Nvidia Data Center GPUs: Hopper, Blackwell and Vera Rubin
Nvidia announced the Hopper H100 in March 2022 and said it began shipping the GPU in November that year; Hopper was later extended with the H200. Nvidia introduced Blackwell in March 2024, but reports of a delay were followed by Nvidia’s disclosure that it had changed the GPU mask to improve yields before its fourth-quarter production ramp. Blackwell entered full production, then generated billions of dollars in its first reported sales quarter, while Blackwell Ultra ramped during 2025. Vera Rubin is now in production shipments and being deployed by system makers and cloud providers, but Nvidia’s descriptions leave unresolved timing and configuration differences, including whether Rubin has six or seven chips and how Rubin Ultra’s schedule relates to reports of a Kyber rack delay.
Key moments
- Mar 22, 2022Nvidia announces the Hopper architecture and H100 GPU at GTC 2022
- Nov 16, 2022NVIDIA begins shipping Hopper-based H100 GPUs
- Mar 18, 2024Blackwell platform announced at GTC 2024 with B200 and GB200
- Aug 5, 2024Blackwell shipments reported delayed to early 2025; Nvidia says production is on track
- Aug 28, 2024Nvidia discloses a Blackwell mask change to improve yields on its Q2 FY2025 call
- Feb 26, 2025First quarter of Blackwell sales: Nvidia cites billions in revenue, data center at $35.6 billion
- Mar 18, 2025Roadmap set: Vera Rubin NVL144 in H2 2026, Rubin Ultra in H2 2027
- May 31, 2026Vera Rubin ramping across the supply chain; production shipments set for fall 2026
- Aug 26, 2026Nvidia reports $96.2 billion quarter, data center $89.0 billion, Vera Rubin in full production
- Aug 27, 2026NVIDIA says Vera Rubin began production shipments in fiscal Q3 2027
- Sep 23, 2026Supermicro says it is shipping Vera Rubin NVL72 racks
- Sep 30, 2026CoreWeave makes Vera Rubin NVL72 available with Cognition workloads
In numbers
Data and sources (8 points)
| Date | Nvidia data center revenue in the quarter | Entry |
|---|---|---|
| Nov 20, 2024 | $30.8B | Nvidia says Blackwell is in full production; data center revenue hits $30.8 billion |
| Feb 26, 2025 | $35.6B | First quarter of Blackwell sales: Nvidia cites billions in revenue, data center at $35.6 billion |
| May 28, 2025 | $39.1B | NVIDIA reports $39.1 billion in Data Center revenue |
| Aug 27, 2025 | $41.1B | NVIDIA reports $41.1 billion in Data Center revenue |
| Nov 19, 2025 | $51.2B | NVIDIA reports $51.2 billion in Data Center revenue |
| Feb 25, 2026 | $62.3B | NVIDIA reports $62.3 billion in Data Center revenue |
| May 20, 2026 | $75.2B | NVIDIA reports $75.2 billion in Data Center revenue |
| Aug 26, 2026 | $89B | Nvidia reports $96.2 billion quarter, data center $89.0 billion, Vera Rubin in full production |
Timeline
Earlier years show only their entries' dates and titles. Click a year heading to show the details.
202614 entries
- Announcement
NVIDIA contributes Vera Rubin rack and compute-tray designs to OCP
Image: NVIDIA, press image · source · editorial use NVIDIA contributed Vera Rubin rack and compute-tray designs to the Open Compute Project through its open NVIDIA MGX architecture. NVIDIA said the contribution gives system builders a common foundation for next-generation AI-factory infrastructure.
- NVIDIA Blog · Oct 12, 2026 ·
NVIDIA contributed key NVIDIA Vera Rubin rack and compute tray designs to OCP through the open NVIDIA MGX architecture, giving system builders a common foundation for developing next-generation AI factory infrastructure.
Editor's note: The NVIDIA blog explicitly states that NVIDIA contributed Vera Rubin rack and compute-tray designs to OCP through open NVIDIA MGX and that this provides a common foundation for next-generation AI-factory infrastructure. The page text does not state the event date.
- NVIDIA Blog · Oct 12, 2026 ·
- Report
Cognition reports Vera Rubin SWE-2 throughput comparison
CoreWeave reported that Cognition measured a 4.8 times increase in total token throughput for SWE-2 inference workloads on Vera Rubin NVL72 compared with GB200 NVL72. NVIDIA reported that Cognition observed up to a 4.8 times increase in early tests.
- CoreWeave · Sep 30, 2026 ·
In independent benchmarks run on CoreWeave Cloud, Cognition measured a 4.8 times increase in total token throughput for its SWE-2 inference workloads on NVIDIA Vera Rubin NVL72 compared to a GB200 NVL72 baseline.
- NVIDIA Blog · Sep 30, 2026 ·
In its early tests, Cognition saw Vera Rubin NVL72 deliver up to a 4.8x increase in total token throughput for SWE-2 inference workloads over GB200 NVL72.
