Vera Rubin price, performance and timing: what buyers should plan for

This is the buyer's part of our three-part guide to NVIDIA Vera Rubin: what drives the price, how much faster it really is, and when it reaches most buyers. For the short answers, start with Vera Rubin explained. For the full spec tables and the cable-free build, see the Vera Rubin hardware deep dive.
Figures are as of 24 October 2026. We separate published specs, NVIDIA's own claims and analysts' estimates, and label ratios we calculated as "our arithmetic".
What drives the price of Vera Rubin
Up, roughly double per rack. NVIDIA publishes no prices, but analysts estimate a Vera Rubin NVL72 rack at about US$7.8 million to US$9.1 million, about twice a GB300 NVL72 on the same analysts' numbers. Each estimate, and who made it, is in what Rubin costs. These are purchase prices, not rental prices.
Why it costs more: memory
Memory is now the biggest driver. Morgan Stanley estimates memory at about a quarter of a Vera Rubin rack's cost, up from about 9% in a GB300 rack. Bernstein puts memory and storage together at about 35%. And memory is getting dearer.
TrendForce forecasts the average HBM price to rise 121% in 2027. Industry estimates put a 36 GB HBM4 stack at about US$600 in 2026 and about US$1,300 in 2027. Michael Dell said in October 2026 that memory, not demand, is the industry's bottleneck, and that it gets worse in 2027.
Memory takes a bigger share of the rack’s cost
Share of the rack’s cost. Bernstein’s figure counts memory and storage together.
Source: Morgan Stanley; Bernstein
NVIDIA appears to be passing this on. In August 2026 Bloomberg, citing unnamed sources, reported that NVIDIA had told some of its biggest customers of price rises, above 15% in many cases. The rises cover Grace Blackwell and Vera Rubin systems shipping from early 2027; NVIDIA did not comment.
On its results call the same month, NVIDIA's finance chief said memory prices were above expectations. Price increases NVIDIA had already made take effect in the first quarter of its 2028 financial year (about February to April 2027); products and amounts were not named.
Cheaper per unit of compute, not per box
A Rubin rack costs more but does more. On Morgan Stanley's figures, a Vera Rubin rack costs about 1.95 times as much as a GB300 rack. It delivers about 2.3 times the dense FP4 compute, so it costs about 16% less per unit of dense FP4 compute (our arithmetic; the result moves a lot with the GB300 price used).
Against the measured MLPerf results further down, the same price gap works out to anywhere from about 12% more to about 24% less hardware cost per token, depending on the test (our arithmetic, hardware price only).
NVIDIA's own claim is far bigger: one tenth the cost per million tokens (the word pieces a model reads and writes). But that is a projection for one large reasoning model, Kimi-K2-Thinking, compared with the older GB200 NVL72. Real savings depend on your model and on how fast your users need answers.
Blackwell is not getting cheaper
Normally a new generation pushes the old one's prices down. H100 rental prices at specialist GPU clouds fell from peaks of about US$8 an hour in 2023 to US$1.50 to US$3.00 by late 2025, over the period in which Blackwell arrived (Silicon Data). In 2026 that pattern broke.
AWS's published price for reserved 8-GPU B300 capacity blocks in Oregon rose about 38% between December 2025 and October 2026. Its B200 price in Ohio rose about 52%. One public price tracker shows the same direction: on-demand list prices for 17 of the 21 GPU models it follows, including the H100, were higher in October 2026 than a year earlier.
SemiAnalysis estimated in September 2026 that three-year rental rates in July 2026 were over US$8.50 per GPU-hour for Rubin. That compares with about US$5 per GPU-hour in a GB300 NVL72.
Rubin costs more per GPU-hour
Three-year rental rates in July 2026, per GPU-hour. SemiAnalysis also estimates about 62% more output per rental dollar for Rubin, at 80 tokens per second per user.
Source: SemiAnalysis, September 2026
But SemiAnalysis estimates Rubin gives about 62% more output per rental dollar, at a typical speed of 80 tokens per second per user. In short: expect to pay more per GPU-hour for Rubin, and judge it on cost per result, not on the hourly rate.
Vera Rubin vs Blackwell performance: how much faster, and why
About twice the work per GPU in the first peer-reviewed benchmark. In MLPerf Inference v6.1 (September 2026), a Vera Rubin NVL72 rack did 1.7 to 2.6 times the DeepSeek-R1 work per GPU of a GB300 NVL72 (MLCommons results). No result for an 8-GPU Rubin server has been published yet.
The gain comes from four places, roughly in order of importance for serving AI models:
- Memory bandwidth: writing each word (token) of an answer means reading the model's weights from memory, so speed is often limited by memory bandwidth, not compute. Rubin's is about 2.4 times Blackwell's in the rack.
- Low-precision compute: about 2.3 times the dense FP4 compute of a GB300 GPU, and about 3.5 times the FP8. Gains in BF16, a common format for training, are smaller: about 1.8 times per 8-GPU server.
