The Nvidia GB300 workstation, which HP claims is the most powerful Windows AI PC ever created, can manage one trillion parameters thanks to its 784GB unified memory, but it won't be inexpensive.
- At Computex, HP recently unveiled a number of AI-focused products powered by Nvidia's GPUs, such as the DGX Spark and the GB300 Grace Blackwell Ultra Desktop Superchip.
- Utilizing Nvidia's top-tier GPU, HP's premium deskside solution, the ZGX Fury GB300, provides up to trillion-parameter inference with up to 784GB of memory.
- Later in 2026, power users and AI fans should be able to purchase the ZGX Fury GB300.
Today is the last day of Computex 2026, and the natural topic of discussion was artificial intelligence (AI) and its advancements since ChatGPT was introduced in November 2022.
Nvidia's GTC 2026 announcements coincided with the expo, which had a "AI Together" theme and featured keynotes from several prominent figures in the field.
The most notable of Nvidia's announcements was the DGX Station, a potent supercomputer that can be installed deskside and provides computing equivalent to small datacenters because of its massive GB300 superchip and vast amounts of memory.
In front of more than 30,000 people from more than 190 nations on May 31, 2026, at GTC Taipei, Nvidia unveiled the DGX Station for Windows, a potent but expensive AI solution.
Nvidia markets the DGX Station as a desk-side AI supercomputer that can handle up to 1 trillion parameters locally.
Previously, only specialized datacenter-class hardware, such as Nvidia's DGX GB300 and its rack-scale GB300 NVL72 solutions, could accomplish this feat.
According to Chris Marriott, vice president of enterprise platforms at NVIDIA, "enterprises need AI infrastructure that can connect directly to the applications and workflows that power their business as they scale AI agents across their organizations."
In contrast to some of its more reasonably priced (in terms of AI) solutions, like the DGX Spark, the DGX station and its derivatives, like HP's ZGX Fury GB300 and Dell's Pro Max GB300, are anticipated to cost hundreds of thousands of dollars.
This is not surprising for the majority of their target market, which is enterprise consumers.
These clients could need the ability to fine-tune multi-billion-parameter models to meet their needs, frequently by localizing data to lessen dependency on the cloud, or the trillion-parameter inference that these setups can provide.
In addition to specialised enterprise networking solutions, products like HP's enterprise-focused DGX station, ZGX Fury GB300, offer up to 784GB of coherent memory and up to 20 petaflops FP4 compute for these applications.
"Over 70% of enterprise PCs run Windows, and our customers have asked for AI supercomputing power that can seamlessly integrate into their existing environments,
" said Jim Nottingham, SVP and Division President, Advanced Compute and Solutions, HP Inc., who also confirmed support for Windows in the future.
Although HP has not disclosed the price, it can be assumed that it would roughly mirror the DGX station's pricing,
which has seen some resellers offer mid-range configurations for $94,000 or more, with higher-end SKUs rising out at no more than $200,000.
Additionally, the product's ultimate release date has not been revealed. However,
it is anticipated to appear in Q4 2026 alongside the Nvidia DGX Station, working with additional partners like Dell, MSI, ASUS, and Supermicro.