Editor's note: The page states that Cognition measured a 4.8 times increase in total token throughput for SWE-2 inference workloads on Vera Rubin NVL72 versus a GB200 NVL72 baseline, and separately attributes an "up to" 4.8 times result to Cognition. It is published September 30, 2026.
- CoreWeave · Sep 30, 2026 ·
- LaunchKey moment
CoreWeave makes Vera Rubin NVL72 available with Cognition workloads
Image: NVIDIA, press image · source · editorial use CoreWeave announced the availability of NVIDIA Vera Rubin NVL72 on its cloud, with Cognition as the first customer anywhere running production workloads on the system.
- CoreWeave · Sep 30, 2026 ·
CoreWeave Inc. (Nasdaq: CRWV), The Essential Cloud for AI™, today announced the availability of NVIDIA Vera Rubin NVL72 on CoreWeave, with Cognition as the first customer anywhere running production workloads on the system.
- NVIDIA Blog · Sep 30, 2026 ·
At CoreWeave Fully Connected, running this week in San Francisco, CoreWeave announced availability of NVIDIA Vera Rubin NVL72 systems with Spectrum-X 102.4T Ethernet networking.
Editor's note: Editor: dropped a sentence crediting NVIDIA with a Spectrum-X 102.4T networking detail the announcement does not carry.
- CoreWeave · Sep 30, 2026 ·
- ShipmentKey moment
Supermicro says it is shipping Vera Rubin NVL72 racks
Supermicro said it was shipping NVIDIA Vera Rubin NVL72 racks. It said the racks were integrated with its Data Center Building Block Solutions and DLC-2 direct-liquid-cooling stack.
- Supermicro · Sep 23, 2026 ·
“Super Micro Computer, Inc. (NASDAQ: SMCI),an AI, Enterprise, Storage, and 5G/Edge IT Total Solution Provider, is now shipping NVIDIA Vera Rubin NVL72 racks integrated with Supermicro's Data Center Building Block Solutions®(DCBBS) and its direct liquid cooling stack (DLC-2).”
Editor's note: The page states that Supermicro is now shipping NVIDIA Vera Rubin NVL72 racks integrated with its Data Center Building Block Solutions (DCBBS) and DLC-2 direct liquid cooling stack. The page’s PR Newswire metadata indicates September 23, 2026.
- Supermicro · Sep 23, 2026 ·
- Shipment
CoreWeave brings up a multi-rack Vera Rubin NVL72 cluster
CoreWeave announced the bring-up of a multi-rack NVIDIA Vera Rubin NVL72 system on CoreWeave Cloud. The cluster connected hundreds of NVIDIA Rubin GPUs for agentic AI.
- CoreWeave · Sep 16, 2026 ·
CoreWeave, Inc. (Nasdaq: CRWV), The Essential Cloud for AI™, today announced the bring up of multi-rack NVIDIA Vera Rubin NVL72 on CoreWeave Cloud, putting hundreds of NVIDIA Rubin GPUs into a single scale-out cluster for agentic AI.
Editor's note: The page states that CoreWeave announced the bring up of multi-rack NVIDIA Vera Rubin NVL72 on CoreWeave Cloud, connecting hundreds of Rubin GPUs in a scale-out cluster for agentic AI. It is dated September 16, 2026.
- CoreWeave · Sep 16, 2026 ·
- Report
MLCommons releases MLPerf Inference v6.1 results with Rubin previews
MLCommons released MLPerf Inference v6.1 results listing NVIDIA Rubin and NVIDIA Vera Rubin NVL72 as preview platforms. NVIDIA said its first MLPerf Inference preview submission for Vera Rubin NVL72 delivered up to 3.7 times higher throughput than GB300 NVL72.
- MLCommons · Sep 16, 2026 ·
The results include five new processors or accelerators: AMD Ryzen AI Max+ 395, AMD Instinct MI350P, and Intel Arc Pro B70 are all available, and NVIDIA Rubin and NVIDIA Vera Rubin NVL72 is in preview.
- NVIDIA Blog · Sep 16, 2026 ·
In its first MLPerf Inference preview submission, the NVIDIA Vera Rubin NVL72 system delivered up to 3.7x higher throughput than NVIDIA GB300 NVL72, demonstrating the performance gains possible with NVIDIA’s next-generation AI infrastructure.
- NVIDIA Blog · Sep 16, 2026 ·
“NVIDIA Vera Rubin NVL72 system debuts with leading performance: In its first MLPerf Inference preview submission, NVIDIA Vera Rubin NVL72 delivers up to 3.7x better throughput than GB300 NVL72.”
Editor's note: Both cited pages are readable. MLCommons states that NVIDIA Rubin and NVIDIA Vera Rubin NVL72 are in preview, and NVIDIA states that its first preview submission for Vera Rubin NVL72 delivered up to 3.7x higher throughput than GB300 NVL72.
