NVIDIA DGX Station

NVIDIA DGX Station is a line of deskside artificial intelligence workstations from NVIDIA, each marketed as a "personal AI supercomputer" that puts data-center-class compute next to a developer's desk rather than in a server room. The line began in 2017 with a water-cooled tower built around four Tesla V100 GPUs, was rebuilt in 2020 around four A100 GPUs as the DGX Station A100, and was reinvented again in 2025 around a single NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip. ^[1][2][4] The current GB300 model carries up to 748GB of coherent memory and up to 20 petaflops of FP4 compute, and NVIDIA says it can run AI models of up to 1 trillion parameters locally. ^[2][11]

Field	Value
Product type	Deskside AI workstation, marketed as a "personal AI supercomputer"
Manufacturer	NVIDIA (reference design); built and sold by OEM and system-integrator partners
Product line introduced	May 2017, at GTC 2017
Generations	DGX Station (2017), DGX Station A100 (2020), DGX Station GB300 (2025)
Current model	GB300 Grace Blackwell Ultra DGX Station
Current superchip	NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip
Coherent memory (current)	Up to 748GB
FP4 compute (current)	Up to 20 petaflops (NVIDIA figure)
Local model size (current)	Up to 1 trillion parameters (per NVIDIA)
Operating system	NVIDIA DGX OS (Ubuntu-based Linux); a separate Windows variant exists
Availability (current)	Orders opened March 2026; OEM systems on general sale from mid-2026
OEM / integrator partners	ASUS, Dell, Exxact, GIGABYTE, HP, MSI, Supermicro, BIZON

The DGX Station sits between NVIDIA's smaller NVIDIA DGX Spark desktop unit and its rack-scale data-center systems, and it is part of the broader NVIDIA DGX family. This article covers the product line and its history. For the exhaustive GB300 specification table, the trillion-parameter claim in detail, NVIDIA OpenShell, and the separate Windows edition, see NVIDIA DGX Station for Windows. ^[2][17]

History of the DGX Station

Across three generations the DGX Station has kept one idea constant, a full data-center-class AI machine that plugs into a wall outlet and sits under a desk, while the silicon inside changed completely. The first two generations packed four discrete GPUs into a tower; the third replaced that with a single CPU-plus-GPU superchip.

The original DGX Station (2017)

NVIDIA unveiled the first DGX Station at its GPU Technology Conference in May 2017, alongside a Volta-based refresh of the rack-mounted DGX-1. It was an entirely new product rather than an upgrade of an earlier machine, positioned as a quiet, standalone workstation for organizations that wanted to train deep neural networks without access to a climate-controlled data center. ^[4][5] NVIDIA's own white paper titled it "The First Personal AI Supercomputer." ^[6]

The machine held four Tesla V100 Tensor Core GPUs (the initial configuration used the 16GB V100, and NVIDIA later offered a 32GB version) fully connected in a four-way topology using second-generation NVLink. That gave it 20,480 CUDA cores and 2,560 Tensor Cores. The rest of the system was a 20-core Intel Xeon E5-2698 v4 CPU, 256GB of DDR4 RDIMM memory, four 1.92TB SSDs (three in a RAID 0 data array plus one for the operating system), three DisplayPort outputs, and dual 10 gigabit Ethernet. The whole thing was water-cooled, ran inside a 1,500W power envelope quiet enough for an office (NVIDIA quoted under 35 dB), and launched at 69,000 US dollars. ^[5][6][7] NVIDIA rated the 16GB launch configuration at up to 480 teraflops of mixed-precision performance and the later 32GB configuration at 500 teraflops. ^[5][6]

DGX Station A100 (2020)

The second generation, the DGX Station A100, was announced on November 16, 2020, at the SC20 supercomputing conference. NVIDIA kept the deskside tower form factor and the self-contained cooling system that let the machine plug into a normal wall socket, but swapped in four Ampere-architecture A100 Tensor Core GPUs and pitched it as an "AI data-center-in-a-box." ^[8][9] Charlie Boyle, NVIDIA's vice president and general manager of DGX systems, said the product "brings AI out of the data center with a server-class system that can plug in anywhere." ^[8]

