Dell Technologies’ announcement of a Dell Pro Max Plus laptop that will use a Qualcomm AI 100 PC Inference Card shows how OEMs are seizing on the increasing competition between semiconductor companies to differentiate against their own rivals.
Since last year, Qualcomm has been making inroads in the PC market with its Arm-compatible Snapdragon X Series processors, thanks to growing adoption by the likes of Dell Technologies, HP Inc., Lenovo and several other vendors.
Mostly known for its smartphone chips, the San Diego, Calif.-based chip designer signaled that it wanted to make a renewed push with PCs when it acquired chip design startup Nuvia in 2021 to deliver “step-function improvements in CPU performance and power efficiency” for laptops, among other kinds of devices.
[Related: With Nvidia’s DGX Spark Mini AI PC, Dell Sees Big Edge Computing Potential]
The acquisition, combined with internal development efforts, has allowed Qualcomm to secure more than 85 PC design wins with OEMs so far, and that number is expected to surpass 100 by next year, the company said last month.
But if you thought the Snapdragon X Series represented the extent of its compute ambitions in the PC market, you’d be wrong.
In a move that caught me by surprise, Dell announced on Monday a new Dell Pro Max Plus laptop that will use a Qualcomm AI 100 PC Inference Card. This will make it the “world’s first workstation with an enterprise-grade discrete NPU,” it said.
It’s not news that Qualcomm is in the business of providing NPUs (neural processing units) for PCs. After all, the NPU is a key component in the Snapdragon X Series system-on-chips, and their NPU performance is what helped Qualcomm secure a deal to supply processors for the first wave of devices in Microsoft’s Copilot+ PC program last year.
I did not foresee, however, that Qualcomm would provide a discrete NPU product for a laptop that consists of two of the company’s Cloud AI 100 data center processors housed on a circuit board, according to an image from Dell.
With these two Cloud AI 100 processors combined, the inference card features a total of 32 AI cores, 64 GB of on-board LPDDR4x memory, allowing it to run more than 450 tera operations per second of 8-bit integer math in a thermal envelope of up to 75 watts, Dell said.
This, in turn, allows the Dell Pro Max Plus to “offer fast and secure on-device inferencing at the edge for large AI models typically run in the cloud,” including models with up to 109 billion parameters, the company added.
It’s an interesting move in part because it shows that Qualcomm has ambitions in providing an alternative to power-hungry GPUs that can consume more energy have traditionally gone into high-performance desktops and laptops as discrete parts.
But it also demonstrates how OEMs like Dell are seizing on the increasing competition between semiconductor companies to differentiate against their own rivals. About 10 years ago, the PC world largely revolved around Intel, but things have significantly changed since then, with AMD, Nvidia, Qualcomm and Apple all offering a variety of computer chip options.
“I think we see there’s a competitive world around silicon, and it’s been good,” Sam Burd, president of Dell’s Client Solutions Group, told CRN in an interview this week. “To me, competition and the desire to go win and do better stuff for customers is really good. So we’re going to ship the best parts for our customers.”