- MLCommons · Sep 16, 2026 ·
- RegulatoryKey moment
NVIDIA says Vera Rubin began production shipments in fiscal Q3 2027
NVIDIA stated that Vera Rubin began production shipments in the third quarter of fiscal 2027. The filing described Vera Rubin as its next-generation Data Center architecture.
- NVIDIA Form 10-Q · Aug 27, 2026 ·
Our next-generation Data Center architecture, Vera Rubin, began production shipments in the third quarter of fiscal year 2027.
contradicts Vera Rubin ramping across the supply chain; production shipments set for fall 2026.
Editor's note: The filing states that Vera Rubin, NVIDIA’s next-generation Data Center architecture, began production shipments in the third quarter of fiscal year 2027. The filing supports the node date and does not give a different event date.
- NVIDIA Form 10-Q · Aug 27, 2026 ·
- FinancialKey moment
Nvidia reports $96.2 billion quarter, data center $89.0 billion, Vera Rubin in full production
Second-quarter fiscal 2027 results showed revenue of $96.2 billion, up 106% year on year, with data center revenue of $89.0 billion. Jensen Huang said Vera Rubin was in full production, and the release noted racks running at partners.
- Nvidia Newsroom · Aug 26, 2026 ·
Vera Rubin, now in full production, was built to power exactly this moment.
- Nvidia Newsroom · Aug 26, 2026 ·
- Response
NVIDIA says its roadmap is intact
NVIDIA said, “Our roadmap is intact.” Tom’s Hardware published the statement in an article about the Rubin Ultra Kyber NVL144 rack.
- Tom’s Hardware · Jul 6, 2026 ·
“Our roadmap is intact.”
contradicts TrendForce: HBM4 mass production slips to the end of the first quarter of 2026 after Nvidia raised the spec on HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4E.
Editor's note: The article’s update dated July 6, 2026 says an Nvidia representative told Tom’s Hardware, “Our roadmap is intact.” It appears in an article about the reported delay of the Kyber NVL144 rack for Rubin Ultra.
- Tom’s Hardware · Jul 6, 2026 ·
- AnnouncementKey moment
Vera Rubin ramping across the supply chain; production shipments set for fall 2026
At GTC Taipei ahead of Computex, Nvidia said hundreds of supply chain partners across more than 350 factories in 30 countries were ramping Vera Rubin, with production shipments to begin in the fall.
- Nvidia Newsroom · May 31, 2026 ·
Production shipments of Vera Rubin are set to begin starting this fall.
is contradicted by NVIDIA says Vera Rubin began production shipments in fiscal Q3 2027.
- Nvidia Newsroom · May 31, 2026 ·
- Financial
NVIDIA reports $75.2 billion in Data Center revenue
NVIDIA reported $75.2 billion in Data Center revenue for the first quarter of fiscal 2027. It said the revenue was up 21% from the prior quarter and 92% from a year earlier.
- NVIDIA · May 20, 2026 ·
First-quarter revenue was a record $75.2 billion, up 21% from the previous quarter and up 92% from a year ago.
Editor's note: The cited NVIDIA release states that first-quarter fiscal 2027 Data Center revenue was $75.2 billion, up 21% sequentially and 92% year over year. It is dated May 20, 2026.
- NVIDIA · May 20, 2026 ·
- Announcement
Vera Rubin declared in full production at GTC 2026
Image: NVIDIA, press image · source · editorial use Nvidia announced that the seven chips of the Vera Rubin platform were in full production and described the platform as five racks and one supercomputer. This kept to the second-half-2026 timing set a year earlier.
- Nvidia Newsroom · Mar 16, 2026 ·
Vera Rubin is a generational leap — seven breakthrough chips, five racks, one giant supercomputer — built to power every phase of AI
- Nvidia Newsroom · Mar 16, 2026 ·
- Financial
NVIDIA reports $62.3 billion in Data Center revenue
NVIDIA reported $62.3 billion in Data Center revenue for the fourth quarter of fiscal 2026. It said the revenue was up 22% from the prior quarter and 75% from a year earlier.
- NVIDIA · Feb 25, 2026 ·
Fourth-quarter revenue was a record $62.3 billion, up 22% from the previous quarter and up 75% from a year ago, driven by the major platform shifts — accelerated computing and AI.
Editor's note: The cited NVIDIA press release, dated February 25, 2026, reports fourth-quarter fiscal 2026 Data Center revenue of $62.3 billion, up 22% quarter over quarter and 75% year over year. The cited quote appears in the release’s Data Center highlights.
- NVIDIA · Feb 25, 2026 ·
- Financial
NVIDIA unveils the six-chip Rubin platform
NVIDIA unveiled the Rubin platform, which it said comprised six chips. It said AWS, Google Cloud, Microsoft Azure and Oracle Cloud Infrastructure would be among the first to deploy Vera Rubin-based instances.
