Nvidia has officially revealed its new DGX Spark and DGX Station "personal AI supercomputers" at the GTC conference, both powered by the company's Grace Blackwell platform. The DGX Spark, previously known as "Digits," is a Mac Mini-sized AI desktop that will be available for preorder today, while the larger DGX Station desktop counterpart is aimed at AI developers, researchers, data scientists, and students.

The DGX Spark is a compact, $3,000 AI desktop that boasts impressive specs, including Nvidia's GB10 Blackwell Superchip, featuring a GPU with fifth-generation Tensor Cores and FP4 support. This powerful combination enables the Spark to deliver up to 1,000 trillion operations per second (TOPS) of AI compute for fine-tuning and inference with the latest AI reasoning models. Additionally, the Spark comes with 128GB of unified memory and up to 4TB of NVMe SSD storage.

In contrast, the DGX Station is a larger, more powerful desktop that accommodates Nvidia's GB300 Blackwell Ultra desktop super chip, providing 20 petaflops of AI performance and 784GB of unified system memory. While the Spark is designed for individual users, the DGX Station is geared towards teams and organizations working on large-scale AI projects.

Nvidia has also announced that OEM partners, including Asus, Dell, HP, Boxx, Lambda, and Supermicro, will be building their own versions of the DGX computers. These customized DGX Stations will be available later this year, while the DGX Spark will have versions made by Asus, Dell, HP, and Lenovo. Interested customers can reserve a DGX Spark on Nvidia's website, with deliveries expected this summer.

The launch of the DGX Spark and DGX Station marks a significant milestone in the development of AI computing, as it brings powerful AI capabilities to the desktop. This move is likely to have a profound impact on the AI research and development community, enabling users to work on complex AI models without relying on datacenter connections.

It's worth noting that Nvidia is not the only company pushing the boundaries of AI computing. Rival AMD has its Ryzen AI Max+ "Strix Halo" GPU, which also features a large amount of unified memory for local large language models (LLMs). HP has integrated this GPU into a laptop, while Framework has built it into a $2,000 desktop. These developments highlight the intense competition in the AI computing space, as companies strive to deliver more powerful and efficient solutions for AI researchers and developers.

In conclusion, the launch of the DGX Spark and DGX Station marks a significant step forward in the development of AI computing, bringing powerful AI capabilities to the desktop and enabling users to work on complex AI models without relying on datacenter connections. As the AI computing landscape continues to evolve, it will be interesting to see how these new products shape the industry and drive innovation in the years to come.