- NVLink 6: about 1.7 times the GPU-to-GPU bandwidth. This matters most for mixture-of-experts (MoE) models, which split each layer into many "experts" spread across GPUs that constantly exchange data.
- Rack-scale serving: the larger CPU memory holds the KV cache of long conversations, and BlueField-4 adds a fast storage tier for it. NVIDIA's Dynamo software also splits the work, so some GPUs read the prompt while others write the answer.
NVIDIA's claims, and what they compare against
NVIDIA claim | Compared with | The catch |
|---|---|---|
One tenth the cost per million tokens | GB200 NVL72 | A projection for one model (Kimi-K2-Thinking) |
One quarter of the GPUs to train a 10-trillion-parameter MoE model in a month | GB200 NVL72 | A projection |
10 times the token output of an 8-GPU server | HGX B200 | A projection that compares Rubin's sparse figure with the B200's dense one |
Up to 30 times the tokens per MW for AI coding agents | GB300 NVL72 | Early results, best case only; NVIDIA states no speed target, precision or GPU count |
Up to 2.5 times on DeepSeek-R1 in MLPerf | GB300 NVL72 | The best of three MLPerf scenarios (interactive, 2.6x on our arithmetic); the other two gave 1.7x and 2.0x |
The first benchmark results
The first peer-reviewed numbers came on 16 September 2026, in MLPerf Inference v6.1, the industry benchmark run by MLCommons. Vera Rubin NVL72 was entered in the preview category, for systems not yet on general sale, by NVIDIA and by one GPU cloud. The submitters run the tests themselves, and the results are then peer-reviewed.
DeepSeek-R1, tokens per second per GPU | GB300 NVL72 | Vera Rubin NVL72 | Gain |
|---|---|---|---|
Offline (bulk processing) | 9,441 | 16,435 | 1.7x |
Server (steady online traffic) | 8,291 | 16,332 | 2.0x |
Interactive (fast responses) | 3,521 | 9,066 | 2.6x |
MLPerf DeepSeek-R1: Rubin rack against GB300 rack, per GPU
Tokens per second per GPU. Per rated watt, the gain shrinks to about 1.05 to 1.6 times (our arithmetic).
Source: MLCommons, MLPerf Inference v6.1, 16 September 2026
So a Vera Rubin rack did about twice the work per GPU of a GB300 rack, and the gap was widest when users need fast answers. On Qwen3-VL-235B, a large vision model run in BF16 on both systems, the gains were 1.8 to 3.7 times. Three cautions:
- Power: each Rubin GPU is rated at up to 2,300 W, against 1,400 W for a GB300 GPU. Per rated watt, the DeepSeek-R1 gain shrinks to about 1.05 to 1.6 times (our arithmetic). MLPerf did not measure power in these runs.
- Blackwell keeps improving: software alone made GB300 NVL72 1.6 to 2.9 times faster on DeepSeek-R1 between MLPerf v5.1 and v6.1, a year apart. Rule changes may explain part of that.
- Claims shrink in practice: NVIDIA claimed 1.5 times the dense FP4 compute for GB300 over GB200; in MLPerf, throughput rose 1.45 times offline and only about 1.25 times in the server test.
Nobody has yet published a training benchmark for Rubin, or any result for an 8-GPU Rubin server. In MLPerf v6.1 the Rubin rack did about 1.9 to 2.2 times the work per GPU of NVIDIA's 8-GPU DGX B300. That compares a 72-GPU rack with an 8-GPU node, at about twice the rated power per GPU (our arithmetic).
And the best 8-GPU B300 server beat the GB300 rack per GPU in the server test: 8,847 against 8,291 tokens per second. So for steady, high-volume serving that fits in eight GPUs, a B300 node keeps pace with a GB300 rack.
Who gains the most
The biggest gains are for large MoE reasoning models with long inputs, served at fast response speeds across a full NVL72 rack. Expect less on an 8-GPU server, on smaller models, on batch work and on BF16 training.
FP4 gains need your model converted to NVIDIA's 4-bit format, NVFP4. NVIDIA reports an accuracy loss of 1% or less on DeepSeek-R1, but test your own model.
For double-precision science (FP64), Rubin brings back what the B300 cut. Its native FP64 is slightly below the B200's (33 against 37 TF per GPU), but NVIDIA lists 200 TF per GPU for FP64 matrix maths (DGEMM) using emulation on the Tensor Cores. Test your own code before you count on that.
Our Hopper vs Blackwell guide covers how to match a GPU to a workload.
Vera Rubin release timeline: when it reaches buyers
For most buyers, from 2027. NVIDIA says production shipments began in August 2026, and Google Cloud plans general availability of its Rubin instances in November 2026, but early volume is limited and goes to the largest buyers first.