- NVIDIA · Feb 25, 2026 ·
Unveiled the NVIDIA Rubin platform, comprising six new chips to deliver up to a 10x reduction in inference token cost, compared with the NVIDIA Blackwell platform; cloud providers Amazon Web Services (AWS), Google Cloud, Microsoft Azure and Oracle Cloud Infrastructure will be among the first to deploy Vera Rubin-based instances.
Editor's note: The page is readable and dated February 25, 2026. Its Data Center highlights state that NVIDIA unveiled the Rubin platform comprising six new chips and name AWS, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure as early deployers of Vera Rubin-based instances.
- NVIDIA · Feb 25, 2026 ·
20259 entries
- Financial
NVIDIA reports $51.2 billion in Data Center revenue
NVIDIA reported $51.2 billion in Data Center revenue for the third quarter of fiscal 2026. It said the revenue was up 25% from the prior quarter and 66% from a year earlier.
- NVIDIA · Nov 19, 2025 ·
Third-quarter revenue was a record $51.2 billion, up 25% from the previous quarter and up 66% from a year ago.
Editor's note: The cited NVIDIA press release, dated November 19, 2025, states that third-quarter Data Center revenue was a record $51.2 billion, up 25% from the previous quarter and 66% from a year earlier.
- NVIDIA · Nov 19, 2025 ·
- Financial
NVIDIA says Blackwell sales and cloud GPUs are sold out
NVIDIA reported Blackwell sales. It said cloud GPUs were sold out.
- NVIDIA · Nov 19, 2025 ·
Blackwell sales are off the charts, and cloud GPUs are sold out.
Editor's note: The page is readable and dated November 19, 2025. It reports Blackwell sales and states that cloud GPUs are sold out, using the quoted wording.
- NVIDIA · Nov 19, 2025 ·
- Financial
NVIDIA reports $41.1 billion in Data Center revenue
NVIDIA reported $41.1 billion in Data Center revenue for the second quarter of fiscal 2026. It said the revenue was up 5% from the prior quarter and 56% from a year earlier.
- NVIDIA · Aug 27, 2025 ·
Second-quarter revenue was $41.1 billion, up 5% from the previous quarter and up 56% from a year ago.
Editor's note: The cited NVIDIA press release, dated August 27, 2025, states that second-quarter Data Center revenue was $41.1 billion, up 5% from the previous quarter and 56% from a year earlier.
- NVIDIA · Aug 27, 2025 ·
- Financial
NVIDIA says Blackwell Ultra production is ramping
NVIDIA said production of Blackwell Ultra was ramping at full speed. It identified Blackwell as its AI platform.
- NVIDIA · Aug 27, 2025 ·
Blackwell is the AI platform the world has been waiting for, delivering an exceptional generational leap — production of Blackwell Ultra is ramping at full speed, and demand is extraordinary.
Editor's note: The page is readable and is dated August 27, 2025. It contains the quoted statement verbatim and identifies Blackwell as the AI platform while stating that Blackwell Ultra production is ramping at full speed.
- NVIDIA · Aug 27, 2025 ·
- Financial
NVIDIA reports $39.1 billion in Data Center revenue
NVIDIA reported $39.1 billion in Data Center revenue for the first quarter of fiscal 2026. It said the revenue was up 10% from the prior quarter and 73% from a year earlier.
- NVIDIA · May 28, 2025 ·
First-quarter revenue was $39.1 billion, up 10% from the previous quarter and up 73% from a year ago.
Editor's note: The cited NVIDIA press release, dated May 28, 2025, states that first-quarter fiscal 2026 Data Center revenue was $39.1 billion, up 10% from the previous quarter and 73% from a year earlier. The cited quote appears on the page.
- NVIDIA · May 28, 2025 ·
- Shipment
NVIDIA says Blackwell NVL72 is in full-scale production
NVIDIA said its Blackwell NVL72 AI supercomputer was in full-scale production. It said production covered system makers and cloud service providers.
- NVIDIA · May 28, 2025 ·
Our breakthrough Blackwell NVL72 AI supercomputer — a ‘thinking machine’ designed for reasoning— is now in full-scale production across system makers and cloud service providers.
Editor's note: The page, dated May 28, 2025, states that NVIDIA's Blackwell NVL72 AI supercomputer is in full-scale production across system makers and cloud service providers. The quoted statement appears on the page.
- NVIDIA · May 28, 2025 ·
- AnnouncementKey moment
Roadmap set: Vera Rubin NVL144 in H2 2026, Rubin Ultra in H2 2027
Image: NVIDIA, press image · source · editorial use In the GTC 2025 keynote, Nvidia dated the next two generations. Vera Rubin NVL144 systems were slated for the second half of 2026 and Rubin Ultra systems for the second half of 2027.
