- Google Cloud will deploy Intel’s Xeon 6 processors for AI, cloud, and inference workloads under the expanded multiyear agreement.
- The two companies will extend co-development of custom ASIC-based infrastructure processing units (IPUs), a program that has been underway since 2021.
- Intel CEO Lip-Bu Tan stated CPUs and IPUs are central to meeting the demands of modern AI infrastructure, citing the need for balanced systems beyond GPU accelerators.
- The announcement comes as the AI industry faces a growing shortage of CPUs, which are critical for inference workloads and data center management.
What Happened
Google and Intel on Thursday announced an expansion of their existing multiyear technology partnership, deepening collaboration on both processor deployment and jointly developed silicon. According to reporting published April 9, 2026 by TechCrunch, Google Cloud will continue deploying Intel’s Xeon processor family—including the recently released Xeon 6—across AI, cloud, and inference workloads. The two companies will also broaden their co-development of custom infrastructure processing units (IPUs), with Intel CEO Lip-Bu Tan confirming the deal in a company press release.
Why It Matters
The expansion arrives as CPU supply has become a material constraint on AI deployment across the industry. While GPUs dominate model training infrastructure, CPUs play a comparably important role in inference—running trained models in production—and in managing data center operations at scale. Demand for CPUs has outpaced supply in recent months. Arm Holdings, owned by SoftBank, moved to address the same gap, announcing the Arm AGI CPU in the same period: the first chip the semiconductor company has produced itself, a sign that the industry broadly recognizes that AI infrastructure requires more than GPU accelerators.
Technical Details
The IPU co-development program, which began in 2021, has shifted focus toward custom ASIC-based designs under the expanded agreement. IPUs are purpose-built hardware distinct from both CPUs and GPUs: they handle data center management tasks—including network packet processing, storage orchestration, and workload offloading—freeing CPUs to focus on compute-intensive inference operations. Intel’s Xeon 6, the latest generation of its flagship server chip line, will be deployed across Google Cloud AI and cloud workloads. Intel declined to disclose pricing terms or the financial scope of the deal. “AI is reshaping how infrastructure is built and scaled,” Tan said in the press release. “Scaling AI requires more than accelerators — it requires balanced systems. CPUs and IPUs are central to delivering the performance, efficiency and flexibility modern AI workloads demand.”
Who’s Affected
Google Cloud customers running AI inference workloads—enterprise clients deploying large language models and AI-assisted applications at production scale—are the most directly affected. If the Xeon 6 and ASIC-based IPU integrations perform as both companies described, those customers would see efficiency gains in per-query compute cost and latency. For Intel, the agreement provides a high-visibility anchor customer at a time when the company faces competitive pressure in the server chip market from Nvidia’s H-series GPUs and AMD’s EPYC line. Because Google also develops its own Tensor Processing Units (TPUs) for internal training workloads, the Xeon and IPU relationship is primarily positioned around inference and data center infrastructure management rather than model development.
What’s Next
Intel did not provide a deployment timeline for Xeon 6 availability within Google Cloud or for the next generation of co-developed ASIC-based IPUs. Custom silicon programs of this type typically require multi-year development cycles from specification to production readiness, meaning the performance claims associated with the new IPU program cannot yet be independently verified. Google has not disclosed whether the co-developed IPUs will be made available to third-party Google Cloud customers or remain internal infrastructure.
