NVIDIA's liquid-cooling work with Vertiv can deliver more compute from a fixed power budget. Liquid cooling yields roughly 22% more compute and more tokens from the same power on average, a benefit modeled in an April report. NVIDIA's scale, balance sheet and command of its supply chain make it the economic winner of the AI compute cycle, with Broadcom a close-ish second. x.com  x.com 

Liquid cooling is the must-have upgrade for AI data centers, making Vertiv, Eaton, Trane and Modine the next big AI infrastructure winners. It unlocks racks above 100kW, cuts energy use and keeps hyperscalers ahead of the heat. Vertiv and Modine span direct-to-chip cold plates and immersion cooling, Vertiv, Eaton and Trane supply coolant distribution units, Trane and Vertiv build facility chillers and heat rejection, and Vertiv and Eaton offer integrated thermal management. x.com 

Data center cooling stocks lose as NVIDIA's Vera Rubin chips need no chillers. Per @HedgieMarkets, Jensen Huang said Rubin can be cooled with 45°C water, sending Johnson Controls down 6.2%, Modine 7.4%, Trane 4% and Carrier nearly 1%; the chips ship in the second half of 2026. With NVIDIA reporting $1 trillion of Blackwell and Rubin orders through 2027, AI capex is doubling, and on an Elliott Wave view the cooling stocks may be starting their own wave 3 to the downside. x.com  x.com 

Inference's move toward edge devices and dense, memory-hungry racks should drive demand for specialized chips, links, power delivery and cooling. McKinsey projects inference at 60% of AI computing demand by 2030. Ambarella makes on-device AI chips; Astera Labs links GPUs, memory and networks; Credo supplies short copper connections; Rambus earns from memory interfaces; Navitas reduces power waste; Modine cools dense racks; and Super Micro builds inference-specific server racks at volume. x.com  … Show more

Vertiv benefits from every AI data-center buildout because rising rack power makes power management, UPS, distribution and liquid cooling, its core businesses, ever more critical. Wall Street keeps raising estimates, with a consensus target near $375 and revenue seen rising from about $10.8B to over $21B by 2028. NVIDIA's shift to 800V DC power, which Vertiv already supports, and rising liquid cooling add growth. A months-long wedge could break out on Aug 5 earnings if orders and guidance beat. … Show more

Vertiv could be a big thematic loser if data centers move to space, where liquid cooling is not needed. Given the SpaceX IPO and Vertiv's valuation above 45x NTM earnings, its terminal value deserves questioning. x.com 

AI data-center spending should reach chips first, then racks, power and cooling, benefiting Nvidia, AMD, Vertiv, Bloom Energy, Credo, Marvell, Lumentum and Micron. BofA forecasts the global data-center market at about $1T in 2026, $2.1T in 2028 and nearly $3T by 2030, a 37% annual rise; roughly 80% of a new AI site's $57M per MW cost goes to servers, networking and storage. x.com 

AI data-center spending flows beyond GPUs and operators into networking, optics, power and cooling. Nvidia, AMD, Broadcom and Marvell supply chips; Arista, Nokia and Credo connectivity; Applied Optoelectronics and Lumentum optics; Vistra, Constellation and Bloom power; Vertiv and Modine cooling. Meta's $140B spending this year illustrates the scale, per a commenter with whom the point is agreed. x.com  x.com 

Vertiv and Vicor hold the leverage as AI's bottleneck shifts to rack-level power and cooling, whichever custom chip wins. Amazon CEO Andy Jassy said the bottleneck is no longer just chip availability but rack-level power delivery, thermal management and silicon efficiency, as AWS custom silicon hits a $25 billion run rate. At 100-plus kW and over 1,000 amps per rack, Vertiv's direct-to-chip liquid cooling and Vicor's vertical power delivery become critical; air cooling fails above 80-100 kW. … Show more

Liquid-cooling suppliers AVC, LOTES and FIT, plus ODMs Hon Hai and Supermicro, benefit as Nvidia's Vera Rubin racks sharply raise cooling content. GPU power rises from 1,000-1,200W for Blackwell to up to 2,300W for Rubin and 2,600W for Rubin Ultra, with racks heading toward 600kW-1MW. VR NVL72 cold plates are about $34.2k per rack, up 58% more for VR300; quick-disconnect content is about $10.8k per rack, up 77% versus GB300, and another 103% for VR300. Fositek holds over 50% QD share. x.com