Recently, an interesting report was released stating that Intel is now binning down and selling high-performance Xeon dies that would otherwise have been scrapped by customer requests. What challenges is the AI industry presenting to supplies, what exactly have customers asked of Intel, and what does this tell us about the ongoing situation?
What Challenges is The AI Industry Presenting to Supplies?
The ongoing demand for AI across the globe is mind boggling. Stock markets are ballooning, data-centers are getting built daily, and the power that these mammoths of technology consume is unbelievable.
But while all of this is happening, the supply of key electronic components is dwindling. RAM, for example, has seen major issues now that manufacturers are switching over to the production of HBM RAM, specifically for AI. GPU manufacturers, such as NVIDIA, are failing to make major improvements in their tech, as they look towards AI hardware.
SSDs and NVMe drives are now all increasing massively in price, virtually destroying the PC market, and now, CPU availability is shrinking, making it harder to get the next generation of Intel CPUs for both desktop and server environments.
As such, it is now becoming increasingly more difficult for both businesses and individuals alike to get new hardware. In fact, the situation has become so bad that engineers are now turning to DDR3 in their designs, just so they can keep producing equipment.
Customers Asking Intel for Scrapped CPU Chips
Recently, it was reported that Intel is seeing increasing revenue through the sale of CPU chips that would otherwise be discarded or used in low-end products. This shift is primarily driven by surging demand for CPUs in AI workloads, especially inference, which has expanded beyond GPUs to require more CPU and memory resources. Intel’s Xeon processors, commonly deployed in datacenters and AI infrastructure, are now in high demand as a result.
Typically, chips located at the edge of a wafer have lower yields and are not as good as those in the center. As such, they are downgraded to be used in low-end products, but when supply becomes tight, even these are being purchased.
Intel has been able to successfully bin down these lower quality dies and sell them to customers, demonstrating the severity of the shortage. Strong demand means buyers are accepting lower-performance chips to meet capacity needs, highlighting how acute the CPU supply crunch has become. This approach of selling previously discarded dies has contributed to better-than-expected financial performance for Intel, alongside improvements in manufacturing execution and supply chain management. In addition, it is likely that other companies, including AMD, are also looking at lower-tier dies and foundry production as a potential source of additional CPUs.
This strategy turns what would have been manufacturing waste into a new revenue stream for Intel, underscoring the extent to which AI-driven demand is reshaping semiconductor economics. Other chip-makers are likely adopting similar tactics to meet customer needs as well.
What Does This Reveal About the Ongoing Situation
If there is one fact that this story demonstrates, it is that chips previously considered undesirable are now being purchased out of necessity by customers operating large AI-based data-centers, reflecting a high level of urgency and compromise in response to hardware shortages.
Interestingly, what makes this unusual is that it is often the RAM and GPU that are needed for AI, not CPUs, so to see CPUs suffering is unusual. However, CPUs remain essential for orchestrating and managing the complex software stacks, drivers, front-end services, and data pipelines required by large-scale AI deployments.
Even if the core AI inference runs on GPUs, CPUs are needed to pre- and post-process data, coordinate workloads, and manage thousands of concurrent agents. As a result, larger multi-core CPUs become increasingly valuable, even if their performance is not top-tier.
The growing scarcity of not just CPUs, but also RAM and GPUs, points to a deeper supply crunch across the entire hardware ecosystem. If component shortages persist, it will become even more challenging to build and operate large AI systems, potentially slowing innovation and deployment.
This situation may prompt new entrants or countries with advanced semiconductor capabilities, such as China, to accelerate their efforts to develop domestic alternatives and fill gaps in the global supply chain. China's ongoing investment in its own processors, 3D memory technologies, and RISC-V architectures positions it as a key player in the evolving landscape.
You may also like
