MiTAC Presents Modular AI Data Center and Waterless Two-Phase Cooling PoC at OCP Global Summit 2026
Редакция ЦОД:Главное
MiTAC Computing has presented an integrated AI-infrastructure stack at OCP Global Summit 2026, including the TonoForge modular data center developed with Tonomia and a liquid-cooling proof of concept with ZutaCore. The offering combines liquid-cooled MiTAC G4826Z5 AI servers, rack integration and chip-level cooling for high-density GPU deployments.
- TonoForge is a modular data center developed by MiTAC Computing with Tonomia and built around liquid-cooled MiTAC G4826Z5 AI servers.
- MiTAC cites a 12-week TonoForge deployment as evidence of accelerated GPU-cluster delivery.
- MiTAC and ZutaCore demonstrated a proof of concept using HyperCool Waterless Two-Phase Liquid Cooling.
- ZutaCore’s cited thermal metrics—more than 4,000 W per chip and more than 250 W/cm² average heat flux—are potential capabilities, not independently verified operating results.
MiTAC Computing used OCP Global Summit 2026 to present infrastructure components for scalable AI factories, distributed modular data centers and liquid cooling.
The central deployment concept is TonoForge, a modular data center developed with Tonomia. It combines liquid-cooled MiTAC G4826Z5 AI servers with rack integration. MiTAC said a proven TonoForge deployment took 12 weeks, positioning modular construction as a route to faster deployment of high-performance GPU AI clusters.
Separately, MiTAC demonstrated a proof of concept integrating the G4826Z5 server with ZutaCore’s HyperCool Waterless Two-Phase Liquid Cooling technology. According to MiTAC, the technology has the potential to support more than 4,000 W per chip and an average heat flux exceeding 250 W/cm².
The announcement reflects the market’s shift toward integrated AI-infrastructure stacks in which servers, cooling and rack integration are supplied as a combined system. For operators designing dense GPU environments, modular deployment and chip-level thermal management address two linked constraints: the time required to bring infrastructure online and the ability to remove heat at rising processor power levels.
The stated performance figures remain supplier claims. The available information does not identify the location, customer, MW scale or operating status of the cited 12-week TonoForge deployment, and the cooling figures are described as technology potential rather than verified commercial operating results.
As AI compute density rises, deployment speed, rack-level integration and heat removal are becoming material infrastructure constraints. MiTAC’s approach points to an integrated delivery model spanning servers, modular data center architecture and cooling; if its stated deployment and thermal-performance claims are validated in commercial deployments, it could help reduce constraints on high-density GPU cluster rollouts.
- — / 13 октября 2026 г. · АнглийскийПервоисточник →
Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.