Qualcomm aims to return to the data center space, and this time it doesn’t seem to be a minor experiment. The company has confirmed that it is working with a major hyperscale client on custom silicon development, with initial shipments expected by the end of this year. The client’s name hasn’t been disclosed, but Cristiano Amon, Qualcomm’s CEO, made it clear: they are talking about a “multigenerational” collaboration and promised more details at their Investor Day on June 24.
The announcement comes at a delicate yet exciting time. Delicate because Qualcomm still heavily relies on the mobile business, which is affected by memory cost pressures and uneven demand in China. Exciting because large cloud operators are ramping up their investments in proprietary chips, AI accelerators, custom CPUs, and high-speed connectivity to reduce dependence, improve energy efficiency, and tailor hardware to their actual workloads.
A Quarter with Bright and Dark Spots
Qualcomm closed its second fiscal quarter of 2026, ending March 29, with revenue of $10.599 billion, down 3% year-over-year. GAAP net income amounted to $7.370 billion, although this figure was significantly influenced by a $5.700 billion tax benefit resulting from the release of a previous tax valuation. On a non-GAAP basis, net income was $2.840 billion, with non-GAAP earnings per share of $2.65.
The QCT division, which includes the semiconductor business, reported $9.076 billion in revenue, a 4% decrease year-over-year. Within that segment, mobile devices remained the dominant part, bringing in $6.024 billion, but this was down 13%. Conversely, automotive grew by 38%, reaching $1.326 billion, and IoT business increased by 9%, totaling $1.726 billion. Qualcomm emphasized that the combined automotive and IoT segments grew by 20% year-over-year—a sign of diversification the company has been working to strengthen for years.
| Q2 FY2026 Metrics | Data |
|---|---|
| Total Revenue | $10.599 billion |
| QCT Revenue | $9.076 billion |
| QTL Revenue | $1.382 billion |
| Handsets | $6.024 billion |
| Automotive | $1.326 billion |
| IoT | $1.726 billion |
| Diluted GAAP EPS | $6.88 |
| Non-GAAP Diluted EPS | $2.65 |
| Q3 FY2026 Revenue Guidance | $9.2–$10 billion |
The guidance for the third fiscal quarter also shows caution. Qualcomm expects revenue between $9.2 billion and $10 billion, with QCT estimated at $7.9–$8.5 billion and QTL between $1.15 billion and $1.35 billion. The company acknowledges that supply chain constraints and rising memory costs are impacting demand from several phone manufacturers, but expects QCT handset revenue from Chinese clients to stabilize in Q3 and then increase sequentially afterward.
Data Center Returns to the Radar
The most significant aspect of the quarter isn’t just the numbers. It’s the statement regarding the hyperscale client. Qualcomm assures that its first custom silicon collaboration for data centers is on track to initiate shipments before the end of the year. During the analyst call, Amon avoided naming the client but spoke of a long-term commitment and a strategic opportunity aligned with industry transformation driven by AI.
This isn’t an isolated move. Qualcomm had already indicated in 2025 its intention to re-enter the data center market, from which it retreated in 2018 after abandoning its focus on ARM servers to concentrate on mobile. The market has since evolved. The center of gravity isn’t solely in general-purpose CPUs anymore but in AI inference, chip-to-chip connectivity, efficient acceleration, and semi-custom designs for large cloud customers.
In May 2025, Qualcomm signed a memorandum of understanding with HUMAIN, a Saudi AI company, to develop next-generation data centers, hybrid cloud-to-edge services, and CPU and AI solutions for cloud infrastructure. The collaboration aimed to build AI data centers in Saudi Arabia using Qualcomm solutions for hybrid inference across cloud and edge.
One month later, the company announced the acquisition of Alphawave Semi for $2.4 billion, a move designed to strengthen its position in high-speed connectivity, IP, chiplets, and custom silicon for data centers. Qualcomm completed the acquisition in December 2025 and highlighted that Alphawave complements its Oryon and Hexagon processors for computing and interconnect solutions for AI.
The talent pool is also shifting in this direction. Qualcomm has brought in profiles from Intel to bolster operations, supply chain, and data center CPUs. In a market where architecture decisions are made years in advance, having teams capable of designing, validating, and delivering complex chips for large clients is as critical as owning good IP.
Why Would an Hyperscaler Look to Qualcomm?
The major hyperscalers no longer settle for off-the-shelf hardware. AWS, Google, Microsoft, Meta, and others want chips tailored to their workloads, software, and energy costs. AI has accelerated this trend. Every watt, every millisecond of latency, and each dollar spent on inference across millions of requests becomes a strategic decision.
Qualcomm can attempt to play in this arena with three advantages. First is energy efficiency. Having decades of experience designing low-power chips for mobile, edge, and connected devices, this expertise could be valuable in inference-oriented data centers. Second is integration: combining CPU, NPU, connectivity, and specialized blocks in more optimized designs than generic solutions. Third is customization, now reinforced by Alphawave and the demand for bespoke silicon.
This path won’t be easy. NVIDIA dominates AI infrastructure with GPUs, networking, software, and a mature ecosystem. AMD is gaining ground with Instinct accelerators and EPYC CPUs. Intel aims to defend its role across CPUs, accelerators, and foundry services. Meanwhile, hyperscalers develop their own internal chips and leverage enormous bargaining power. Qualcomm will need to prove not only that it can design high-quality silicon but also that it can produce, support, and sustain it across multiple generations.
Nevertheless, the opportunity is real. If the unidentified client confirms a long-term relationship, Qualcomm could open a new growth avenue less dependent on the smartphone cycle. Mobile will remain a key segment, but the market no longer values volume alone; it also considers who can capture a slice of the massive AI, cloud infrastructure, connected automotive, and edge computing spend.
The company seems poised to tell this story in June. The Investor Day on June 24 will be the moment to clarify whether its return to data centers is a tactical move with a specific client or part of a broader strategy. For Qualcomm, the difference is huge. A one-off design can generate revenue and prestige, but a multi-generational platform for hyperscalers could reshape market perception of a company long seen mainly as the provider of chips for Android smartphones.
For now, the message is cautious yet ambitious: Qualcomm continues to navigate a weak mobile business and a complicated memory environment, but it has an opening in data centers. In the race for AI, that may be enough to turn investor attention back to San Diego—with fresh eyes.
Frequently Asked Questions
What has Qualcomm announced regarding data centers?
Qualcomm has confirmed a custom silicon collaboration with a major hyperscale client, with initial shipments expected by late 2026 and more details to come at its June Investor Day.
Who is the hyperscale client?
The company has not disclosed the name. It could be a large cloud provider or a tech firm with its own infrastructure, but any specific identification would be speculative.
Why does Qualcomm want to return to data centers?
Because AI is increasing demand for custom chips, efficient inference, advanced connectivity, and cloud-to-edge solutions—areas where Qualcomm aims to leverage its experience in low-power design and integrated processing.
What role does Alphawave Semi play?
Alphawave provides high-speed connectivity IP, chiplets, and custom silicon capabilities—assets that are crucial for designing chips and interconnection solutions in AI data centers.