Dell Technologies executive Kevin Terwilliger tells CRN that he views edge computing as a ‘bigger opportunity’ than AI developers who need a compact, power-efficient PC for its iteration of Nvidia’s DGX Spark, which features a small but powerful chip for AI inference.
While Nvidia is pushing its new DGX Spark mini AI supercomputer as a lightweight desktop for developers prototyping, fine-tuning and running inference on AI reasoning models, Dell Technologies sees a bigger opportunity for the product: edge computing.
Announced this week at Nvidia’s GTC 2025 event, Dell is among a handful of OEMs that plan to release later this year their own version of the DGX Spark, which is expected to deliver up to 1,000 trillion operations per second, or 1 petaflop, of 4-bit floating point performance.
[Related: In DeepSeek Era, Nvidia Fixes On Blackwell Ultra’s AI Money-Making Potential]
This, in turn, will make the 2.6-pound PC suitable for fine-tuning and inferencing AI models with up to 200 parameters, according to Nvidia.
What makes this possible is a compact motherboard that features a smaller version of Nvidia’s Grace Blackwell Superchip called the GB10—which consists of a 20-core, Arm-based CPU and a Blackwell GPU. It also features 128 GB of unified coherent system memory, 273 GBps of memory bandwidth and Connect-X networking ports to connect with another GB10 device.
Slated for release this summer with an expected price point around $3,000, Dell’s version of the DGX Spark will be known as the Dell Pro Max with GB10, but the target audience for the device won’t be limited to AI developers who want a compact, power-efficient PC.
In an interview with CRN, Dell PC executive Kevin Terwilliger said he believes the “bigger opportunity” for its GB10-powered Dell Pro Max will be high-performance edge computing use cases like video surveillance in operational technology environments.
“If you think about an AI developer: swipe the credit card, get up and running super quick with that as an option. So that’s kind of the starting point of AI: doing that development and then it’s going to get deployed into the data center,” said Terwilliger, who is Dell’s vice president and general manager of commercial, consumer and gaming PCs.
“But then we actually think for big-model AI inference, it’s going to come back to a form factor like that. So big opportunity for us to actually enable that out in [operational] technology-type fields,” he added.
With the GB10-based Dell Pro Max providing significantly higher AI performance than a standard PC with similar dimensions, Terwilliger said the 180-watt-maximum device is intended for businesses who want a lot more horsepower for edge computing purposes.
“They’re going to want to do more and more edge compute from an AI perspective. And so we think that’ll be a big inference box as well,” he said.
While Terwilliger said he expects the GB10-powered device to appeal to industries like oil and gas and manufacturing, the executive thinks it could also play well with federal government customers who want air-gapped edge systems that can process AI models locally.
“It really comes down to what you’re able to get out of that data and the intelligence that you can go and pull. So it’s going to be use case by use case, and it’s also going to depend a little bit on when the models are ready for that,” he said.
“So I think this is going to happen over time as more of the models mature for things like video surveillance or manufacturing production—those types of things,” Terwilliger said.
With Dell planning to make an emphasis on the mini PC’s edge AI potential, the executive said the company plans to lean on channel partners who understand the space to sell it.
“A lot of times those channel partners have the expertise in those industries. And so big opportunity for us to enable them with that expertise,” he said. “We want to provide the infrastructure, but a lot of times those system integrators, VARs [and] others really own the industry, and so we’ll obviously work with their expertise.”
