The CPU Comeback Is Upon Us

Amazon Web Services (AWS) has instructed its engineers to conserve CPU cycles at all costs as AI workloads, particularly agentic AI systems, strain its cloud infrastructure and cause wait times for CPU server capacity to explode. According to Moor Insights & Strategy vice president Matt Kimball, 2026 has brought a spike in CPU demand largely due to agentic AI, where agents spawn sub-agents and make API calls, tasks that are inherently CPU-based. Intel senior staff research scientist Souvik Kundu co-authored a paper with Georgia Institute of Technology researchers proposing scheduling optimizations that can cut end-to-end latency by up to 1.8x under sustained load.

Earlier this year, leaders at Amazon Web Services https://aws.amazon.com/free/?trk=dc9b9d60-cc82-4cd5-8a61-0b33d6a79fab&sc channel=ps&ef id=CjwKCAjw1vXTBhB-EiwAEKr k 57Xoz6QKQSrRqDF4mGRYOudA A99MTHZL6alVjPDxdliBQfEMPihoC0RUQAvD BwE&gads camp=23532472510&gads ag=199502799824&gads ad=795877020713&gads kw=amazon%20web%20services&gads matchtype=e&gads network=g&gads device=c&gads geo=9198314&gad campaignid=23532472510&gbraid=0AAAAADjHtp-SgwVQvE7H9V8lK49jVtMDw&gclid=CjwKCAjw1vXTBhB-EiwAEKr k 57Xoz6QKQSrRqDF4mGRYOudA A99MTHZL6alVjPDxdliBQfEMPihoC0RUQAvD BwE delivered a new mandate to their engineers: they need to conserve CPU cycles at all costs. AWS has reportedly https://www.theinformation.com/articles/aws-tells-engineers-cut-cpu-waste-amid-crunch experienced an explosion in wait times for CPU server capacity as AI workloads strain the company’s cloud infrastructure. The issue seemingly took AWS off-guard, and for good reason. The AI boom led to a surge in demand for GPUs https://spectrum.ieee.org/nvidia-gpu and, later, memory https://spectrum.ieee.org/high-bandwidth-memory-shortage . CPUs were mostly left out of the story, as their relative lack of parallelization made them a poor fit for AI model inference, the process of running and serving large language models LLM to users. But the rise of agentic AI systems, which allow AI models to operate autonomously and call on sub-agents, is changing the narrative. Matt Kimball https://moorinsightsstrategy.com/team/matt-kimball/ , vice president and principal datacenter analyst at Moor Insights & Strategy https://moorinsightsstrategy.com/ , says 2026 has brought a spike in CPU demand, much of it due to agentic AI https://spectrum.ieee.org/ai-agents . “It’s one thing to have this agentic workload, and let’s say, it spawns 100 agents. If I’m going to roll this out across my enterprise, those 100 become tens of thousands, hundreds of thousands, or millions of agents,” says Kimball. “You have agents spawning sub-agents, making application programming interface API calls, and talking to more agents through Anthropic’s model context protocol.” Kimball’s comments refer in part to “tool use,” which is shorthand for an LLM’s ability to access the internet, open files on a desktop, and generally use a variety of software to accomplish its task. LLMs trained for tool use learn how to call on other software. While the LLM’s inference is still primarily executed on a GPU or similar AI accelerator, the tool calls the LLM makes are typically pushed to the CPU. “Many components of an agentic AI task are inherently CPU based jobs,” explains Souvik Kundu https://www.linkedin.com/in/souvik-kundu-64922b50/ , senior staff research scientist at Intel https://www.intel.com/content/www/us/en/homepage.html . “The CPU does the job of parsing output, figuring out which tool to invoke, making the API call or running the code, collecting the result, and feeding it back.” Madhu Rangarajan https://www.linkedin.com/in/mrangarajan/ , vice president of compute and enterprise AI at AMD https://www.amd.com/en.html , makes a similar claim, saying “in our testing, seven of the eight stages in realistic agentic AI pipelines run entirely on the CPU.” An LLM tasked with programming software, for example, will likely make tool calls to write code to files, move or replace files, download required packages, and build the software once the LLM believes it’s complete. Kundu co-authored a paper https://arxiv.org/pdf/2511.00739 on agentic AI optimization alongside researchers from the Georgia Institute of Technology. They found the CPU is often idle while LLM inference is executed on a GPU and that, conversely, the GPU is often idle when tool calls are executed on the CPU. To optimize this, Kundu and his colleagues propose scheduling optimizations that can cut end-to-end latency the time between the start and finish of the agentic workload by up to 1.8x under sustained load. It’s a start, but the gains chase a moving target. Agentic systems generate work at machine speed and multiply it as they go. OpenAI’s inadvertent hack https://spectrum.ieee.org/hugging-face-openai-cyberattack?itm source=homepage&itm medium=hero&itm campaign=hero-2026-08-10&itm content=hero6 of Hugging Face saw its model fire off as many as 300 actions an hour, and a single agent can spawn sub-agents that make tool calls of their own. And there’s one more important complication that may increase the workload on a CPU as models become more complex: safety guardrails. Safety and policy checks on