- Nvidia Blog · Mar 18, 2025 ·
Systems built on Rubin Ultra, including the Vera Rubin NVL 144, will arrive in the second half of next year. And due for the second half of 2027: systems built on Rubin Ultra.
- Nvidia Blog · Mar 18, 2025 ·
- Announcement
Blackwell Ultra announced at GTC 2025 for the second half of 2025
Nvidia announced Blackwell Ultra and the GB300 NVL72 rack, claiming 1.5x the AI performance of the GB200 NVL72. Partner products were expected from the second half of 2025.
- Nvidia Investor Relations · Mar 18, 2025 ·
Blackwell Ultra-based products are expected to be available from partners starting from the second half of 2025.
- Nvidia Investor Relations · Mar 18, 2025 ·
- FinancialKey moment
First quarter of Blackwell sales: Nvidia cites billions in revenue, data center at $35.6 billion
Fourth-quarter fiscal 2025 results reported data center revenue of $35.6 billion. Nvidia said Blackwell AI supercomputers achieved billions of dollars in sales in their first quarter, without giving an exact figure in the release.
- Nvidia Newsroom · Feb 26, 2025 ·
Fourth-quarter revenue was a record $35.6 billion, up 16% from the previous quarter and up 93% from a year ago.
- Nvidia Newsroom · Feb 26, 2025 ·
- Nov 19, 2025NVIDIA reports $51.2 billion in Data Center revenue
- Nov 19, 2025NVIDIA says Blackwell sales and cloud GPUs are sold out
- Aug 27, 2025NVIDIA reports $41.1 billion in Data Center revenue
- Aug 27, 2025NVIDIA says Blackwell Ultra production is ramping
- May 28, 2025NVIDIA reports $39.1 billion in Data Center revenue
- May 28, 2025NVIDIA says Blackwell NVL72 is in full-scale production
- Mar 18, 2025Roadmap set: Vera Rubin NVL144 in H2 2026, Rubin Ultra in H2 2027
- Mar 18, 2025Blackwell Ultra announced at GTC 2025 for the second half of 2025
- Feb 26, 2025First quarter of Blackwell sales: Nvidia cites billions in revenue, data center at $35.6 billion
20247 entries
- Financial
Nvidia says Blackwell is in full production; data center revenue hits $30.8 billion
In its third-quarter fiscal 2025 results, Nvidia described Blackwell as being in full production and reported record data center revenue of $30.8 billion, up 17% from the prior quarter.
- Nvidia Newsroom · Nov 20, 2024 ·
Third-quarter revenue was a record $30.8 billion, up 17% from the previous quarter
- Nvidia Newsroom · Nov 20, 2024 ·
- Shipment
NVIDIA says Blackwell samples are shipping
NVIDIA said Blackwell samples were shipping to partners and customers. The company identified the products as Blackwell samples.
- NVIDIA · Aug 28, 2024 ·
Blackwell samples are shipping to our partners and customers.
Editor's note: The page is readable and dated August 28, 2024. It states, “Blackwell samples are shipping to our partners and customers,” supporting the summary as written.
- NVIDIA · Aug 28, 2024 ·
- ResponseKey moment
Nvidia discloses a Blackwell mask change to improve yields on its Q2 FY2025 call
On the earnings call, CFO Colette Kress said Nvidia had changed the Blackwell GPU mask to improve production yields and that the production ramp would begin in the fourth quarter. Jensen Huang said Nvidia meant to ship product, not merely start production, in Q4. Nvidia also said Blackwell demand was well above supply.
- The Register · Aug 29, 2024 ·
We executed a change to the Blackwell GPU mask to improve production yields.
responded to Blackwell shipments reported delayed to early 2025; Nvidia says production is on track. led to TrendForce: HBM4 mass production slips to the end of the first quarter of 2026 after Nvidia raised the spec on HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4E.
- The Register · Aug 29, 2024 ·
- ReportKey moment
Blackwell shipments reported delayed to early 2025; Nvidia says production is on track
Citing The Information, The Register reported that Nvidia had told Microsoft of delays to the most advanced Blackwell parts, with volume shipments pushed to the first quarter of 2025 and TSMC's CoWoS packaging suspected as the cause. Nvidia's statement did not confirm a delay: it said sampling had started and production was on track to ramp in the second half.
- The Register · Aug 5, 2024 ·
As we've stated before, Hopper demand is very strong, broad Blackwell sampling has started, and production is on track to ramp in 2H. Beyond that, we don't comment on rumors.
was answered by Nvidia discloses a Blackwell mask change to improve yields on its Q2 FY2025 call.
- The Register · Aug 5, 2024 ·
- Announcement
Rubin platform and Vera CPU revealed ahead of Computex 2024
Image: NVIDIA, press image · source · editorial use In his Computex keynote in Taipei, Jensen Huang showed the Rubin platform as Blackwell's successor, together with an Arm-based CPU called Vera and new networking parts. He described a one-year release rhythm for the whole data center stack. No launch year was given in Nvidia's own write-up.
