Nvidia’s DGX Spark desktop supercomputer is on sale now, but hard to find

Nvidia’s “personal AI supercomputer,” the DGX Spark, may run fast but it’s been slow getting here. It finally went on sale on Oct. 15, five months later than the company initially promised, and early units are hard to find: Nvidia’s online shop was out of stock on launch day, and retailers have few units in store.

When Nvidia unveiled the machine, then known as Project DIGITS, at the CES 2025 electronics show in January, it said it would go on sale in May. But by the time Nvidia gave the product its final name and released more detailed specifications in March, the delivery date had slipped to “later this year”.

[ Related: More Nvidia news and insights ]

CEO Jensen Huang announced on Oct. 13 that Nvidia would begin sales of the DGX Spark two days later, but by 5am Eastern Time on the first day of sale Nvidia’s own online store was already showing a “sold out” message hard coded into the web page’s HTML: It wasn’t even checking stock availability in real time.

As a consolation, Nvidia’s online store offered links to four partners it said were selling the device, but none of them offered an online “buy” button.

Micro Center’s website reported that the retail chain had the DGX Spark in stock and available for in-store pickup at 29 of its 31 outlets. While a few of the stores showed “25+” units available, most had stock levels in the teens, and the Indianapolis store had just nine units available. Micro Center wouldn’t accept reservations or advance payments and warned that its stores will impose a one-per-household limit on DGX Spark buyers.

With the DGX Spark priced at a penny short of $4000 — almost double the cost of a high-end gaming PC — it’s unlikely that householders will be interested in the device. Back in March, when the pre-order price was just $3000, PC World was already predicting that it wouldn’t find many buyers.

Industrial demand

Nvidia’s DGX chips are in high demand in industry, though, and it’s more likely that Micro Center’s one-Spark limit is to prevent businesses scooping them up by the rack-load to run AI applications in their data centers.

The DGX Spark contains an Nvidia GB10 Grace Blackwell chip, 128GB of unified system memory, a ConnectX-7 smart NIC for connecting two Spark’s in parallel, and up to 4TB of storage in a package just 150mm (about 6 inches) square.

It consumes 240W of electrical power and delivers 1 petaflop of performance at FP4 precision — that’s one million billion floating point operations with four-bit precision per second. In comparison, Nvidia said, its original DGX-1 supercomputer based on its Pascal chip architecture and launched in 2016 delivered 170 teraflops (170,000 billion operations per second) at FP16 precision, but cost $129,000 and consumed 3,200W. It also weighed 60kg, compared to the Spark’s 1.2kg or 2.65 pounds.

Nvidia won’t be the only company selling compact systems based on the DGX Spark design: It said that partner systems will be available from Acer, Asus, Dell Technologies, Gigabyte, HP, Lenovo, and MSI.

More Nvidia news:

Inside Nvidia’s ‘grid-to-chip’ vision: How Vera Rubin and Spectrum-XGS advanceAI giga-factories

Nvidia and Fujitsu team for vertical industry AI projects

Nvidia and OpenAI open $100B, 10 GW data center alliance