It came in two memory configurations: four A100 40GB GPUs for 160GB of total GPU memory, or four A100 80GB GPUs for 320GB. The four GPUs were fully connected with third-generation NVLink, the CPU was a single 64-core AMD EPYC 7742, and the system supported up to 512GB of DDR4 memory. NVIDIA rated it at 2.5 petaflops of AI performance. A distinctive feature was support for Multi-Instance GPU (MIG), which could partition the GPUs into as many as 28 separate instances so a small team could share one machine without interfering with each other's jobs. ^[8][9][10] NVIDIA sold the DGX Station A100 through its reseller partner network rather than publishing a fixed consumer price. ^[8][10]

DGX Station GB300 (2025)

The current generation was announced on March 18, 2025, at GTC, launched together with the smaller DGX Spark under the banner of "personal AI computers." ^[1][3] It is a clean break from the previous two designs. Instead of four discrete GPUs, it is built around a single NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, which fuses a 72-core NVIDIA Grace CPU and an NVIDIA Blackwell Ultra GPU into one coherent-memory package. NVIDIA CEO Jensen Huang framed the two new machines as a new class of computer "designed for AI-native developers." ^[1][3]

The GB300 Grace Blackwell Ultra model

At the heart of the current DGX Station is the GB300 Grace Blackwell Ultra Desktop Superchip. It pairs one 72-core Grace CPU (built on Arm Neoverse V2 cores) with one Blackwell Ultra GPU, joined by NVLink-C2C, a chip-to-chip link that carries 900 GB/s between the two and lets them share a single coherent pool of memory. This is the same class of silicon NVIDIA uses in its rack-scale NVIDIA GB300 NVL72, packaged here for one desktop unit. ^[2][11][13]

The 748GB figure combines the GPU's high-bandwidth memory with the CPU's larger but slower memory, so a very large model and its working data can live in one address space rather than being copied across a slow bus. The built-in ConnectX-8 SuperNIC runs at up to 800 Gb/s, which is what lets two or more DGX Stations be lashed together for a job that outgrows one box. ^[2][11][12] This page summarizes the hardware; the full specification table, including every precision tier, storage, and total system power, lives on the NVIDIA DGX Station for Windows page, which documents the identical GB300 platform. ^[2][17]

Software: DGX OS versus the Windows variant

The DGX Station ships with NVIDIA DGX OS, an Ubuntu-based Linux distribution (ServeTheHome reported the shipping systems run a DGX OS build derived from Ubuntu 24.04) preloaded with NVIDIA's CUDA-X AI stack, container runtime, and developer tools. Linux has been the native environment for the whole DGX line since 2017, and it remains the default for the GB300 model. ^[2][12][6]

In June 2026 NVIDIA introduced a second software flavor of the same hardware, the DGX Station for Windows, developed with Microsoft so that the NVIDIA AI stack runs on Microsoft Windows (with Windows Subsystem for Linux available for Linux-only tooling) and adding an agent runtime called NVIDIA OpenShell. That variant has its own detailed page, NVIDIA DGX Station for Windows (linked in the intro and See also). The two share the same GB300 silicon, up to 748GB of coherent memory, and the trillion-parameter claim; they differ in operating system and target buyer. ^[17]

Deskside AI for agents (SIGGRAPH 2026)

At SIGGRAPH 2026, described in an NVIDIA blog post dated July 20, 2026, NVIDIA positioned the DGX Station as the deskside home for agentic AI. The pitch is that a single machine can run a frontier agent stack entirely on-premises. According to NVIDIA, "On DGX Station, NVIDIA Agent Toolkit brings together NVIDIA NemoClaw, the NVIDIA Nemotron 3 Ultra open model ... and a secure runtime in a single local system," with "no internet required." ^[14]

The NVIDIA NemoClaw agent blueprint is part of the NVIDIA Agent Toolkit stack on DGX Station, alongside Nemotron 3 Ultra, which NVIDIA describes as a "frontier 550-billion-parameter open model" optimized to run on DGX Station GB300 systems. NVIDIA says the stack "can be running in roughly 30 minutes" and published step-by-step playbooks on its DGX Station project page, including "Connect Two DGX Stations for Distributed Workloads" and a guide to running NemoClaw with a local LLM on a dual DGX Station. The framing NVIDIA used was blunt: "Super agents have arrived on the desktop." ^[14][15] Running a 550-billion-parameter model this way trades the recurring per-token cost of a cloud API for a one-time hardware purchase, which is the economic argument NVIDIA makes for keeping agents local. ^[14]

