Intel and Google just locked in a multiyear expansion of their AI infrastructure partnership, TechCrunch AI reports. Google Cloud will keep running on Intel’s Xeon processors, including the latest Xeon 6 chips, for AI, cloud, and inference workloads. The two companies will also deepen their co-development of custom infrastructure processing units (IPUs).
This isn’t a new relationship. Google has used Intel’s Xeon chips for decades, and the IPU co-development track started back in 2021. What’s new is the expanded scope and the timing.
Why this matters right now
The AI industry has a CPU problem. Everyone’s been focused on GPUs for training models, but running those models at scale? That takes CPUs. And there aren’t enough of them.
“AI is reshaping how infrastructure is built and scaled,” Intel CEO Lip-Bu Tan said in a 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.”
He’s not wrong. As AI inference demand explodes, the bottleneck is shifting. Training a model is a one-time cost. Serving it to millions of users is ongoing, and that’s where CPUs do the heavy lifting.
The IPU angle
The custom ASIC-based IPUs are worth watching. These chips offload data center management tasks from CPUs, freeing up compute for actual workloads. Think of them as specialized traffic controllers for data centers. Google and Intel building these together signals that hyperscalers want more control over their infrastructure stack, not less.
Intel declined to share pricing details, according to TechCrunch AI. That’s typical for deals at this scale, but it also makes it hard to gauge how much Intel is willing to discount to keep Google in its ecosystem.
The bigger picture
Google isn’t the only company feeling the CPU crunch. Arm Holdings recently unveiled the Arm AGI CPU, the first chip the semiconductor giant has produced itself. SoftBank-backed Arm making its own silicon is a clear signal: demand for AI-capable CPUs is outpacing supply.
For Intel, this deal is a lifeline and a statement. The company has struggled to compete with Nvidia in the GPU race, but CPUs remain its core strength. Positioning Xeon as essential AI infrastructure, not just legacy server hardware, is smart strategy.
For Google, it’s about securing supply. When every cloud provider and AI company is scrambling for compute, locking in a multiyear CPU partnership removes one variable from the equation.
What to watch
The shift from GPU-centric to balanced AI infrastructure is accelerating. Companies building AI products should pay attention to inference costs and CPU availability, not just training budgets. The next bottleneck in AI scaling won’t be model quality. It’ll be the infrastructure to serve those models at scale.
More details are available in the original report from TechCrunch AI.