Qualcomm’s $15 Billion Data Center Bet Faces Its Test

By Priya Raman

Qualcomm targets more than $15 billion in annual data-center AI chip revenue by fiscal 2029, up from essentially zero now.

What must Qualcomm show on November 4?

Qualcomm reports fiscal fourth-quarter and full-year 2026 results after the close on November 4. Investors will test whether expansion beyond phones can support the revenue targets Qualcomm set in June, and whether commercial commitments can support the schedule. The data-center business remains at the beginning of its planned ramp.

Guidance sets three revenue milestones:

• Fiscal 2026: about $300 million
• Fiscal 2027: $5 billion
• Fiscal 2029: more than $15 billion

The HBC product should contribute in the second half of fiscal 2027. Dragonfly C1000 CPUs should begin contributing in the second half of fiscal 2028. The roadmap spans custom silicon (customer-designed chips), AI accelerators (specialized machine-learning processors), Dragonfly CPUs (central processing units) and connectivity products through 2028. The AI target sits inside a plan to double non-handset revenue to $40 billion. Most forecast value remains prospective. Qualcomm must convert commercial commitments into shipments and recognized revenue.

How large is Nvidia’s lead?

Nvidia reported $89 billion in quarterly data-center revenue for the period ended July 26. Qualcomm provides annual forecasts for an early-stage segment. The periods and reporting scopes differ, so the figures indicate the scale of Nvidia’s position rather than a like-for-like quarterly market-share calculation.

How does the hardware target inference?

Qualcomm unveiled AI200 and AI250 on October 27, 2025. AI200 is due in 2026; AI250 is slated for 2027. The associated racks are built around 768GB of memory per card. Qualcomm says that exceeds memory on comparable Nvidia and AMD hardware.

Qualcomm is targeting inference, the production use of already-trained models to generate text, power chatbots and serve real-time applications inside commercial data-center systems. It is not pursuing training workloads that the company credits for making Nvidia’s H100 and Blackwell indispensable. The thesis assumes inference will grow faster as companies move from model development to deployment.

This narrower bet avoids the training market Nvidia defined with H100 and Blackwell. It does not remove Nvidia’s software advantage or developer lock-in. Qualcomm must overcome an ecosystem that has kept developers on Nvidia hardware for nearly two decades.

Which customers have committed?

Saudi state-backed AI venture HUMAIN plans to deploy 200 megawatts of Qualcomm AI systems beginning in 2026. Analysts estimated the deployment could generate roughly $2 billion for Qualcomm. Neither company disclosed a contract value, so the estimate cannot be checked against disclosed deal economics.

Meta agreed in June to deploy Dragonfly C1000 CPUs in future server generations. Deployment starts in the second half of 2028. Neither side disclosed volume or pricing.

In September, Qualcomm announced a multi-generational collaboration with Amazon covering custom AI inference silicon and connectivity. Qualcomm granted warrants (rights to buy shares at a set price) to an Amazon affiliate for up to 25 million shares at $161.26 each under the agreement. Vesting depends on commercial agreements, binding orders and purchases that can reach $60 billion over the warrant term.

The companies also are building 1.6-terabit-per-second optical connectivity for data-center networking. Qualcomm shares rose more than 4% after the announcement. The $60 billion ceiling concerns commercial orders and purchases associated with the arrangement; it is not a disclosed Qualcomm revenue commitment.

Why did Qualcomm buy Modular?

Qualcomm entered advanced talks to acquire AI infrastructure startup Modular Inc. in June, according to Bloomberg. The June 22 report put the valuation at $4 billion, two days before Qualcomm’s investor day. Modular was an inference and compiler (code-translation software) company co-founded by Chris Lattner.

Qualcomm then closed the acquisition in July. It valued the completed transaction at approximately $3.1 billion based on its share price at closing, below the roughly $3.9 billion value announced in June.

Modular’s Mojo programming language and MAX inference engine can run AI workloads across Nvidia, AMD, Intel and Qualcomm hardware without code rewrites for each platform. Lattner, who created Swift at Apple and co-created LLVM, now runs advanced AI software and platforms at Qualcomm.

The acquisition targets CUDA (Nvidia’s proprietary development platform), a central barrier to hardware displacement. Qualcomm still needs developer adoption and production proof. The announcement does not quantify feature parity, migration effort or performance at comparable scale.

Why does Alphabet matter before the call?

Alphabet reported mixed third-quarter results on October 28, one week before Qualcomm’s scheduled release. Revenue rose 25% to $127.58 billion but missed analyst expectations. An $8.7 billion loss on equity investments reduced earnings.

Google Cloud revenue climbed 103% to $30.7 billion, while total revenue backlog reached $567.7 billion. The stock fell sharply as investors weighed the earnings miss, slower-than-expected advertising growth and the cost of the AI buildout.

Alphabet maintained 2026 capital-expenditure guidance of $195 billion to $205 billion. It said spending would increase significantly in 2027. That backdrop raises scrutiny of AI investment returns just as Qualcomm asks investors to underwrite a three-year revenue ramp.

What remains unproven?

Qualcomm does not need to beat Nvidia to validate its target. It needs credible evidence that $15 billion arrives on schedule. Analysts are likely to press management on customer timing, product contribution and remaining forecast gaps.

The stock still trades mainly on smartphone royalties and Snapdragon shipments. A credible data-center ramp could broaden that earnings base. Commercial announcements alone will not settle the question. Execution against the fiscal 2027 and 2028 milestones will determine whether Qualcomm’s product roadmap becomes recognized revenue on the promised schedule.