Availability and partners

When it was announced in March 2025, the GB300 DGX Station was expected to be available "later this year," meaning later in 2025. ^[1][3] In practice, orders opened about a year later, around GTC 2026. On March 17, 2026, Tom's Hardware reported that the DGX Station was "now available to order and will begin shipping in the coming months," and ServeTheHome titled its March 20, 2026 coverage "Available At Last." ^[11][12] OEM systems then reached the market over the following months: ASUS announced worldwide availability of its version on June 15, 2026, HP's version became orderable on September 9, 2026, and by October 2026 reviewers had published hands-on tests of GB300 systems from ASUS, GIGABYTE, and MSI. ^{[21][23][24][27][28]}

NVIDIA designs the reference system and its OEM and system-integrator partners build and sell their own branded configurations. NVIDIA's March 2025 announcement named ASUS, BOXX, Dell, HP, Lambda, and Supermicro as expected manufacturing partners. ^[1] When orders opened at GTC 2026, ServeTheHome listed ASUS, Dell, GIGABYTE, HP, MSI, and Supermicro among the vendors confirmed to offer systems, and reported that HP said its system was targeting August. ^[12] NVIDIA's product page shows ASUS, Dell, Exxact, GIGABYTE, HP, MSI, and Supermicro as OEM partners, and BIZON sells a GB300 "AI Station" built to the DGX Station architecture. ^[2][19] NVIDIA strictly defines the motherboard, the soldered CPU, GPU and memory, the networking, and even the front and rear I/O, so the OEM versions differ mainly in chassis, cooling, power supply, and bundled software. ^[12][23]

NVIDIA did not publish a list price. At GTC 2026, ServeTheHome reported hearing prices around 100,000 US dollars, with the higher end at 120,000 to 125,000 US dollars, and called it "highly unlikely" that the systems would sell for under 80,000 to 85,000 US dollars. ^[12] Later retail listings fell in that range. In June 2026 Notebookcheck found the ASUS version listed by several US retailers, the cheapest at 99,999 US dollars from Central Computers, and in August 2026 Tom's Hardware reported that Exxact's Valence VWS-158270643 was listed from 94,930 US dollars, rising to 108,350 US dollars in a configuration with an RTX PRO 6000 Blackwell card and two extra 2TB SSDs. ^[25][26]

Dell and Supermicro were also confirmed as DGX Station vendors at GTC 2026. ^[12]

ASUS ExpertCenter Pro ET900N G3

ASUS's DGX Station is the ExpertCenter Pro ET900N G3, which ASUS describes as "Based on NVIDIA DGX Station Architecture." ^[22] ASUS first showed it at NVIDIA GTC Paris in June 2025, quoting up to 784GB of coherent memory, ^[36] and showed it again at CES 2026, in a press release dated January 6, 2026 that quoted 775GB of coherent memory; by the product's launch on June 15, 2026, ASUS listed the standard 748GB figure and said the system was "now available worldwide," with future support "planned" for Windows-based AI development. ^[20][21] ASUS sells it through pre-sales consultation rather than a published price. Its product page says higher-education institutions and NVIDIA Inception members may qualify for special offers, and that customers who qualify for the NVIDIA Fast Start Program may receive up to 50 percent off a specific eligible base configuration for their first two units, subject to NVIDIA approval. ^[22]

StorageReview, which tested a unit in September 2026, found three ASUS-specific touches compared with MSI's WS300: two carry handles on top, a dedicated fan aimed at the ConnectX-8 optics cages, and a power supply rated 80 PLUS Titanium rather than Platinum. ^[23] ASUS also bundles its own Zenni AI Hub software, which it says can deploy and route LLMs, manage MIG partitions, cluster several ET900N G3 systems over a LAN, and run multi-agent workflows including OpenClaw and Hermes; the system also supports NVIDIA NemoClaw and NVIDIA AI Enterprise. ^[21][22] In its launch release ASUS said its engineers' vLLM stress tests on an unnamed "massive Qwen open-source AI model" reached about 864 output tokens per second and about 1,600 tokens per second combined input and output, a vendor-run figure. ^[21]

Reviews and real-world performance

Reviews of OEM DGX Stations appeared from August 2026, starting with a remotely run StorageReview test of the MSI WS300. Every OEM uses the same NVIDIA baseboard, and StorageReview, which ran identical benchmarks on the MSI and ASUS systems, measured almost no performance difference between them.

