Qualcomm CEO Cristiano Amon. Image processed by AI

Author: Bo Yang, Tencent Technology

Editor: Xu Qingyang

On June 25 local time in the U.S., Qualcomm held its 2026 Investor Day in New York.

Qualcomm announced a complete roadmap for data center AI infrastructure, launching the Dragonfly C1000 CPU, AI300 inference accelerator and High-Bandwidth Computing (HBC) technology, while also unveiling a multi-generation partnership with Meta, deepened collaboration with Hugging Face, and the acquisition of AI software company Modular.

Qualcomm's non-handset business revenue target for fiscal year 2029

Financially, Qualcomm raised its non-handset revenue target for fiscal year 2029 to $40 billion, nearly double the previous long-term target. Among that, data center business revenue is expected to exceed $15 billion in that fiscal year.

In after-hours trading, Qualcomm's stock price rose as much as 16%.

01. Data Center Revenue to Exceed $15 Billion

Qualcomm CFO Akash Palkhiwala forecast during the event that the company's data center business would generate "billions" of dollars in revenue in fiscal year 2027. And by fiscal year 2029, the business's annual revenue will surpass $15 billion.

From the perspective of the company's overall revenue structure, by fiscal year 2029, the non-handset revenue of the QCT (semiconductor) division will reach $40 billion, whereas the long-term target set for this figure in 2024 was $22 billion.

In fiscal year 2029, Qualcomm's handset business will account for only about one-third of QCT revenue.

The remainder will be shared by several growth engines: automotive business revenue of $10 billion, and IoT business revenue exceeding $14 billion. The IoT business includes industrial, networking and robotics ($8 billion), as well as personal AI and computing ($6 billion).

Profit guidance was also raised.

The average analyst estimate for Qualcomm's adjusted earnings per share in fiscal year 2029 is $15.26, while the target set by Qualcomm itself is over $18. This gap directly explains the after-hours jump in the stock price.

When explaining the growth logic, CEO Cristiano Amon focused on the shift in how AI is being used. He believes AI is moving from simple Q&A to agentic applications, meaning models capable of autonomously executing multi-step tasks. These workloads demand more low-power computing, which is precisely the capability Qualcomm has built up through mobile chips.

Amon also stated that AI computing is entering automobiles, everyday electronic devices, and robotics, and chip demand in these areas will continue to "open up."

02. Dragonfly C1000 Debuts, Meta Becomes First Customer

The highlight of the hardware launch was the Dragonfly C1000, a CPU designed by Qualcomm specifically for data centers.

The Dragonfly C1000 is based on a custom-designed Oryon core, adopting a multi-chiplet architecture with over 250 integrated cores running at frequencies above 5GHz. Performance tests provided by Qualcomm show its performance-per-watt is more than double that of existing server CPU competitor benchmarks.

The Dragonfly C1000 supports PCIe Gen 7 and CXL connectivity, uses low-power memory technology for the memory system, and features built-in RAS functions such as ECC, fault isolation, and error recovery. The thermal solution is compatible with both air and liquid cooling, and the rack complies with the OCP ORv3 standard.

Rack configurations equipped with the Dragonfly C1000 were also announced: featuring 43TB of DRAM, with sampling expected in fiscal year 2026.

Qualcomm has planned three sub-segments for this CPU:

The first is the Agent CPU, aimed at high-throughput agent orchestration and low-latency interactive AI tasks.

The second is the General-Purpose CPU, balancing two needs: pursuing optimal TCO (Total Cost of Ownership) performance when running first-party workloads, and optimal vCPU (virtual CPU) performance for third-party elastic usage.

The third is the AI Head Node CPU, designed to complete host processing with low overhead, allowing XPUs to run at full capacity in generative AI computing.

What really gave the Dragonfly C1000 its weight was Meta's endorsement.

Qualcomm announced that the two companies signed a "multi-year, multi-generation" agreement, under which Meta will use the Dragonfly C1000 in its next-generation server clusters, with the chip planned for mass production in the second half of 2028. Subsequent CPU iterations are also included in the scope of cooperation.

Qualcomm CFO Palkhiwala stated that through handset chips and other existing products, Qualcomm already has business relationships with nearly all hyperscale enterprises, "This is not a newly established relationship." This statement implies that Meta is likely not the only company being negotiated with, and more customers may still be in discussion.

In response to external doubts about whether "Qualcomm is too late to enter the data center market," Qualcomm CEO Amon responded, "When people ask if it's too late to enter the data center now, you should think about scale and execution capability, engineering capability, or operations and supply chain."

His implication is that Qualcomm's large-scale system engineering capabilities, accumulated during the mobile phone era, remain effective in this market.

03. AI Accelerator Plus HBC, Breaking the "Memory Wall"

Beyond CPUs, Qualcomm also updated its AI accelerator roadmap.

Following the previously released AI200 and AI250, the AI300 inference accelerator made its debut at this Investor Day, with the three products iterating on an annual cadence.

