Qualcomm Eyes AI Data Centers as Smartphone Sales Slow

Qualcomm wants a bigger role in the computers that run AI services, just as the smartphone market that built its business gets harder to count on. On its fourth-quarter 2026 earnings call, the company outlined plans to enter the AI data-center business; smartphone sales are reportedly running about 20% below 2025 levels. Earlier partnerships involving Amazon and Meta give Qualcomm relationships with two of the companies shaping AI infrastructure.

The immediate question is what those relationships can become. A partnership might lead to engineering work, a product integration or purchases at scale. Those outcomes carry very different consequences for a chipmaker's revenue. Qualcomm's data-center ambitions have yet to come with identified chips, deployment volumes, production dates or revenue targets in the account of its earnings call. For now, investors and potential customers have an ambition to evaluate, rather than a rollout to measure.

Why the Phone Market Adds Pressure

Qualcomm sells into a world where phone makers plan products far ahead of launch, choose components within tight cost and power budgets, and depend on consumers replacing devices. When demand softens, even a company with strong technology faces fewer opportunities to turn a design win into growing sales. That helps explain the attraction of another market, although a phone-sales estimate should not be mistaken for a like-for-like measure of Qualcomm's own revenue.

The pressure extends beyond demand. A weak earnings forecast and memory shortage weighing on smartphone sales helped send Qualcomm shares down 5% after its results. Memory is a useful reminder that chip companies operate inside a supply chain, not a solo speedrun. A phone can become more expensive or harder to build because of a component outside its main processor, affecting the market Qualcomm serves even if its own design holds up.

Moving toward data centers offers a different potential source of growth, but it also changes the customer conversation. Phone makers buy chips for devices constrained by battery life and size. Data-center operators buy computing capacity that must fit within power, cooling and operating budgets across entire facilities. Both care about efficiency. They calculate it over very different systems.

Qualcomm's strength in designing for power-constrained devices gives the move a coherent rationale. If its engineers can carry useful lessons from mobile computing into AI infrastructure, operators have a reason to listen. That is a case for investigating the technology, though, rather than assuming a phone chip's efficiency will automatically translate into a server purchase.

What an AI Data-Center Customer Buys

AI infrastructure can mean several jobs. Training builds or updates a model from large amounts of data. Inference runs a trained model to answer a request, such as a chatbot prompt or an image search. A chip that performs well at one job need not be the best choice for the other. Buyers also weigh how easily they can move existing applications onto new hardware, how reliably systems operate and what the full installation costs to run.

Software belongs in that calculation. A data-center buyer needs tools that let developers prepare models, manage workloads and diagnose problems. If adopting a processor requires extensive rewrites, a promising price or power figure may lose its appeal. The processor, software and surrounding equipment have to work together, rather like a game console whose hardware looks great until you discover your library won't run on it.

This is where an established supplier's advantage becomes concrete. Operators have already invested time and money in systems they know how to deploy. A challenger must offer enough benefit to justify testing, integration and the possibility of disruption. Qualcomm could compete on efficiency, cost or a workload-specific feature, depending on what it brings to market. Buyers will want results from systems they can actually use, not an abstract claim that mobile expertise belongs in the data center.

Amazon and Meta make the picture more interesting because large technology companies can influence what AI hardware gets built. A close working relationship can help a chip designer understand a customer's workload early enough to shape the product around it. Yet partnership announcements alone leave open who is buying what, and for which job. The path from joint work to recurring orders is where this story will become measurable.

The Forecast Versus the Contract

Big numbers can make that path look shorter than it is. A Seeking Alpha contributor has projected data-center revenue of $15 billion in three years. That is an outside forecast, not a Qualcomm revenue target. To reach any large sales figure, a chipmaker needs products that customers qualify, manufacture and deploy in sufficient quantities, followed by repeat purchases. Each step takes work, and each can change the eventual revenue.

The strongest case for Qualcomm is straightforward: AI services need computing capacity, major operators have incentives to examine alternatives, and Qualcomm knows how to design processors under demanding power constraints. More competition could give buyers additional choices about price, performance and supply. Qualcomm also has a reason to look beyond phones while the smartphone market is under strain.

The hard part is that data-center customers buy a whole working system, not a processor's backstory. A chip must perform the intended workload well enough to justify switching or expanding a deployment. Its software needs to fit the customer's development process. Manufacturing and delivery must keep pace if a trial becomes a substantial order. None of those requirements disappears because AI demand is high; high demand can make a buyer's tolerance for interruptions even lower.

These pressures also shape how to read Qualcomm's next announcements. A named product would clarify which AI job it is built to handle. A production schedule would show when customers could put it to work. A customer commitment with a stated deployment would reveal more than the word “partnership.” Eventually, separately identifiable data-center revenue would show whether those deployments have become a business.

Qualcomm has a plausible route into the discussion because efficiency and customer relationships matter in AI infrastructure. Its next challenge is to turn those advantages into hardware that earns a place in a data center, and then into orders that show up on the balance sheet.