- Nvidia Blog · Jun 2, 2024 ·
Revealed for the first time, the Rubin platform will succeed the upcoming Blackwell platform, featuring new GPUs, a new Arm-based CPU — Vera — and advanced networking with NVLink 6, CX9 SuperNIC and the X1600 converged InfiniBand/Ethernet switch.
- Nvidia Blog · Jun 2, 2024 ·
- Financial
NVIDIA says Blackwell is in full production
NVIDIA said the Blackwell platform was in full production. It said the platform formed the foundation for trillion-parameter-scale generative AI.
- NVIDIA · May 22, 2024 ·
We are poised for our next wave of growth. The Blackwell platform is in full production and forms the foundation for trillion-parameter-scale generative AI.
Editor's note: The page is readable and dated May 22, 2024. It states verbatim that the Blackwell platform was in full production and formed the foundation for trillion-parameter-scale generative AI.
- NVIDIA · May 22, 2024 ·
- AnnouncementKey moment
Blackwell platform announced at GTC 2024 with B200 and GB200
Nvidia announced the Blackwell architecture in San Jose, listing the B200 Tensor Core GPU and the GB200 Grace Blackwell Superchip, which pairs two B200 GPUs with a Grace CPU. The release claims up to 25x lower cost and energy for LLM inference compared with H100.
- Nvidia Investor Relations · Mar 18, 2024 ·
Packed with 208 billion transistors, Blackwell-architecture GPUs are manufactured
- Nvidia Investor Relations · Mar 18, 2024 ·
- Nov 20, 2024Nvidia says Blackwell is in full production; data center revenue hits $30.8 billion
- Aug 28, 2024NVIDIA says Blackwell samples are shipping
- Aug 28, 2024Nvidia discloses a Blackwell mask change to improve yields on its Q2 FY2025 call
- Aug 5, 2024Blackwell shipments reported delayed to early 2025; Nvidia says production is on track
- Jun 2, 2024Rubin platform and Vera CPU revealed ahead of Computex 2024
- May 22, 2024NVIDIA says Blackwell is in full production
- Mar 18, 2024Blackwell platform announced at GTC 2024 with B200 and GB200
20231 entry
- Announcement
Nvidia announces the H200, the first GPU with HBM3e memory
The H200 kept the Hopper architecture but moved to HBM3e memory. Nvidia said systems from server makers and cloud providers would begin shipping in the second quarter of 2024.
- Nvidia Newsroom · Nov 13, 2023 ·
With HBM3e, the NVIDIA H200 delivers 141GB of memory at 4.8 terabytes per second, nearly double the capacity and 2.4x more bandwidth compared with its predecessor.
led to Micron starts volume production of HBM3E for Nvidia's H200 on HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4E.
- Nvidia Newsroom · Nov 13, 2023 ·
20222 entries
- ShipmentKey moment
NVIDIA begins shipping Hopper-based H100 GPUs
Image: NVIDIA, press image · source · editorial use NVIDIA said it began shipping the Hopper-based H100 Tensor Core GPU. It said the first H100 systems were available.
- NVIDIA Newsroom · Nov 16, 2022 ·
Began shipping the NVIDIA® H100 Tensor Core GPU based on the new NVIDIA Hopper™ architecture, with first systems available now.
led to Data-center revenue of $10.32 billion, up 171 percent in a year on How Nvidia Came to Own AI Compute: From CUDA to the H100 Shortage.
Editor's note: The page is dated November 16, 2022, and its Data Center highlights state that NVIDIA began shipping the H100 Tensor Core GPU based on Hopper, with first systems available now.
- NVIDIA Newsroom · Nov 16, 2022 ·
- AnnouncementKey moment
Nvidia announces the Hopper architecture and H100 GPU at GTC 2022
Image: NVIDIA, press image · source · editorial use Nvidia introduced Hopper, the successor to Ampere, with the H100 as its first GPU. The press release cites 80 billion transistors on a TSMC 4N process and says the H100 will be available in the third quarter.
- Nvidia Newsroom · Mar 22, 2022 ·
NVIDIA H100 will be available starting in the third quarter.
result of A100 in full production: Ampere unifies training and inference, adopted by every major cloud on How Nvidia Came to Own AI Compute: From CUDA to the H100 Shortage.
- Nvidia Newsroom · Mar 22, 2022 ·
Questions this timeline answers
Each one is an ordered set of dated entries, most of them from this page, with a line on why each step led to the next.
Cause and effect
- Blackwell shipments reported delayed to early 2025; Nvidia says production is on trackwas answered by →Nvidia discloses a Blackwell mask change to improve yields on its Q2 FY2025 callThree weeks after saying production was on track, Nvidia acknowledged a mask change to fix Blackwell yields.
