A "Coming Out Party" for a Reinvented Company
The most recent Computex show in Taiwan functioned as a coming-out party for a reinvented Intel. The company used the event to launch new products — most notably its Xeon 6 Plus processor — and to signal to the market that it is ready to get back into building systems, not just selling individual chips. A central theme was Intel's close collaboration with AI leaders on what the agentic servers of the future should look like. It began to bring that vision into focus by pairing the new CPU with blueprints for new server racks specifically designed to put those servers to work creating and running AI agents.
A Full-Speed AI Build-Out
The broader environment is decisively risk-on for technology, because the AI build-out is proceeding at full speed. GPUs, XPUs, and CPUs are all in high demand simultaneously, with memory acting as the key driver that creates bottlenecks across every one of those processor types. A clear takeaway from Computex is that AI is the future not only of the data center but also of personal computing. New products are being launched specifically to adapt to the needs of the AI community and to make every category of computer better suited to AI workloads.
What It Means to Call Intel "an AI Lab" — and to Say It "Saves the Hyperscalers"
Describing Intel as an "AI lab" captures a major break from how the company historically engaged with the tech ecosystem. In the past, Intel offered CPUs that achieved great efficiency and could be deployed across a multitude of form factors, whether in the cloud or on premises. What distinguishes the new Intel, under the leadership of its CEO, is that its data center and AI group now works closely with the AI community itself — learning directly from startups and AI model developers, and co-designing the software and servers those customers actually need.
The claim that Intel "saves the hyperscalers" rests on a second pillar: Intel Foundry. On the hyperscaler side, Intel works with these customers on CPUs, but it can also collaborate with them through its foundry business. Intel Foundry is progressing rapidly in terms of the yield it can deliver for its newest CPUs, and it presents a compelling value proposition in advanced packaging. That packaging advantage is the specific element that can save hyperscalers money when they build their own custom chips. By utilizing more of the wafer and providing denser interconnects on the chip, customers can save a significant portion of their costs — up to 22% on the total package cost, according to Future Market Insights.
These amount to two distinct value propositions Intel presents to the AI community: model co-design on one side and processor co-design on the other. Together they allow the company to work across both silicon and systems, positioning it as a partner to the AI build-out as a whole rather than a supplier of one component.
Geopolitics, Inference, and the Demand for Control
The AI build-out is global. Other countries are increasing their spending, and the home government has been working closely with Intel. There is a genuine competitive battle underway in the world of AI, but the guiding principle is that everything that can be built for AI will be deployed wherever it is needed.
The key use case Intel serves within this landscape is the ability to perform inference across a variety of models — both closed-source and open-source — and to deploy CPUs in the places enterprises are accustomed to having them: on premises, or in a custom-built data center focused specifically on the inference that actually runs AI and improves the economics for users. This is the direction in which AI is heading, and it was a major focus of the Computex show. The industry is running into high costs and into models that are difficult to trust. Enterprise customers want control over their models; they want to be able to run them only on the specific tasks they need, rather than on every possible task. To achieve that, they are increasingly working directly with Intel.
A briefing reinforced how seriously Intel is taking this shift: the company has profiled over 300 customer agentic use cases, studying how customers actually use agents and what they use GPUs versus CPUs for. That research surfaced directly in Intel's keynote, where the company demonstrated that for a given Python script, you use two CPUs for every GPU to get the job done. There will still be a need for large GPU clusters, but Intel is well positioned to maintain its leading market share in server CPUs.
Will Intel's CPU Market Share Keep Growing?
A natural question is whether Intel can expand its CPU market share, especially given a period when competitors such as AMD were eating into Intel's position. On the investment side, the numbers suggest the story is still underappreciated: 16% of funds hold Intel, up 3% month over month — meaning interest is rising but the stock is far from being held as universally as one of the Mag 7 names. Even so, Intel can reasonably be described as the pillar of the AI build-out, capturing more and more market share in both CPUs and AI sequencing logic packaging.
The answer to the market-share question is that the CPU pie itself is growing very rapidly. A striking feature of Computex was that everyone focused on the CPU opportunity — not only Intel, already a leader in that space, but also Nvidia, which shared the view that CPUs are the future and that more of them will be needed for every GPU. In an expanding pie of that kind, there is room for every participant to grow significantly.
This points to a clear near-term opportunity: data center and AI could become Intel's largest business unit, surpassing both client computing and foundry. Beyond that, the expanding ratio of CPUs to GPUs has the potential to multiply what has been a sub-$100 billion market into one worth hundreds of billions, depending on exactly how the AI community puts these CPUs to use. A great deal remains in flux, but Intel holds a structural advantage: it can expand its own supply because it owns that supply — a benefit that flows directly to its customers.
The Catalysts Ahead
Taken together, the forces that can continue to pull Intel forward span foundry progress, the surging demand for inference, fiscal and enterprise-controlled AI, hybrid cloud deployments, and the emerging category of the AI PC. With shifting industry dynamics — including model developers recalibrating their own product roadmaps — and with intensifying global and geopolitical spending on AI, Intel's ability to operate across silicon and systems leaves it positioned to benefit from nearly every layer of the AI build-out at once.
