Companies
Partnership
AWS and NVIDIA plan 2 million additional GPUs for 2027–2028
The August 26 NVIDIA Newsroom release is a multi-year capacity plan, not the fiscal Q2 2027 8-K. Official results remain on the sourced recap. This page does not invent conference-call Q&A.
TickerGrove Newsroom · Published 08-26-2026 · NVIDIA Corporation and Amazon Web Services
Sources
NVIDIA Newsroom, “AWS and NVIDIA to Deliver 2 Million Additional GPUs and Next-Generation Infrastructure for Agentic and Physical AI,” August 26, 2026, independently re-read the same evening. Fiscal Q2 2027 figures cited for contrast are from Exhibit 99.1 to 8-K accession 0001045810-26-000073 and Exhibit 99.2, already recapped at /stories/nvidia-q2-fy2027-results.
Amazon Web Services and NVIDIA said on August 26, 2026 that they plan to deploy 2 million additional NVIDIA GPUs across AWS’s global infrastructure in 2027 and 2028. NVIDIA’s Newsroom release names Blackwell Ultra, Rubin, and Rubin Ultra as the additional mix. That statement is a capacity plan. It is not fiscal Q2 2027 revenue, EPS, or the Q3 outlook furnished with Form 8-K accession 0001045810-26-000073.
What changed
The same release says that at NVIDIA GTC 2026, AWS announced plans to add more than 1 million NVIDIA GPUs starting in 2026, and that demand has since exceeded those expectations. The August 26 add-on is the extra 2 million GPUs in 2027–2028, not a restatement of the 8-K print recapped at /stories/nvidia-q2-fy2027-results.
The 2 million figure is incremental and dated. NVIDIA said AWS plans to deploy an additional 2 million NVIDIA Blackwell Ultra, Rubin, and Rubin Ultra GPUs in 2027–2028 across AWS Global Infrastructure, including AI factories. That sits on top of the GTC 2026 plan of more than 1 million GPUs starting in 2026, which NVIDIA said demand has exceeded. Jensen Huang, in the same release, said demand is running ahead of every forecast. Those are issuer words from the Newsroom copy, not from a TickerGrove transcript of the 5:00 p.m. ET call.
The release also lists work beyond GPU count. AWS and NVIDIA said they are working to bring NVIDIA Vera CPU-based infrastructure to AWS, to extend NVIDIA NVLink Fusion with custom high-bandwidth memory branded NVHBM, and to build AI factories for the U.S. government, including plans to deliver 100,000 GPUs on secure AWS infrastructure for federal and national-security workloads classified at Impact Level 6 and above. Matt Garman, CEO of AWS, said the expanded collaboration is meant to give frontier labs, enterprises, and governments more ways to build and deploy AI on AWS. Those lines are in the August 26 release; they are not booked NVIDIA revenue.
Keep the objects separate. Fiscal Q2 2027 GAAP revenue of $96.221 billion and Data Center revenue of $89.023 billion belong on /stories/nvidia-q2-fy2027-results from the 8-K and 10-Q. This page is the same-day Newsroom capacity plan. TickerGrove has not independently verified conference-call Q&A, a fiscal 2028 growth rate, or hyperscaler capex commentary that is not in this release.
A GPU deployment plan is not a reported quarter
A company can say it plans to install more chips in future years without that number appearing in the quarter it just reported. The 2 million additional GPUs are dated 2027–2028. The $96.221 billion is the quarter ended July 26, 2026.
Treat 2027–2028 capacity as a forward-looking arrangement
The release is explicit that the extra 2 million GPUs are planned for 2027–2028 and that the GTC 2026 “more than 1 million starting in 2026” plan is the prior object. The 100,000 federal GPUs are another planned delivery, not Q2 shipments. Do not fold any of those counts into the August 26 8-K table.
What we do not know
This article does not invent conference-call Q&A, a share-price reaction, a consensus comparison, or a recommendation. It does not convert the 2 million additional GPUs into NVIDIA revenue, a booking, or a shipment schedule inside fiscal Q2 2027. Huang’s and Garman’s sentences are from the Newsroom release. Secondary live blogs are not a source here.
Related Learn
Corrections
We do not silently rewrite a published line. A correction is a new labeled piece, not an invisible edit.
Discuss this story. Join the TickerGrove community to talk companies, earnings, and markets, or request future coverage.
Sources & evidence
Primary documents used for this piece. Internal claim-lineage notes stay off this page.
Amazon Web Services and NVIDIA
NVIDIA Newsroom AWS expansion, August 26, 2026
Planned deployment in 2027–2028, announced August 26, 2026
NVIDIA Corporation
NVIDIA Q2 FY2027 Exhibit 99.1, August 26, 2026
Fiscal Q2 2027 ended July 26, 2026
NVIDIA Data Center
NVIDIA Q2 FY2027 Exhibit 99.2, August 26, 2026
Fiscal Q2 2027 ended July 26, 2026
Figures used in this article
GPU counts are from the August 26, 2026 NVIDIA Newsroom release. The $96.221 billion and $89.023 billion figures are the separate fiscal Q2 2027 8-K recap, not AWS contract values.
Figure
2 million additional GPUs
- Entity
- Amazon Web Services and NVIDIA
- Period / as-of
- Planned deployment in 2027–2028, announced August 26, 2026
- Unit / basis
- issuer-stated incremental NVIDIA GPU deployments across AWS global infrastructure
Figure
more than 1 million GPUs
- Entity
- Amazon Web Services
- Period / as-of
- GTC 2026 plan, starting in 2026, restated August 26, 2026
- Unit / basis
- prior AWS/NVIDIA GPU deployment plan as characterized in the August 26 release
Figure
100,000 GPUs
- Entity
- AWS and NVIDIA for U.S. government AI factories
- Period / as-of
- planned delivery, announced August 26, 2026
- Unit / basis
- issuer-stated GPUs on secure AWS infrastructure for federal and national-security workloads
Figure
$96.221 billion
- Entity
- NVIDIA Corporation
- Period / as-of
- Fiscal Q2 2027 ended July 26, 2026
- Unit / basis
- GAAP revenue from the separate 8-K recap, not from the AWS release
Figure
$89.023 billion
- Entity
- NVIDIA Data Center
- Period / as-of
- Fiscal Q2 2027 ended July 26, 2026
- Unit / basis
- issuer-stated Data Center revenue from the separate 8-K recap