Qualcomm lands its first major AI-chip customer and moves into data centers

Qualcomm said it will begin shipping a custom AI chip to a major cloud infrastructure operator as early as 2026. For a company whose name has been associated mainly with smartphones for decades, this is the clearest signal yet that its push into data centers is no longer just a slide deck about the future.

A new customer

The news came with Qualcomm's quarterly report on April 29, 2026. The company said its project with a "leading hyperscaler" is on schedule, with first shipments expected later this calendar year. The customer's name has not been disclosed, but in the industry the term usually refers to the largest cloud players, which build and scale giant AI clusters. On the analyst call, Qualcomm CEO Cristiano Amon added that the company sees this not as a one-off delivery but as a project spanning several product generations.

Even with a less-than-ideal forecast for the next quarter, the market responded to this signal. Qualcomm reported revenue of $10.6 billion and non-GAAP earnings per share of $2.65, and its shares rose about 15% in over-the-counter trading after the results. The investors' logic is clear: the smartphone market remains cyclical and jittery, while contracts with large cloud customers are now seen as a ticket into the fastest-growing segment of the semiconductor business.

What Qualcomm is betting on

Qualcomm is entering data centers not with one bet but with several types of products at once. On the earnings call, the company said explicitly that it is working on several fronts:

• server CPUs for general-purpose computing and AI orchestration
• inference accelerators, meaning hardware for running already trained models
• custom ASIC chips for the specific tasks of individual cloud customers
• the AI200 and AI250 line, announced in October 2025 as a separate roadmap for AI inference

This matters because Qualcomm is not trying to sell one universal chip to everyone. Amon described the company's approach as a mix of standard solutions and custom projects: in some cases Qualcomm will offer ready-made products, in others it will build a system around a specific customer's needs. After buying Alphawave, the company also gained more IP and expertise in high-speed connectivity, which is especially important for modern AI systems, where the value lies not only in the chip itself but in how it connects to memory, the network and the entire rack.

Why now

Qualcomm has a simple motive for the pivot: dependence on smartphones has become too risky. Last quarter, management again spoke of weak demand and said that rising memory prices had hit the Android device and PC markets. At the same time, Apple and Samsung are expanding their own chip programs, so Qualcomm's traditional mobile business no longer looks as secure as it once did. Against this backdrop, data centers look less like a new line of business and more like necessary diversification.

The timing is good. Hyperscalers continue to pour tens of billions of dollars into AI infrastructure, and the emphasis is gradually shifting from training models to running them at scale. This is where Qualcomm is playing its old trump card: energy efficiency. As early as fall 2025, the company showed AI200 and AI250 as inference solutions focused on cost of ownership, memory and performance per watt. Now the story has what any attractive roadmap lacks: a real customer and first shipments on the horizon. Qualcomm promised more details at its investor day on June 24, 2026.

What it means

For Qualcomm, this is not yet a victory over Nvidia or a sharp turn in its financial model, but a first commercial test in the most expensive chip market. But if shipments really start in 2026 and grow into a multi-year contract, the company will gain not just new revenue but a claim to be considered a full-fledged player in AI infrastructure, not only a supplier of hardware for smartphones.