The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Yes. NVIDIA announced a Windows version of its GB300-based DGX Station, with Windows availability expected in Q4 2026 through six named hardware partners. Windows is the new software path—not the end of Linux: NVIDIA says developers can use Windows Subsystem for Linux (WSL) to access Linux AI toolchains. The headline figures also need context: 748 GB is combined coherent CPU and GPU memory, while 20 PFLOPS refers specifically to FP4 Tensor Core compute with sparsity.
What changed with the Windows DGX Station?
NVIDIA’s June 1, 2026 announcement brings the DGX Station into a Windows environment aimed at enterprise and workstation users. The system is built around NVIDIA’s GB300 Grace Blackwell Ultra Desktop Superchip. NVIDIA lists ASUS, Dell Technologies, GIGABYTE, HP, MSI, and Supermicro as expected providers in Q4 2026. That is an expected availability window, not a confirmed ship date for every partner or configuration. NVIDIA’s announcement and its Windows product page do not establish prices or order dates.
As an Amazon Associate I earn from qualifying purchases.
The change is the operating-system option and its integration into a Windows enterprise environment. NVIDIA describes WSL as the way to use Linux AI toolchains on Windows. That does not mean every Linux application is guaranteed to work, nor does it establish dual boot or native compatibility for all Linux software. NVIDIA’s DGX Station development guide documents a separate Ubuntu 24.04 software environment. The development guide and Windows product information describe distinct software paths.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsWhat does 748 GB of coherent memory mean?
NVIDIA advertises up to 748 GB of coherent memory, composed of two different memory pools: 252 GB of HBM3e GPU memory and 496 GB of LPDDR5X CPU memory. It is not 748 GB of GPU VRAM. The Grace CPU and Blackwell Ultra GPU are connected through NVLink-C2C, which NVIDIA lists at 900 GB/s.
#1 Best Overall
- All-in-One Connectivity: Build a tidy workstation with 17 ports: 3×USB-A 10Gbps, 3×USB-C 10Gbps (2 ports share 60W charging), SD/TF 4.0, 3×3.5mm audio, 2.5G Ethernet, DP1.4, and 2×Thunderbolt 5 downstream ports
- 120Gbps Thunderbolt 5 Speed & M.2 High-Speed Storage: Intel-certified Thunderbolt 5 provides up to 120Gbps bandwidth. Built-in NVMe PCIe Gen4x4 M.2 slot supports 2230–2280 SSDs up to 8TB for fast external storage. Note: Compatible with M.2 NVMe SSDs with heatsink (height≤12 mm; width≤24 mm), Not compatible with SSDs wider than 24 mm, including Seagate FireCuda 530 with Heatsink, Samsung 9100 PRO with Heatsink, WD_BLACK SN850X Premium Model with OEM Heatsink, and Lexar NM1090 Active Fan Version
- Multi-Monitor 8K/6K Output: Supports triple independent displays on Windows (2×TBT5+DP): dual or single 8K@60Hz. Mac supports dual displays. Native compatible with latest Windows and MacBooks for expanded workspace. Note: The DP port does not support DP++. To connect an HDMI monitor, please use an active DP-to-HDMI adapter or cable; passive adapters and cables will not work
- 240W Total System Power & High-Speed Networking (required) Equipped with up to 240W total power and a built-in 2.5GbE network port for fast and reliable wired connections, with lower latency and a more stable connection
- AI Cooling & Aluminum Chassis: AI smart cooling with 60mm ultra-thin fan ensures stable performance under heavy load. Premium aluminum unibody offers great heat dissipation, scratch resistance, and a clean matte finish, paired with a reliable security lock. Note: It is normal for the cooling fan to make noise when it is running and the noise is below 30 dB (data from UGREEN lab)
A coherent pool can make it easier for CPU and GPU work to access system memory without treating the full advertised capacity as dedicated graphics memory. The distinction matters when estimating whether a workload fits: the GPU’s HBM3e capacity is 252 GB, while the rest is CPU memory. NVIDIA’s stated capacity does not by itself guarantee that a particular model or application will fit or run at a given speed.
Is the 20 PFLOPS figure a general speed rating?
