The collaboration targets up to 340MW of AI-ready capacity by 2028, with plans to scale beyond 3GW through integrated data centre, power and semiconductor technologies.
Overwatch Capital and Dublin-based semiconductor company CHIPX have entered a strategic partnership to advance high-density, AI-ready data centre infrastructure for sovereign AI campuses and cloud platforms. The collaboration aims to align data centre development, power delivery and next-generation semiconductor technology to address the growing infrastructure demands of enterprise and government AI deployments.
The partnership brings together Overwatch’s expertise in developing and operating data centres with CHIPX’s semiconductor design and reference architecture. Its focus is on improving how available power is converted into usable AI computing capacity, particularly as high-density workloads place increasing pressure on energy infrastructure.
A key area of collaboration will be the adoption of 800V DC power distribution, an emerging approach designed to reduce power-conversion stages between the grid and processors, potentially enabling more electricity to reach computing workloads.
“The limiting factor is power. Our partnership with CHIPX connects semiconductor innovation with the mission-critical infrastructure required to deploy sovereign AI capacity at scale.”
— Sean Lyle, Managing Partner, Overwatch Capital
Overwatch plans to develop U.S.-based campuses, including on-site power solutions for markets constrained by grid capacity. CHIPX will contribute technologies addressing power conversion, data movement, memory and thermal management through its CHIPX Connect platform.
“Data centres today face multiple choke points, all falling under the general category of power,” said Chinmoy Baruah, CEO, CHIPX, highlighting the platform’s focus on power delivery, data transmission and thermal management.
The companies will jointly pursue infrastructure planning, site strategy, technology integration and customer engagement. They plan to deploy up to 340 megawatts by 2028 and scale capacity to more than 3 gigawatts over the following years, targeting secure and scalable infrastructure for the expanding global AI economy.