an agent’s actions are often specific rules that inspect syntax and log files, Kundu says. Guardrails may also use small models under a billion parameters to analyze task complexity or intent. Though they could be executed on a GPU, they often aren’t, because their small size and the need to minimize latency keeps the work on the CPU. Increasing the number of CPUs available significantly decreases the latency for Llama-8B responses over longer sequence lengths.Source: Euijun Chung, Yuxiao Jia, et al. https://arxiv.org/pdf/2603.22774 Euijun Chung https://ejchung0406.github.io/ , a PhD student at the Georgia Institute of Technology, recently co-authored another paper https://arxiv.org/abs/2603.22774 , with findings that complement Kundu’s work. Chung and his co-authors found that when a server has too few CPU cores, it falls behind on dispatching work to the GPUs. That causes the GPUs to stall as they wait for instructions. In addition to that, the paper touches on another key element of LLM workloads: tokenization https://seantrott.substack.com/p/tokenization-in-large-language-models . Tokenization is a key first step in LLM inference. It converts text into integer token IDs that can be processed by the model. Unlike the matrix math required for most LLM inference, tokenization is branchy, data-dependent sequential string manipulation. Though it can be parallelized by chunking text, it’s not massively parallel in the same way as the bulk of LLM inference is. Tokenization of small prompts is a relatively trivial task and won’t tax even an entry-level CPU. However, an agentic model that makes tool calls must parse and tokenize the results of the call. “If you have an ongoing sequence of, say, 100,000 tokens, and you have a tool result of a 1,000 tokens, the tokenizer will have to tokenize the whole sequence again. And you have to do tokenization at every agentic tool call,” says Chung. This both increases the frequency of tokenization and increases the number of tokens involved. It’s conceivable that future tokenizers will find ways to mitigate this, Chung says, but it remains a problem for modern LLM inference. The paper finds that time-to-first-token latency the time required for the model to produce the first word of its reply can increase dramatically as the sequence length grows. CPUs with more cores can reduce the problem. In test runs at longer sequence lengths, increasing CPU core counts can reduce time-to-first-token latency by roughly 1.5x to 7x. Chung and his colleagues were only able to test smaller models, such as Alibaba’s Qwen 3-30B and Meta’s Llama 3.1-70B, due to limitations of the hardware available for testing. He speculates that larger models will experience less dramatic bottlenecks due to their higher overall GPU demand, but also expects agentic AI will push token lengths far beyond the longest he and his co-authors tested. “If you think about something like Anthropic’s Claude, you can easily hit 500,000, even a million tokens,” says Chung. “In the world of agentic AI, the average sequence length will grow and grow, so I’m expecting this problem to get worse in future workloads.” Amazon’s crackdown on use of CPU resources is one of several indicators that Kundu and Chung have identified issues with real-world relevance. Intel has sold out https://finance.yahoo.com/news/intel-turnaround-no-one-saw-141000146.html?guccounter=1&guce referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce referrer sig=AQAAAJqbhttNhPiUgZBGw-QWbgpaSpMnF0qwXSRBrSMQvCVWxhrkdpDYWyavDbXWb-9coPNWysaUbJK l5uY bZH6ACvfNpR8JrxKbKXsw8dVSlzZLYe9ydhDqS0SAyZbc4KSK0bqLw8WJSIYt-yoFGlGWDRP1UIaWov2q1bpzbcLPE- of server CPUs through at least the end of the year. AMD has doubled https://wccftech.com/amd-doubles-server-cpu-forecast-to-120-billion-as-agentic-ai-rewrites-demand-ceo-says-epyc-verano-built-purely-for-ai/ its server CPU forecast. Arm https://www.arm.com/products/cloud-datacenter/arm-agi-cpu and Qualcomm https://www.cnbc.com/2026/06/24/qualcomm-data-center-cpu-meta.html have both announced new CPUs designed to accelerate agentic AI. Even Nvidia https://www.nvidia.com/en-us/ has prioritized Vera https://nvidianews.nvidia.com/news/nvidia-unveils-vera-the-cpu-for-agents , its Arm-based CPU for agentic AI, which is part of Nvidia’s Vera Rubin https://spectrum.ieee.org/nvidia-rubin-networking platform. Kimball says these developments make it clear that the AI industry is placing more emphasis on CPU performance. He sees the surge in demand as an “absolute tell” that CPUs are now considered a key part of an agentic AI system. Unfortunately, this may translate to broader CPU shortages and increased prices, much as has already occurred with GPUs and memory. “You’re already seeing a CPU crunch to some degree. When you look at the constraints in the market, it even trickles down into the consumer space,” says Kimball. He adds that Intel has cut production https://www.techpowerup.com/345535/intel-reallocates-pc-production-capacity-to-server-cpus-amid-tight-wafer-supply of client CPUs in favor of server CPUs even as Intel’s new 18A production process has grown the company’s sales in the client segment. Kimball sees that as a sign that CPU makers will follow the money.