Foxconn and Intel have unveiled a strategic collaboration to jointly develop next-generation AI infrastructure, marking one of the most consequential pairings of a hyperscale contract manufacturer with a US chipmaker since the boom began. Foxconn chairman Young Liu and Intel CEO Lip-Bu Tan announced the partnership at Computex 2026 in Taipei on June 4, 2026, with effort spanning silicon, rack, system and application layers.
Rack-Scale AI Built On Intel Xeon
At the heart of the agreement is a plan to design and manufacture CPU-dense rack-scale AI systems based on Intel Xeon processors. The companies will combine Intel’s semiconductor capabilities with Foxconn’s manufacturing scale, system integration and global data center deployment footprint. Initial reference designs pair Intel Xeon 6+ CPUs with Intel AI accelerators and high-speed interconnects, alongside liquid-cooled rack architectures aimed at the inference workloads now dominating new build-outs. A demonstration version also pairs Xeon with SambaNova SN-50 Reconfigurable Dataflow Units for cost-optimised AI inference.
Targeting Edge AI, Robots And Smart Factories
The collaboration extends well beyond the data center. Foxconn and Intel said they will jointly define next-generation Edge AI and Physical AI platform architectures, targeting agentic AI workloads, smart factories, smart cities and robotics. They will also explore custom ASICs, SoCs and integrated systems for partners that want full-stack designs rather than off-the-shelf components, a path Foxconn has been opening through its iPhone-supplier-to-AI-platform pivot.
A Strategic Reset For Both Companies
For Intel, the deal is a much-needed proof point for new CEO Lip-Bu Tan that the company can capture meaningful share of the AI compute build-out beyond its existing accelerator business. Wall Street rewarded the announcement with a 4.4% jump on the day. For Foxconn, the partnership accelerates Young Liu’s “3+3+3” strategy that pushes the contract manufacturer into AI, semiconductors and next-generation communications across smart manufacturing, smart EVs and smart cities. It also reinforces Foxconn’s broader push to become an end-to-end systems supplier, complementing existing moves such as the NVIDIA DRIVE Hyperion robotaxi deal in Kaohsiung and follows other Intel collaborations including the SambaNova rack-scale inference demo and the OpenVINO Physical AI framework.
What Has Not Been Disclosed
Foxconn and Intel did not disclose financial terms, a timeline for commercial availability, or the scale of initial deployment. Both companies framed the announcement as a multi-year framework rather than a single project, with deliveries expected to follow Intel’s Xeon 6+ ramp later this year.
Reporting based on coverage from DataCenterDynamics, Foxconn, Intel and Yahoo Finance.