StorageReview ran 14 models through two vLLM workloads (512 input and 512 output tokens, and a prefill-heavy 8,192 input and 1,024 output tokens) at 1 to 128 concurrent streams. On the ASUS unit, gpt-oss-120B reached 9,280 output tokens per second at 128 streams, about three times a single RTX PRO 6000 (3,064) and 2.7 times an H200 NVL (3,376). GLM-5.2 in NVFP4, with about 215GB of its expert weights held in Grace memory, ran at 139 output tokens per second at 32 streams, against about 40 for a server with four RTX PRO 6000 cards. NVIDIA's 550-billion-parameter Nemotron 3 Ultra in NVFP4 peaked at 159 output tokens per second on the ASUS unit and 168 on the MSI. Across 28 model-and-workload pairs, 24 of the ASUS results landed within 2 percent of the MSI WS300's, and the reviewers concluded that "The GB300 DGX Station remains the most capable thing we've ever put on a desk." ^[23] ServeTheHome reported running Qwen3.6-35B-A3B Flash in NVFP4 on the GIGABYTE system at "well over 26,000 tokens/second," and in its MSI review linked two stations over ConnectX-8 for an 800Gbps aggregate RDMA link. ^[28][29]

Concurrency test by Harrison Kinsley (sentdex)

Harrison Kinsley, who publishes machine learning videos on YouTube as sentdex, tested an ASUS ET900N G3 after ASUS offered to send him a demo unit for review, and published a video and a written report on October 18, 2026. ^[30][31] In an X post on October 17, he wrote that he was running "nemotron 120B A12B" with "64 concurrents at a bit over 100 tok/sec each for an aggregate of almost 7K tok/sec on a single GB300 (Asus DGX Station)"; ASUS's official account quoted the post on October 20 with the line "64 concurrent requests. Nearly 7,000 tokens/sec. One machine." ^[32][33]

His written report identifies the model as Nemotron 3.5 Super VL, NVIDIA's vision-language hybrid Mamba and mixture-of-experts model with 120 billion total and 12 billion active parameters, whose NVFP4 checkpoint NVIDIA released on Hugging Face on October 15, 2026. ^[30][34] It is a different model from the March 2026 Nemotron 3 Super, which carries the same 120B-A12B size label. ^[35] Kinsley ran an NVIDIA NVFP4 checkpoint (which he labeled "EA") on the single GB300 GPU without tensor or pipeline parallelism, with three-token multi-token prediction and an FP8 KV cache. ^[30] His code-generation sweep (1,000-token outputs) produced these results:

Kinsley described this as 13.9 times the throughput for 64 times the concurrency and called it "the best-scaling model I think I've ever seen." ^[30] The figures are one reviewer's measurements on a vendor-supplied unit with a specific serving configuration, not a standardized benchmark. In the same report he ran DeepSeek V4.1-Flash with an RTX PRO 6000 card added to hold 99 of each layer's cold experts and Grace memory holding the model's n-gram tables, measuring roughly 44,000 input tokens per second of prefill, and recorded a peak GB300 temperature of 75 degrees C on a unit whose GB300 was capped at 1,300W. He wrote that he would recommend the machine over a build with four to six RTX PRO 6000 cards, and for professional or enterprise use, and suggested that buyers configure it with an RTX PRO 6000 card. ^[30]

Where it sits in NVIDIA's lineup

The DGX Station is the middle tier of NVIDIA's "personal AI computer" range and the top of its deskside range.

Below the DGX Station is the DGX Spark, a much smaller and cheaper unit built on the GB10 Grace Blackwell Superchip with 128GB of unified memory and about 1 petaflop of FP4, marketed as "the world's smallest AI supercomputer" for individual developers; its pricing history is covered on the DGX Spark page. ^[1][11] Above the DGX Station are the rack-scale machines, the DGX B300 and systems such as the GB300 NVL72 and GB200 NVL72, which deliver many times the compute and memory but live in a server room rather than on a desk. The DGX Station is best understood as the deskside step between those two worlds. ^[2][11][12] For buyers who prioritize double-precision HPC throughput over AI density, ServeTheHome reported that Supermicro also offers a GB200-based desktop system, largely the same as a DGX Station but built on the earlier Blackwell B200 GPU and not sold under the DGX Station name. ^[12]

