Intel used the opening day of Computex 2026 in Taipei to unveil a sweeping rackscale AI inference push, announcing a three-way partnership with SambaNova Systems and Foxconn alongside the launch of its first Intel 18A-built data-center CPU, Xeon 6+. The announcements, made by Intel CEO Lip-Bu Tan on June 2, target the rapidly growing market for cost-efficient agentic inference where, Intel argues, the CPU is reclaiming a central role in the data center.
Intel + SambaNova + Foxconn rackscale AI
The flagship deal is production-ready rack infrastructure that pairs Intel Xeon processors with SambaNova SN-50 Reconfigurable Dataflow Units, with Foxconn handling system integration and also manufacturing a CPU-dense variant for cost-optimised inference workloads that don't need additional acceleration. Intel framed the offering as a direct response to the shift from one-CPU-per-four-GPU training topologies to roughly one-CPU-per-GPU ratios for agentic inference.
Vector Core Compute disaggregated inference demo
Onstage, Intel, SambaNova, Vista Equity Partners and Cambium Capital showcased the first real-world demo of fully disaggregated inference under their new Vector Core Compute enterprise cloud — Xeon 6 CPUs for orchestration, SambaNova SN40 RDUs for decode, and NVIDIA Blackwell GPUs for prefill — operating live from a Los Angeles data center. Together.ai is the first commercial customer, and Vista has secured early access for more than 90 of its portfolio companies.
Xeon 6+ on Intel 18A
Intel Xeon 6+ is the company's first data-center CPU built on Intel 18A. Designed for high-density, scale-out workloads, a single liquid-cooled rack can deliver 36,864 cores in 32U of compute space — what Intel calls the highest agent density available at roughly 100 kW per rack. The chip is positioned for orchestration, concurrency and predictable latency, the load profile that defines real-world agentic deployments.
Deep industry partnerships — Siemens, Hitachi, Echo Neurotechnologies, Greenstone
Alongside the rackscale play, Intel announced new or expanded collaborations with Siemens (chip design for edge devices, HPC and robotics), Hitachi (foundry tools and quantum computing), Echo Neurotechnologies (neuromorphic and brain-computer interfaces) and Greenstone Biosciences (AI-driven drug discovery). Intel also reiterated that over 130 customers have already picked its Core Ultra Series 3 silicon for edge AI and robotics designs — a thread that runs through its OpenVINO physical AI framework rollout and recent Crescent Island GPU launch.
Reporting based on coverage from Intel Newsroom, Tom's Hardware, Dataconomy and Focus Taiwan.