborations with American tech leaders NA.
In December 2025, MHI and EXEO Group built and began commercial use of Japan's first GPU servers with two-phase DLC, a landmark validation of the technology for AI workloads M. The system enables stable cooling of 1,000–1,400W GPUs, reducing fan power, lowering PUE, and mitigating CO2 emissions L.
In February 2026, MHI launched DIAVAULT, an industrial-grade edge data center platform designed for on-site AI inference in on-premises environments like manufacturing plants. It is a secure, high-performance edge system enabling local data and AI utilization NI. DIAVAULT can scale from small installations to multi-megawatt inference facilities I.
Multiple analyst reports project strong growth for data center liquid cooling, driven by AI adoption and high-density computing. The most commonly cited consensus forecast for the broad data center liquid cooling market over the 2026–2033 period comes from MarketsandMarkets (July 2026) and GlobeNewswire (May 2026), both projecting growth from $4.07 billion in 2026 to $27.65 billion by 2033, at a 31.5% CAGR MG. Asia Pacific is expected to be the fastest-growing regional market at a 32.8% CAGR M.
Other analyst forecasts offer variations. Yahoo Finance reported $29.2 billion by 2033 at 26.4% CAGR F. A PRNewswire report focused on AI-specific liquid cooling projected $11.2 billion by 2033 at 17.2% CAGR P. GlobeNewswire also published a projection of $38.4 billion by 2033 at 28.7% CAGR G. Direct-to-chip cooling, the method MHI uses, captured 42.85% of the liquid cooling market share in 2025 M.
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
There is no publicly announced formal partnership agreement between Nvidia and Mitsubishi Heavy Industries for AI data centers, but MHI exhibited at Nvidia GTC 2026 and established a Dallas base to collaborate with U.S.
There is no publicly announced formal partnership agreement between Nvidia and Mitsubishi Heavy Industries for AI data centers, but MHI exhibited at Nvidia GTC 2026 and established a Dallas base to collaborate with U.S. MHI has deployed Japan's first GPU servers using two phase direct to chip cooling with EXEO Group and launched DIAVAULT, an industrial edge data center for on site AI inference.
Multiple analyst reports forecast rapid growth in liquid cooling, with MarketsandMarkets and GlobeNewswire citing a 31.5% CAGR from 2026 to 2033.