Qualcomm has long been the leader at the "edge" — the world of devices, including smartphones, cars, industrial systems, and other connected products. The company's defining advantage in those domains has been the ability to perform processing at very low power. Now, that same low-power competency is being deployed in an entirely new arena: the data center. As data centers evolve, low-power processing is becoming critically important, and Qualcomm is positioning the engineering advantage it spent years building in edge devices as a differentiator for large-scale computing infrastructure.
The reception from customers has reportedly been very positive. A key driver is that hyperscalers — the operators of the largest cloud and AI data centers — are actively looking for alternatives. The market is shifting from a model where GPUs were used for essentially everything toward a "disaggregated computing" approach. As the industry transitions from AI training to AI inference, this disaggregation opens up far more room for new players to offer solutions that are highly optimized for specific, low-power workloads, rather than relying on a single general-purpose GPU for all tasks. Qualcomm views its data center announcements as just the beginning of its presence in this space.
A Doubled Revenue Target
One of the most striking elements of Qualcomm's outlook is a dramatic upward revision to its financial targets. The company previously set a 2029 revenue target for its non-handset business at $22 billion. That figure has now been revised upward to $40 billion — nearly a 2x increase in a revenue target just two years out. This effectively doubles the projection and drew strong attention from Wall Street.
The data center business is a major component of this new target. Qualcomm now anticipates $15 billion of data center revenue in 2029, with roughly $5 billion of that arriving the following year (i.e., in 2027–2028 ramp terms, $5 billion of revenue is expected next year as the business builds). The opportunity is enormous: the data center is described as a trillion-dollar market. Against that backdrop, a Qualcomm target of approximately 5% market share is characterized as very reasonable and achievable, reinforcing the company's confidence in its forecast.
The path begins in 2027 with custom chip engagements involving two global hyperscalers. From there, Qualcomm intends to broaden its reach — engaging additional hyperscalers and building out a wide product line across the data center segment.
Meta, Microsoft, and Long-Standing Customer Relationships
Among the early customers, Meta and Microsoft have been identified, and there has been speculation about other names such as Amazon and ByteDance. Importantly, these are not new relationships. Qualcomm has worked with these companies for a very long time across edge devices:
- Meta uses a Qualcomm chip in its Ray-Ban Meta smart glasses.
- Microsoft uses a Qualcomm chip in its Surface PCs.
- Google and Amazon likewise rely on Qualcomm chips in their edge devices.
Because Qualcomm is already deeply familiar with these companies, the conversation has naturally evolved toward how Qualcomm's technology can become relevant to their data center needs. The current moment represents many threads coming together — years of groundwork now aligning with customer needs and Qualcomm's capabilities, which is the root of the company's significant strategic change.
Dragonfly: A New Brand for the Data Center
To differentiate its data center products from its edge-centric Snapdragon brand, Qualcomm is introducing Dragonfly, positioned as the global brand for everything related to the data center. Under this umbrella, Qualcomm is delivering four families of products, all of which fall under the "dragon wing":
1. Custom chips built specifically for individual customers.
2. CPUs — Qualcomm believes it is building the fastest CPU, claiming it will be 2x better than anything currently on the market.
3. AI accelerators built around a unique technology referred to as HBM (high bandwidth compute), designed to deliver very high bandwidth at very low power.
4. Connectivity chips.
The Inference Opportunity: Data Centers as Token Factories
The most exciting part of this strategy centers on AI inference and "decode acceleration" across both merchant and custom silicon. The framing here is that data centers are becoming token factories — the entire objective is delivering tokens (the units of AI model output) at the lowest cost and the lowest power.
Qualcomm's AI accelerator is engineered to deliver tokens at the lowest power, which translates directly into a total cost of ownership (TCO) advantage for hyperscalers. This advantage is amplified by one of the central constraints in modern data center construction: the availability of power. There simply is not enough power for everything that is needed. As a result, when hyperscalers select chips and solutions, picking extremely low-power options becomes essential, because it allows them to minimize the power required when building new data centers. Qualcomm's pitch is precisely this — a token factory operating at the lowest possible power — and that is why it has generated so much interest.
The Modular Acquisition and an Open Software Stack
Qualcomm announced an acquisition of Modular (announced June 24th), which brings an open software stack into the company's portfolio. Both data center customers and edge device makers are seeking an AI-native software stack — one that allows AI models and use cases to be ported very quickly and then used seamlessly across both the cloud and edge devices.
Modular addresses this need as an industry-standard, open-source platform. Critically, it is not locked to Qualcomm hardware: it can be deployed not only on Qualcomm chips but also on competitors' chips. This matches the requirement of the moment — a solution that works across the disaggregated serving environment emerging in data centers as well as across the wide variety of edge devices. Qualcomm intends to keep Modular as a separate entity that continues to support all chips in the industry, remaining an open-source platform while also being optimized for everything Qualcomm delivers.
Beyond the Data Center: Physical AI, Robotics, and Personal AI
While automotive has always been a bread-and-butter business for Qualcomm, the company is expanding into many additional areas, including industrial networking, robotics, the Internet of Things, and personal AI and compute.
Physical AI is a major area of excitement, and it begins with automotive, where Qualcomm is the leader. The company expects to become the largest supplier to the automotive industry from a silicon perspective next year. That automotive foundation extends naturally into industrial use cases and then into robotics. Robotics is identified as potentially the next trillion-dollar market, and Qualcomm believes it possesses all the right chips and software technologies to converge on this opportunity.
The second area highlighted is personal AI — a new category encompassing smart glasses, smart watches, and other wearable devices. These devices can see what the wearer sees and hear what the wearer hears, enabling agentic AI conversations. The expectation is that, globally, everyone will eventually carry one or two such devices in addition to their phone and PC. Qualcomm positions itself as by far the platform of choice worldwide for companies choosing which chip to use when building these devices. Meta is one example, but the trend spans Google, Snap, and Chinese and Indian customers alike — a global movement in which Qualcomm sees itself at the forefront.
Supply Chain Scale as a Competitive Moat
Addressing concerns about demand-and-supply dynamics during the AI buildout, the message is that demand is extremely strong. Qualcomm frames itself as a very large supply-chain-scale company, shipping about 40 billion components a year. It consumes a large volume of wafers on leading process nodes across different foundries and maintains strong relationships with various suppliers.
This scale has become a significant competitive factor. When customers select a supplier today, they want a partner they can rely on to meet their needs. Qualcomm's strong track record, combined with its considerable supply chain scale, becomes an important consideration as customers make their decisions — and a meaningful differentiator in a market constrained by both power and component supply.
Summary of the Strategic Shift
Taken together, the picture is one of a company leveraging decades of low-power, edge-device leadership to enter — and aim for roughly 5% of — a trillion-dollar data center market. The strategy rests on several pillars: the new Dragonfly brand spanning custom chips, fast CPUs, low-power AI accelerators, and connectivity chips; a software approach anchored by the open, cross-platform Modular acquisition; deep, pre-existing relationships with hyperscalers like Meta and Microsoft; and a supply-chain scale that reassures customers. Combined with parallel growth in physical AI, robotics, and personal AI devices, this is what underpins Qualcomm's confidence in doubling its non-handset revenue target to $40 billion by 2029.
