Compal Electronics and ARC Cloud announced on October 27 that they have signed a memorandum of understanding to work together on large-scale AI computing infrastructure, in a deal signed during the 2026 OCP Global Summit in San Jose.
The agreement sets out a framework for collaboration on advanced AI computing systems, high-performance GPU clusters and next-generation data center infrastructure. As part of that broader work, the companies said they will explore opportunities tied to a data center development roadmap of up to 100 MW.
That roadmap aligns with ARC Cloud's plans to keep expanding in the United States through significant capital investments in AI infrastructure and data center development, including in California, Nevada, Texas and Georgia, according to the announcement.
The division of labor is clear. Compal will contribute expertise in AI server manufacturing, rack-scale integration and liquid-cooling technologies. ARC Cloud will bring capabilities in infrastructure investment, strategic technology partnerships, and the engineering and operation of high-performance GPU clusters.
Beyond server supply, the partnership covers system engineering, technical validation, deployment coordination and ongoing support. The companies said they will work to align system configurations, production schedules and data center deployment requirements, and explore cooperation in joint engineering, system validation and lifecycle technical support.
The aim, both sides said, is to deliver computing capacity faster for AI model training, fine-tuning and inference, at a time when customers increasingly need complete environments rather than individual machines.
"AI computing is moving toward increasingly integrated, high-density systems that require closer coordination between technology manufacturers and infrastructure providers," said Alan Chang, vice president of the Infrastructure Systems Business Group at Compal. "Our partnership with ARC Cloud allows us to build on our engineering strengths and explore new opportunities to support the growing NeoCloud market."
William Tiah, co-founder of ARC Cloud, made a similar point from the operator's side. "AI customers need more than access to GPUs. They need computing environments that can deliver consistent performance for demanding workloads," Tiah said. "Working with Compal allows us to bring system engineering and infrastructure planning closer together, helping us build and operate reliable GPU clusters as demand for AI computing continues to grow."
Compal, listed in Taiwan as TWSE: 2324 and founded in 1984, is headquartered in Taipei and describes itself as a Forbes Global 2000 manufacturer of computers and smart devices, with operations in the United States, Taiwan, China, Vietnam, Mexico, Brazil and Poland. It has expanded into cloud servers, automotive electronics, smart medical and healthcare, and advanced communications.
ARC Cloud, short for Accelerated Research Computing, is a U.S.-based full-stack NeoCloud company specializing in the investment, development and operation of large-scale AI computing infrastructure. It provides dedicated GPU compute, managed infrastructure, model deployment and production inference for AI developers, enterprises and cloud service providers.
Investors and customers should read the announcement carefully, however. The companies stressed that the memorandum establishes a framework for cooperation and does not constitute a binding investment or purchase commitment. The potential 100 MW roadmap is exploratory and does not represent a confirmed project, capacity commitment, purchase commitment or revenue forecast, with specific projects, terms and timelines still subject to further agreements.
Even with that caveat, the signing is a useful signal of where AI infrastructure is heading: server makers and NeoCloud operators planning systems, cooling, power and deployment together from the start, rather than handing hardware over the wall.