When | What happened, or is planned |
|---|---|
March 2025 | NVIDIA shows Vera Rubin at GTC, as "NVL144" |
January 2026 | At CES, NVIDIA says Vera Rubin is in full production, with partner products later in the year |
March 2026 | Microsoft is the first hyperscaler to power on a Vera Rubin NVL72, in its labs |
July 2026 | Early racks running at Google Cloud, Microsoft Azure, Oracle Cloud and two GPU clouds |
August 2026 | NVIDIA says production shipments have begun, and expects Vera Rubin to be about 20% of its data centre revenue in the following quarter |
September 2026 | First MLPerf results; a US GPU cloud opens limited access, by request, in one US zone |
October to December 2026 | Server makers list 8-GPU HGX Rubin NVL8 servers as available |
November 2026 | Google Cloud plans general availability of its Vera Rubin (A5X) instances |
Early 2027 | Reported NVIDIA price rises take effect, on Grace Blackwell and Vera Rubin systems |
Mid-2027 | A Saudi AI company plans 70 MW of Vera Rubin, subject to final agreements |
2027 | Rubin Ultra, on NVIDIA's roadmap (analysts report its largest rack slipping to 2028) |
Mid-2028 | First phase of a 240 MW Vera Rubin campus at Taloja, Navi Mumbai, for a US GPU cloud |
2028 | Feynman, the generation after Rubin, on NVIDIA's roadmap |
Supply stays tight
NVIDIA's finance chief said in August 2026 that supply will remain a bottleneck at least until the end of NVIDIA's 2028 financial year, in late January 2028.
TrendForce expects Blackwell to make up about 70% of NVIDIA's high-end GPU shipments in 2026. It sees the GB300 as the main driver until mid-2027, and Vera Rubin as the largest share during 2027.
Most of the large Rubin deployments announced outside the hyperscalers land between 2027 and mid-2028. An anonymous claim on X in October 2026 that Rubin shipments had been cut by half has not been confirmed by any reputable source.
For Blackwell, about 10 to 14 months passed between NVIDIA's announcement in March 2024 and instances at a US GPU cloud (February 2025) and at AWS (May 2025).
Rubin looks similar if you count from its CES launch in January 2026: limited, by-request access came about nine months later, and Google Cloud's planned general availability about ten months later (our arithmetic).
TrendForce's forecasts point to broader supply during 2027. We expect wider availability through our partner network in 2027. That is our estimate, not an NVIDIA date.
Your software
Existing CUDA code (software written for NVIDIA's GPU programming platform) should run. NVIDIA promises full backward compatibility, and its CUDA 13.4 release notes (October 2026) list Rubin support. Support in popular serving frameworks arrives as the platform ships, so check your stack's release notes.
The real work is FP4: to get the headline gains, models need converting to NVFP4 and checking for accuracy.
Your data centre
Rubin needs liquid cooling. Plan for about 24 kW per 8-GPU server, or 156 to 227 kW per NVL72 rack, with coolant at up to 45°C. NVIDIA says the racks can run on dry coolers, without chillers, in normal conditions up to about 38°C outside.
It also says chillers stay important in hot weather, and Schneider Electric says chillers usually remain necessary in hot, humid parts of India. And the bar is high.
In May 2025, one Indian operator said its sites in NVIDIA's DGX-Ready programme for liquid cooling support up to 200 kW per rack. That is below a Vera Rubin rack at full power.
Our India data centre power guide and guide to hosting GPUs in India cover what to check. For the rack's power shelves, coolant and footprint, see the Vera Rubin hardware deep dive.
Planning a GPU deployment for 2027? Register early interest in Rubin. It costs nothing, and we will come back with pricing and availability as our partners confirm allocation.
Still deciding whether to wait or rent Blackwell now? Our overview has a short guide to when to wait and when to rent now.
Where these figures come from
Figures are as of 24 October 2026. Ratios marked "our arithmetic" are calculated from these published figures.
- Specs: NVIDIA's Vera Rubin NVL72, HGX, NVLink, Vera CPU, DGX Rubin NVL8 and GB300 NVL72 pages, the Vera Rubin NVL72 datasheet (October 2026), the DGX Vera Rubin user guide, and NVIDIA's technical blog posts, including Inside the Rubin platform (January 2026) and its July 2026 update.
- Benchmarks: MLCommons MLPerf Inference v6.1 (16 September 2026; results tables) and SemiAnalysis.
- Prices: Morgan Stanley's estimate as reported by Tom's Hardware, Bernstein's as reported by Tiger Brokers, Reuters, Bloomberg, TrendForce, Silicon Data, a US GPU appraisal firm and AWS's published Capacity Blocks prices.
- Timeline and supply: NVIDIA's second-quarter results and results call of 26 August 2026 (third-party transcript), announcements by NVIDIA, Microsoft and Google Cloud, and TrendForce. The Kyber delay is SemiAnalysis's report, via Tom's Hardware.