[header]

At this year’s Computex Taipei, Qualcomm led the show with the first keynote, which was delivered by CEO Cristiano Amon. (While NVIDIA’s keynote earlier in the day predictably grabbed many headlines, it was not part of the official Computex 2026 program, as NVIDIA hosts its own parallel GTC Taipei event.) Qualcomm used Computex as a platform to tell its compute continuum story, showcasing its ability to scale AI from 2-milliwatt wearables to 200KW datacenter racks.

More specifically, Qualcomm used Computex to make numerous announcements across its businesses, including the launch of its new low-cost Snapdragon C processors for PC, the inclusion of its X2 Elite in the ASUS QN10 small-form-factor desktop, new Dragonwing offerings to support physical AI, and the debut of its new Dragonfly datacenter brand.

The Mysterious Snapdragon C Promises $300 PCs

Right before the show, Qualcomm announced the Snapdragon C. This is an entirely new category for the company, since all of its previous PC chips have been branded as Snapdragon X. The core message about the Snapdragon C is that it aims to deliver an improved PC experience in the $300 segment of the PC market. This is basically the absolute bottom of the PC market, and PC OEMs generally sell a lot of these machines, but as you would expect, the experience and quality aren’t necessarily the best. That said, with the PC market being as expensive as it is, bringing some relief to the budget-conscious end of the market is welcome. Acer has already shown off the Snapdragon C inside its new Aspire Go 15, which I got to play with at Computex; I found it acceptable, even though it had a bit of chassis flex. I also wonder how much memory and storage Acer will include for only $299.

Some people believe Qualcomm is mistaken to go after this segment of the market, but I think it could be a necessary move to force ISVs to take Windows on Arm seriously. Obviously, Qualcomm will get help on that front from NVIDIA with its RTX Spark efforts, but if Qualcomm gains enough market share, some ISVs will be forced to ship Arm-native apps that deliver better performance and battery life — for every segment of the PC market. At the show I saw a handful of new Snapdragon X laptop models from ASUS, which very much look like a way for ASUS to save money and still ship decent PCs. I do believe that the Arm-versus-x86 app compatibility issues are a thing of the past for most people, though Qualcomm’s drivers may still need some work to improve stability.

The biggest problem I have with the Snapdragon C is that Qualcomm didn’t disclose which chip it’s based on. We know it’s a mobile chip repurposed for PC applications, just like what Apple did with the A18 in the MacBook Neo — but that’s all we know. I believe this ambiguity has created unnecessary market uncertainty. It could set expectations incorrectly, and it definitely allows the rumor mill to ramp up before the products even launch. In that context, Qualcomm would be well-served to take control of the narrative by clarifying the Snapdragon C’s lineage and owning it.

There’s also work to be done with partners. I believe that Microsoft needs to continue working with Qualcomm to improve Windows performance for low-end PCs with memory capacities under 8GB, which appears to be the $299 entry-point target for OEMs using the Snapdragon C. OEMs like HP and Lenovo are also onboard with Snapdragon C, and I wonder whether all of their SKUs will be Windows PCs, or if we could see some Chromebooks or Googlebooks as well.

Qualcomm Gets Physical with the Dragonwing IQ10 Reference Design

One of the common themes at Computex was that everyone is not only getting on board with physical AI, but that physical AI helps to embody many of the things people are building with AI today. With the IQ10 robotics reference design (RRD), Qualcomm aims to replicate the success it achieved with its auto platform, which combined compute and software to help bring products to market faster. The Dragonwing IQ10 offers a wide range of connectivity for robotics applications, including support for a rich suite of sensors for low-latency sensor fusion. This is all powered by an 18-core Oryon CPU as well as multi-core NPUs and a GPU for added on-device AI acceleration.

It’s important to remember that the IQ10 RRD isn’t just for humanoid applications, even if those are the most exciting ones. There is also still demand for computing for mobile robots and robotic arms as well as humanoids. Qualcomm already has early access partners, including NEURA Robotics, Advantech, APLUX, Booster, Innodisk, MeiG, NEXCOM, Radxa, Thundercomm, and VinMotion. Qualcomm has also talked at length about its ability not only to offer robotics as a standalone solution, but also to provide a complete AI solution for the space where the robot operates. This extends, for example, to how AI-accelerated security cameras can help improve robots’ spatial awareness. There is also a growing need for more local compute, in general to avoid latency and additional cloud costs by deploying local AI devices, and in the context of robotics to help scale to deployments of multiple robots in the same space.

The Dragon Party Continues with Dragonfly at Investor Day 2026

At Computex 2026, Qualcomm added yet another “dragon” brand to the corporate arsenal. Years ago, Qualcomm introduced Snapdragon for its consumer products; at MWC 2025 in Barcelona, it introduced Dragonwing for enterprise and industrial applications; and now we have Dragonfly for the company’s datacenter products. Qualcomm also said that while it wouldn’t make any datacenter announcements at Computex, it does plan to make some significant ones at its investor day in New York City on June 24.

I expect that at the investor day we’ll hear about some major datacenter wins, see a roadmap for the company’s efforts in that space, and hear more about physical AI. Qualcomm last held an investor day in 2024, where it laid out a clear vision for the company towards 2030, so it will be interesting to see how those projections have changed and whether they are accelerating further.

Computex as a Setup for Qualcomm Investor Day

In my view, Qualcomm had a solid Computex 2026, and Cristiano Amon definitely made the case for the company’s milliwatt-to-kilowatt AI continuum strategy and how the company will power new agentic AI experiences. I also think that Qualcomm’s unique approach across wearables, mobile devices, automotive, PCs, robotics, and datacenters positions it differently from AMD, Apple, Intel, MediaTek, and NVIDIA. There’s no doubt that Qualcomm’s datacenter approach has lagged its competitors, but the company also comes from a background in mobile devices where efficiency is king, and that’s quickly becoming more important every day. Everyone wants to be able to talk about an end-to-end strategy, but few companies actually have one. Qualcomm has one.