No. NVIDIA’s 20 PFLOPS headline is peak FP4 Tensor Core compute with sparsity. The same product information lists 15 PFLOPS FP4 without sparsity. Other advertised peak figures vary by precision, so they are not interchangeable:
| Precision or format | NVIDIA-listed compute | Qualification |
|---|---|---|
| FP4 | Up to 20 PFLOPS | Tensor Core compute with sparsity |
| FP4 | 15 PFLOPS | Without sparsity |
| FP8/FP6 | 10 PFLOPS | Advertised figure for these formats |
| FP16/BF16 | 5 PFLOPS | Advertised figure for these formats |
| FP32 | 80 TFLOPS | Advertised figure for this format |
These are NVIDIA specifications, not independent benchmark results for the Windows model. Actual performance depends on the workload, software, precision and use of sparsity; the figures should not be read as a universal measure of application speed.
Recommended Free Tools
Rank #2
- VD8465 Japanese Authorized Distributor Product
- The speed of FP32 calculation is twice as fast as previous generations, which greatly improves the complex 3D processing and graphics simulation workflow
- Up to 2X the throughput compared to previous generations and significantly faster workloads such as video content rendering, architectural design assessments, and virtual prototypes of product design
- Achieve more than twice the previous generation AI performance improvement, support faster FP8 precision data and accelerate the execution of mixed flotation decimal and whole numbers
- It has a large capacity of memory necessary for working with a vast array of data sets and workloads such as rendering, data science, and simulation
What workloads and model sizes is it meant for?
NVIDIA positions the DGX Station for local AI development, fine-tuning, inference, data science and AI agents connected to Windows applications. It says the platform supports models up to one trillion parameters. Treat that as NVIDIA’s stated capability, not a guarantee that an arbitrary trillion-parameter model will run with any chosen settings or workload.
The system has a 72-core Grace CPU alongside the Blackwell Ultra GPU. NVIDIA also lists ConnectX-8 SuperNIC networking of up to 800 Gb/s. Those component specifications may matter for data movement and deployment planning, but they do not promise a particular network throughput in a real environment.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.When will the Windows model be available?
NVIDIA’s June 1, 2026 announcement expects systems from ASUS, Dell Technologies, GIGABYTE, HP, MSI and Supermicro in Q4 2026. At the time reflected by NVIDIA’s product page, the Windows system is marked “Coming in Q4.” Microsoft’s October 7, 2026 post separately says Dell Pro Precision with GB300 and HP ZGX Fury AI Station are coming later in 2026; it does not establish their exact prices, configurations or order dates. Microsoft’s Windows Experience Blog post corroborates those two named systems, not a first-to-ship ranking across all six partners.
Rank #3
- NVIDIA Quadro Sync II1 compatibility
- 3D stereo support with stereo connector
- NVIDIA GPUDirect for Video support
- NVIDIA GPUDirect Remote Direct Memory Access (RDMA) support
- NVIDIA RTX ExperienceTM
What should buyers verify with an OEM?
The product specifications leave important purchase details to each system provider. Before comparing offers, verify the exact workstation SKU and its configuration rather than assuming every DGX Station includes the same options.
- GB300 configuration: Confirm the exact processor platform and included memory and storage.
- Operating system and software: Ask which Windows edition and AI software stack are supported, and what WSL setup and workload compatibility the OEM documents.
- Optional GPU: NVIDIA lists optional RTX PRO models including RTX PRO 6000 Workstation Edition, RTX PRO 6000 Blackwell Max-Q Workstation Edition, RTX PRO 4000 Blackwell SFF Edition and RTX PRO 2000 Blackwell. Support varies by system. NVIDIA identifies the RTX PRO 6000 as an option for ray-traced visualization and simulation; confirm the exact supported card and SKU with the OEM.
- Power and facilities: NVIDIA lists 1,600 W total system power and a 20 A circuit requirement. Check that the intended location can meet the OEM’s requirements.
- Networking: Confirm which networking hardware is included and how it fits the organization’s network; the listed ConnectX-8 capability is up to 800 Gb/s, not a throughput guarantee.
- Commercial terms: Check warranty, enterprise support, regional availability, delivery timing and price against the actual configuration.
Until OEM listings specify these details, there is no sound basis for claiming which partner is shipping first or for comparing prices.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