- Vera Rubin ramping across the supply chain; production shipments set for fall 2026is contradicted by →NVIDIA says Vera Rubin began production shipments in fiscal Q3 2027The existing node said Vera Rubin production shipments were set for fall 2026, while NVIDIA's August 2026 filing said shipments had begun in fiscal Q3 2027.
- A100 in full production: Ampere unifies training and inference, adopted by every major cloud on How Nvidia Came to Own AI Compute: From CUDA to the H100 Shortageled to →Nvidia announces the Hopper architecture and H100 GPU at GTC 2022The A100's widespread cloud and server-maker adoption created the installed base into which Timeline B says the H100 launched.
- NVIDIA begins shipping Hopper-based H100 GPUsled to →Data-center revenue of $10.32 billion, up 171 percent in a year on How Nvidia Came to Own AI Compute: From CUDA to the H100 ShortageNVIDIA's start of Hopper-based H100 shipments preceded Nvidia's reported data-center revenue surge, which the later entry says came as H100 systems shipped to cloud providers and consumer-internet companies.
- Nvidia announces the H200, the first GPU with HBM3e memoryled to →Micron starts volume production of HBM3E for Nvidia's H200 on HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4ENvidia announced that H200 would use HBM3e and ship in the second quarter of 2024, and Micron subsequently began HBM3E production for shipment inside H200 GPUs in that quarter.
- Nvidia discloses a Blackwell mask change to improve yields on its Q2 FY2025 callled to →TrendForce: HBM4 mass production slips to the end of the first quarter of 2026 after Nvidia raised the spec on HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4EAfter Nvidia said Blackwell demand was well above supply, TrendForce said strong Blackwell demand led Nvidia to adjust the Rubin production timeline, contributing to the HBM4 production delay.
- TrendForce: HBM4 mass production slips to the end of the first quarter of 2026 after Nvidia raised the spec on HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4Eis contradicted by →NVIDIA says its roadmap is intactTrendForce said Nvidia had adjusted the Rubin production timeline because of strong Blackwell demand, whereas NVIDIA later said its roadmap was intact.
Where sources disagree or fall silent
- The extent of the 2024 Blackwell delay remains unclear: press reports said volume shipments slipped to the first quarter of 2025, while Nvidia initially said production was on track and later confirmed only a mask change and a fourth-quarter ramp.
- Nvidia’s February 2026 results described Rubin as a six-chip platform, while its March 2026 announcement described Vera Rubin as seven chips.
- Nvidia said in May 2026 that Vera Rubin production shipments would begin in the fall, but its August filing said they had already begun in fiscal Q3 2027; the precise shipment start is not clear from these statements.
- Nvidia’s reported Data Center revenue is aggregate revenue and does not separately quantify revenue from Hopper, Blackwell, Blackwell Ultra, or Vera Rubin.
- Nvidia has said its roadmap is intact after reporting that linked Rubin Ultra Kyber NVL144 to 2028, but the available statements do not clarify whether that report changes the second-half-2027 Rubin Ultra systems schedule.
- The published nodes contain no dated product or deployment details for Feynman.
What to watch next
- Mar 15, 2027 — GTC 2027 in San Jose Nvidia’s annual conference is a scheduled venue for further details on Rubin Ultra and the Feynman roadmap.
- Jul 2027 — Rubin Ultra enters Nvidia’s stated second-half-2027 window Nvidia has scheduled Rubin Ultra systems for the second half of 2027, making this the next disclosed platform transition after Vera Rubin.
Who is involved
- Nvidiacompany
Designer and supplier of the Hopper, Blackwell and Rubin data-center computing platforms.
- Blackwellproduct
Nvidia's successor data-center architecture, whose 2024 yield issue preceded full production and commercial sales.
- Jensen Huangperson
Nvidia chief executive who presented the platform roadmaps and described Vera Rubin’s production status.
- Rubinproduct
2026 generation (Vera Rubin NVL144), with Rubin Ultra due in 2027.
- Vera Rubin NVL72product
NVIDIA rack platform received by CoreWeave.
- Hopperproduct
Nvidia's data-center GPU architecture introduced in 2022, led by the H100 and later refreshed with the H200.
- TSMCcompany
Nvidia’s manufacturing partner; reporting cited its CoWoS packaging as a suspected factor in the reported 2024 Blackwell delay.
- CoreWeavecompany
Cloud provider that brought up a multi-rack Vera Rubin NVL72 cluster and made the system available for production workloads.
29 more
- Veraproduct
Nvidia’s Arm-based CPU for the Vera Rubin platform.
- Vera Rubinproduct
Nvidia’s next data-center platform, combining Rubin GPUs with the Arm-based Vera CPU and now in production deployment.
- Cognitioncompany
CoreWeave’s first stated production customer for Vera Rubin NVL72 workloads.
- Microsoftcompany
Was identified in press reporting as a customer told of delayed advanced Blackwell shipments and is also named as an intended early Vera Rubin cloud deployer through Azure.
- Rubin GPUproduct
NVIDIA GPU used in the reported benchmark and CoreWeave cluster.