The core logic of this platform is "disaggregated rack-scale AI inference." Tony Pialis, Senior Vice President and General Manager of Qualcomm's Data Center Business, explained that agentic workloads require CPUs, AI accelerators, and connectivity technologies to work together, rather than being handled by a single chip. What Qualcomm is doing now is integrating compute, AI, memory, and connectivity into a unified rack-level platform.

Within this platform, the memory issue is an unavoidable challenge, and the solution Qualcomm presents is High-Bandwidth Computing (HBC).

This is a technology designed to break the "memory wall." The so-called memory wall refers to the bandwidth bottleneck of moving data between the processor and memory in AI computing. HBC addresses this by tightly integrating compute units and memory using 3D stacked silicon technology, following a near-memory computing path.

Qualcomm provided several data points to illustrate the potential of HBC.

The AI250 equipped with HBC Gen 1 achieves an effective memory bandwidth of 133 TB/s per card, an 18x improvement over the AI200 using LPDDR5X. The AI300 with HBC Gen 2 will deliver a 54x bandwidth improvement compared to the AI200.

Compared to current mainstream HBM (High Bandwidth Memory), HBC offers 6 times the bandwidth at the same power consumption. Compared to SRAM (Static Random-Access Memory), HBC offers 200 times the capacity at the same power consumption.

In other words, the amount of data that HBC can process per unit of power is significantly increased, directly impacting data center total cost of ownership (TCO). Commercial samples of the AI250 are expected to be available in mid-2027, while commercial samples of the AI300 will come in 2028.

Connectivity products, a traditional stronghold for Qualcomm, were not absent this time. The company offers interconnect solutions ranging from Die-to-Die, copper cables, fiber optics to campus-level, supporting 800G and 1.6T speeds, covering scenarios from within data centers up to distances of 20 kilometers.

Over 35 technology ecosystem companies publicly expressed support for this roadmap, with the list including names such as Supermicro, Lenovo, SK Hynix, Micron, Samsung SDS, and Arista.

04. Acquiring Modular, Partnering with Hugging Face

Beyond hardware, Qualcomm is also making intensive moves in the software ecosystem.

First is the acquisition of AI software company Modular. The consideration is approximately $3.9 billion in Qualcomm stock, with the deal expected to close in the second half of 2026, pending regulatory approvals.

Modular's core product is an open, AI-native software stack that allows models to run on different chip architectures, including CPUs, GPUs, NPUs, and custom ASICs, eliminating the need for developers to rewrite code for each type of hardware. Co-founded by Chris Lattner and others, Modular's platform is seen in the industry as an open alternative to Nvidia's CUDA.

Commenting on the acquisition, Amon stated that as agents scale across data centers and the edge, the industry needs a more open and modern software foundation. Through this acquisition, Qualcomm hopes to provide customers with genuine deployment choices in a diversified computing environment.

Secondly, the company is expanding its collaboration with Hugging Face. The cooperation covers three areas:

* Bringing Hugging Face's internal and developer workloads to data centers powered by Qualcomm Dragonfly;

* Over 3 million open models on the Hugging Face platform can be directly loaded onto devices and data center racks equipped with Qualcomm platforms, simplifying the workflow for developers from experimentation to deployment;

* Developing the "Hugging Face Agent," which orchestrates AI workloads in hybrid on-device and cloud environments, dynamically allocating tasks based on performance, cost, and latency needs.

Hugging Face co-founder and CEO Clément Delangue explained: “We are enabling our 16 million developers to run open models easily anywhere, from the device in your hand to full racks in data centers.”

The collaboration also includes a specific arrangement. Hugging Face will provide Hugging Face PRO access to customers using Qualcomm platform devices or cloud systems, which includes advanced storage, compute, and collaboration features.

This step lowers the barrier for developers to build applications using open models.

05. Automotive, Robotics, China

Beyond the data center mainline, Qualcomm also updated progress data for other businesses.

On the automotive front, the 'automotive design win pipeline' has expanded to $65 billion, and Qualcomm raised the fiscal 2029 revenue target to $10 billion. Behind the demand for automotive chips is the ongoing penetration of ADAS and autonomous driving.

The IoT business has been broken down further. Industrial, networking, and robotics are listed separately, with a revenue target of $8 billion; personal AI and computing target $6 billion. Qualcomm believes that agents will trigger a new upgrade cycle for smart connected devices. The company estimates that the total scale of these businesses will reach $1.7 trillion by 2030.

Regarding the Chinese market, Amon gave a brief response during the event. The U.S. government currently has regulations on exporting AI-related hardware to China, but he said Qualcomm will launch a version of the data center chip that does not trigger export restrictions. He did not elaborate on specific plans, but this statement indicates that opportunities in the Chinese market have not been shelved.

In summary, the signals released at Qualcomm's Investor Day were quite complete. From C1000 to HBC, from acquiring Modular to collaborating with Hugging Face, from the $15 billion data center target to $18 earnings per share, all are verifiable milestones. Customers are secured, products have sampling timelines, and the financial model has been provided.

The financial reports in the coming quarters will be the first test of Qualcomm's roadmaps.