Significance

The DGX Station line tracks NVIDIA's answer to a recurring question: where should serious AI work run when the cloud is too slow, too expensive, or too exposed for sensitive data, but a normal workstation is too weak? In 2017 that answer was four V100s under a desk; by 2025 it was a single Grace Blackwell Ultra superchip capable, on NVIDIA's numbers, of holding a trillion-parameter model in local memory. ^[4][6][2] The 2025 to 2026 generation lands at the moment the industry's attention shifted from chatbots to autonomous agents, and NVIDIA has leaned into that, using SIGGRAPH 2026 to recast the machine as an on-premises home for agent stacks like NemoClaw and Nemotron 3 Ultra. ^[14] The through-line is on-premises control: keeping model weights, proprietary data, and inference cost inside an organization instead of renting them from a cloud. ^[16][17]

Limitations and caveats

Several of the headline numbers are NVIDIA's own vendor claims and deserve context.

• The trillion-parameter and 20-petaflop figures depend on low precision. Running a model of up to 1 trillion parameters locally, and the up-to-20-petaflop compute figure, both rely on FP4, a 4-bit format, which means the model must be quantized. These are NVIDIA's figures under favorable precision assumptions, not a guarantee for any model at full precision. ^[2][14][16]
• The shipping memory is smaller than first announced. NVIDIA's March 2025 announcement described 784GB of coherent memory; the shipping 2026 product carries 748GB. ServeTheHome attributes the difference to a binned Blackwell Ultra (B300) GPU with seven of eight HBM3e stacks enabled, giving 252GB of GPU memory instead of 288GB, roughly 12 percent less memory and bandwidth (about 7.1 TB/s). ^[1][12] ASUS's January 2026 CES preview of its OEM system quoted a third figure, 775GB, before the 748GB specification became standard. ^[20][21] NVIDIA's own specification table lists the 20-petaflop FP4 figure with sparsity and 15 petaflops without it. ^[2]
• No official list price. NVIDIA has not published one, and OEMs such as ASUS and MSI sell through sales inquiries. The prices in circulation are reseller listings, roughly 95,000 to 110,000 US dollars as of August 2026, and they vary with configuration. ^{[12][24][25][26]}
• Single-superchip design. Unlike the four-GPU 2017 and 2020 machines, the current model has one GB300 superchip. The only internal GPU expansion NVIDIA supports is one additional RTX PRO Blackwell card, which shares the system's 1,600W power budget with the superchip; scaling further means networking DGX Stations together over ConnectX-8. ^[2][12]

ELI5

Think of the DGX Station as a supercomputer shrunk down to fit under a desk and plug into a normal wall socket, so a researcher can train and run big AI models in their own office instead of renting time in a giant data center far away. The first one, in 2017, stuffed four powerful graphics chips into a quiet, water-cooled tower. The newest one, from 2025, uses a single super-chip that glues a big processor and a big graphics chip together and shares one huge pile of memory, big enough, NVIDIA says, to hold an AI model with a trillion adjustable numbers in it. In 2026 NVIDIA started pitching it as the machine that runs your own private AI "agents" (programs that can plan and do tasks for you) right on your desk, without sending anything to the internet.

Continue reading

• NVIDIA DGXNVIDIA DGX is NVIDIA's line of integrated artificial-intelligence supercomputers: systems that package the company's highest-end data-center GPUs, CPUs, high-speed NVLink and NVSwitch interconnects…Mentioned 14 times above
• Nemotron 3Nemotron 3 is a family of open-weights large language model systems released by NVIDIA beginning on December 15, 2025, built for agentic AI and consisting of three sparse mixture-of-experts variants named…Mentioned 6 times above
• NVIDIA DGX Station for WindowsNVIDIA DGX Station for Windows is a deskside artificial intelligence supercomputer announced by NVIDIA on June 1, 2026, at NVIDIA GTC Taipei during COMPUTEX 2026.Mentioned 3 times above
• NVIDIA BlackwellNVIDIA Blackwell is a family of graphics processing unit architectures and computing platforms developed by NVIDIA.Mentioned 3 times above
• NVIDIA GraceNVIDIA Grace is an Arm-based data-center central processing unit (CPU) from Nvidia, built from 72 Arm Neoverse V2 cores with co-packaged LPDDR5X memory and a coherent NVLink-C2C interconnect that links it to…Mentioned 3 times above
• NVIDIA DGX SparkNVIDIA DGX Spark is a compact, deskside AI development system, marketed by NVIDIA as a "personal AI supercomputer," that puts a Grace Blackwell superchip on a desktop and delivers up to 1 petaFLOP (1,000 TOPS)…Mentioned twice above