- Spectrum-X Ethernetproduct
NVIDIA networking product associated with the CoreWeave deployment.
- Amazon Web Servicescompany
Cloud provider NVIDIA named as an intended early deployer of Vera Rubin-based instances.
- Blackwell Ultraproduct
Blackwell refresh announced for the second half of 2025 and reported as ramping in production by August 2025.
- Colette Kressperson
Nvidia chief financial officer who disclosed the Blackwell mask change intended to improve production yields.
- Data Center Building Block Solutionsproduct
Supermicro infrastructure integrated into its Vera Rubin NVL72 racks.
- DLC-2product
Supermicro direct-liquid-cooling stack integrated into its Vera Rubin NVL72 racks.
- GB200 NVL72product
System used as the baseline in Cognition's SWE-2 throughput comparison.
- GB300 NVL72product
Blackwell Ultra rack-scale system Nvidia positioned as the successor to GB200 NVL72.
- Google Cloudcompany
Cloud provider identified by Nvidia as an early planned deployer of Vera Rubin-based instances.
- H100product
The first Hopper GPU, announced in March 2022 and reported as shipping in November 2022.
- Kyber NVL144product
Rack configuration discussed in the Tom’s Hardware article about the Rubin Ultra roadmap.
- MLCommonscompany
Industry consortium that listed Rubin and Vera Rubin NVL72 as preview platforms in MLPerf Inference v6.1.
- MLPerf Inferenceproduct
Benchmark suite in which Rubin and Vera Rubin NVL72 were listed as preview systems.
- NVIDIA MGXproduct
Open architecture through which NVIDIA contributed Vera Rubin designs to OCP.
- NVL72product
Blackwell AI supercomputer that NVIDIA said was in full-scale production.
- Open Compute Projectother
Open hardware organization to which Nvidia contributed Vera Rubin rack and compute-tray designs.
- Oraclecompany
Cloud provider NVIDIA named as an intended early deployer of Vera Rubin-based instances.
- Rubin Ultraproduct
The platform scheduled by Nvidia for the second half of 2027; reporting has raised questions about timing for its Kyber NVL144 rack.
- SemiAnalysiscompany
Analyst firm attributed by Tom’s Hardware with the reported Kyber NVL144 schedule claim.
- Supermicrocompany
System maker that said it began shipping Vera Rubin NVL72 racks in September 2026.
- SWE-2product
Cognition software-engineering inference workload used in the throughput comparison.
- The Informationother
Outlet whose August 2024 report of a Blackwell delay triggered Nvidia's statements.
- Tom’s Hardwarecompany
Published the articles reporting NVIDIA’s statements.
- Vera CPUproduct
Arm-based CPU announced with the Rubin platform and used in the Vera Rubin system.
Changelog
- Added 1 entry: 2026-10-12 NVIDIA contributes Vera Rubin rack and compute-tray designs to OCP.
- Added 6 entries: 2025-05-28 NVIDIA reports $39.1 billion in Data Center revenue; 2025-08-27 NVIDIA reports $41.1 billion in Data Center revenue; 2025-11-19 NVIDIA reports $51.2 billion in Data Center revenue; 2026-02-25 NVIDIA reports $62.3 billion in Data Center revenue and 2 more. Linked 1 entry with HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4E.
- Added 1 entry: 2026-09-16 MLCommons releases MLPerf Inference v6.1 results with Rubin previews.
- Linked 1 entry with HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4E. Added 2 entries: 2026-09-30 CoreWeave makes Vera Rubin NVL72 available with Cognition workloads; 2026-09-30 Cognition reports Vera Rubin SWE-2 throughput comparison.
- Added 1 entry: 2026-09-23 Supermicro says it is shipping Vera Rubin NVL72 racks.
- Linked 1 entry with How Nvidia Came to Own AI Compute: From CUDA to the H100 Shortage. Linked 1 entry with HBM Supply Race: SK hynix, Samsung and Micron from HBM3E to HBM4E. Added 1 entry: 2026-09-16 CoreWeave brings up a multi-rack Vera Rubin NVL72 cluster.
- Linked 1 entry with How Nvidia Came to Own AI Compute: From CUDA to the H100 Shortage.
- Added 1 entry: 2026-07-06 NVIDIA says its roadmap is intact.
- Added 8 entries: 2022-11-16 NVIDIA begins shipping Hopper-based H100 GPUs; 2024-05-22 NVIDIA says Blackwell is in full production; 2024-08-28 NVIDIA says Blackwell samples are shipping; 2025-05-28 NVIDIA says Blackwell NVL72 is in full-scale production and 4 more. Added 13 hand-verified entries by the editor.
A research pass records each dated fact with the page it came from and a verbatim quote. A second, independent pass re-opens every cited page and rejects entries it cannot confirm. The editor reviews what remains before it is published, and every change is listed above. Full method.