More from See also

• NVIDIA GB300 NVL72
• NVIDIA RTX PRO 6000 Blackwell

More to read

• Nvidia
• Model
• Artificial Intelligence
• ASUS

Topics:AI ComputeAI HardwareAI InfrastructureNVIDIA

References

1. ^1 ^2 ^3 ^4 ^5 ^6 ^7NVIDIA Newsroom. "NVIDIA Announces DGX Spark and DGX Station Personal AI Computers." March 18, 2025. nvidianews.nvidia.com/...ion-personal-ai-computers
2. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13NVIDIA. "NVIDIA DGX Station: The Ultimate Desktop AI Supercomputer." Product page. nvidia.com/...dgx-station
3. ^1 ^2 ^3Szkutak, Rebecca. "Nvidia announces two 'personal AI supercomputers.'" TechCrunch, March 18, 2025. techcrunch.com/...s-two-personal-ai-supercomputers
4. ^1 ^2 ^3NVIDIA Newsroom. "NVIDIA Advances AI Computing Revolution With New Volta-Based DGX Systems." May 2017. nvidianews.nvidia.com/...w-volta-based-dgx-systems
5. ^1 ^2 ^3Chiappetta, Marco. "NVIDIA Volta-Powered DGX-1 And DGX Station AI Supercomputers Debut At GTC 2017." HotHardware, May 10, 2017. hothardware.com/...upercomputers-debut-at-gtc-2017
6. ^1 ^2 ^3 ^4 ^5NVIDIA. "NVIDIA DGX Station: The First Personal AI Supercomputer." White paper (WP-V01). images.nvidia.com/...DGX-Station-WP.pdf
7. ^Kumar, Rohit. "NVIDIA DGX Station Upgraded to Tesla V100." ServeTheHome, October 28, 2017. servethehome.com/...gx-station-upgraded-tesla-v100
8. ^1 ^2 ^3 ^4NVIDIA Newsroom. "NVIDIA DGX Station A100 Offers Researchers AI Data-Center-in-a-Box." November 16, 2020. nvidianews.nvidia.com/...s-ai-data-center-in-a-box
9. ^1 ^2Robinson, Cliff. "NVIDIA DGX Station A100 Brings 320GB HBM2e to Desktops." ServeTheHome, November 16, 2020. servethehome.com/...brings-320gb-hbm2e-to-desktops
10. ^1 ^2Mujtaba, Hassan. "NVIDIA Announces DGX Station A100 With Upgraded 80 GB A100 Tensor Core GPUs, Up To 320 GB Memory & 2.5 Petaflops of AI Horsepower." wccftech, November 16, 2020. wccftech.com/...graded-80-gb-a100-tensor-core-gpus
11. ^1 ^2 ^3 ^4 ^5 ^6Klotz, Aaron. "Nvidia launches DGX Station with its bleeding-edge GB300 Grace Blackwell Superchip, now available to order and will begin shipping in the coming months." Tom's Hardware, March 17, 2026. tomshardware.com/...-shipping-in-the-coming-months
12. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12Smith, Ryan. "NVIDIA DGX Station Systems Available At Last: GB300 and GB200 Workstations For Your Desktop." ServeTheHome, March 20, 2026. servethehome.com/...-workstations-for-your-desktop
13. ^Mujtaba, Hassan. "NVIDIA DGX Station Upgraded With GB300 Blackwell Ultra Desktop Superchip: 748 GB Memory, 20 PFLOPs AI Compute & AI Ready." wccftech, March 17, 2026. wccftech.com/...-blackwell-ultra-desktop-superchip
14. ^1 ^2 ^3 ^4 ^5NVIDIA Blog. "At SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI." July 20, 2026. blogs.nvidia.com/...siggraph-news-2026
15. ^NVIDIA. "NVIDIA DGX Station" project and playbooks page. build.nvidia.com/station
16. ^1 ^2Nuñez, Michael. "Nvidia's DGX Station is a desktop supercomputer that runs trillion-parameter AI models without the cloud." VentureBeat, March 16, 2026. venturebeat.com/...er-that-runs-trillion-parameter
17. ^1 ^2 ^3 ^4NVIDIA Newsroom. "NVIDIA DGX Station for Windows Puts a Trillion-Parameter AI Supercomputer on Every Enterprise Desk." June 1, 2026. nvidianews.nvidia.com/...-on-every-enterprise-desk
18. ^Exxact Corporation. "Exxact Valence NVIDIA DGX Station, 1x NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip." Product page. exxactcorp.com/Exxact-VWS-158270643-E158270643
19. ^1 ^2BIZON. "AI Station with NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip (NVIDIA DGX Station GB300 architecture)." Product page. bizon-tech.com/gb300
20. ^1 ^2 ^3ASUS Pressroom. "ASUS Showcases ExpertCenter Pro ET900N G3 at CES 2026 to Herald New Era of Deskside AI Supercomputing." January 6, 2026. press.asus.com/...center-pro-et900n-g3-at-ces-2026
21. ^1 ^2 ^3 ^4 ^5 ^6ASUS Pressroom. "ASUS Launches ExpertCenter Pro ET900N G3 Built on NVIDIA DGX Station Architecture." June 15, 2026. press.asus.com/...et900n-g3-nvidia-gb300-blackwell
22. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12ASUS. "ExpertCenter Pro ET900N G3." Product page and specifications. Accessed October 30, 2026. asus.com/...expertcenter-pro-et900n-g3
23. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14Beeler, Brian. "ASUS ExpertCenter Pro ET900N G3 Review: The GB300 DGX Station Gets Handles, Titanium Power, and Cooled Optics." StorageReview, September 23, 2026. storagereview.com/...rtcenter-pro-et900n-g3-review
24. ^1 ^2 ^3 ^4Jain, Divyansh. "MSI XpertStation WS300 Review: 748GB of Coherent Memory and 20 PetaFLOPS on a Desk." StorageReview, August 24, 2026. storagereview.com/...ry-and-20-petaflops-on-a-desk
25. ^1 ^2Brecher, Hannes. "Asus präsentiert Tower mit 748 GB RAM und Nvidia GB300 für über 86.000€." Notebookcheck, June 15, 2026. notebookcheck.com/...uer-ueber-86-000EUR.1322270.0
26. ^1 ^2 ^3Nasir, Hassam. "Nvidia's GB300-powered DGX Station desktop tower listed for nearly $100,000 online." Tom's Hardware, August 23, 2026. tomshardware.com/...mere-mortals-with-lots-of-cash
27. ^1 ^2Beeler, Brian. "HP ZGX Fury Is Now Orderable: GB300 Superchip, 748GB Unified Memory, and a Red Hat AI Factory Plan for the Edge." StorageReview, September 9, 2026. storagereview.com/...-ai-factory-plan-for-the-edge
28. ^1 ^2 ^3 ^4Kennedy, Patrick. "Gigabyte W775-V10-L01 Hands-on Bringing NVIDIA GB300 Deskside." ServeTheHome, October 6, 2026. servethehome.com/...bringing-nvidia-gb300-deskside
29. ^1 ^2 ^3Smith, Ryan. "MSI XpertStation WS300 Review NVIDIA DGX Station GB300." ServeTheHome, October 23, 2026. servethehome.com/...eview-nvidia-dgx-station-gb300
30. ^1 ^2 ^3 ^4 ^5 ^6Kinsley, Harrison. "A Datacenter at Home." Reflections (hkinsley.com), October 18, 2026. hkinsley.com/...a-datacenter-at-home
31. ^1 ^2sentdex. "Datacenter at Home - GB300 GPU - w/ Asus DGX Station (ET900N G3)." YouTube, October 18, 2026. youtube.com/watch
32. ^Kinsley, Harrison (@Sentdex). Post on X, October 17, 2026. x.com/...2111600208966472017
33. ^ASUS (@ASUS). Post on X, October 20, 2026. x.com/...2112506619913482247
34. ^NVIDIA. "NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-NVFP4." Model card, Hugging Face, released October 15, 2026. huggingface.co/...ron-3.5-Super-VL-120B-A12B-NVFP4
35. ^NVIDIA. "NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4." Model repository, Hugging Face, March 2026. huggingface.co/...Nemotron-3-Super-120B-A12B-NVFP4
36. ^1 ^2ASUS Pressroom. "ASUS Announces Key Milestone with Nebius and Showcases NVIDIA GB300 NVL72 System at GTC Paris 2025." June 11, 2025. press.asus.com/...asus-nvidia-gb300-nvl72-ai-factory

Show all 36 references