Amazon and NVIDIA have not disclosed an $8 billion price for their latest GPU expansion. Their August 26, 2026 announcement says AWS plans to deploy two million additional NVIDIA GPUs in 2027–2028, but gives no contract value. The headline figure is therefore unverified; what is clear is that the buildout is part of a much larger and increasingly expensive race to supply AI computing capacity.
What did Amazon and NVIDIA announce?
AWS and NVIDIA announced an expansion of their strategic collaboration on August 26, 2026. AWS plans to add two million NVIDIA GPUs across its global infrastructure during 2027–2028. The announcement names Blackwell Ultra, Rubin and Rubin Ultra GPUs. AWS and NVIDIA’s announcement also describes work on NVIDIA Vera CPUs, NVLink Fusion, high-bandwidth memory and networking, as well as Nemotron open models on Amazon Bedrock and SageMaker, data processing, vector indexing and robotics.
This is an infrastructure commitment, not a consumer graphics-card announcement. The GPUs are intended for AWS data centers and cloud services; the companies did not describe the plan as a purchase consumers can make directly.
How many NVIDIA GPUs is AWS adding, and when?
The announced addition is two million GPUs, with deployment planned across 2027 and 2028. It follows Amazon’s earlier plan to deploy more than one million NVIDIA GPUs starting in 2026. These are separate announcements, and the two-million figure is described as additional capacity—not as a replacement for the earlier deployment. Amazon’s Q1 2026 results also said it had landed more than 2.1 million AI chips over the preceding 12 months, more than half of them Trainium chips.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →#1 Best Overall
- AI Performance: 767 AI TOPS
- OC mode: 2632 MHz (OC mode)/ 2602 MHz (Default mode)
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Axial-tech fan design features a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
- A 2.5-slot design maximizes compatibility and cooling efficiency for superior performance in small chassis
How much will Amazon’s NVIDIA deal cost?
The August announcement does not state a dollar value. TechCrunch reported that the companies did not disclose financial terms; its suggestion that the expansion could be worth tens of billions of dollars is an estimate based on unit costs, not a confirmed contract amount. The report does not establish that the deal costs $8 billion. Unless a reliable source identifies what that figure refers to, it should not be treated as the price of the two-million-GPU plan.
Other large figures help show the scale of Amazon’s AI investment, but they describe different things:
Rank #2
- Chipset: GeForce RTX 3050
- Boost Clock / Memory: 1492 MHz / 14 Gbps
- Video Memory: 6GB GDDR6
- Memory Interface: 96-bit
- Output: DisplayPort x 1 (v1.4a) / HDMI 2.1a x 2
- $220 billion: Amazon’s expected total capital spending for 2026, as reported by the Associated Press after the company’s Q2 results. The outlook covers technology, robots, semiconductors and satellites—not just NVIDIA GPUs. AP reported that Amazon raised its outlook from $200 billion, citing higher memory-chip costs as a main reason. Associated Press report
- More than $20 billion: Amazon said its chip business—including Graviton, Trainium and Nitro—had exceeded this annual revenue run rate in Q1 2026 and was growing at a triple-digit rate year over year. That is a reported run rate for Amazon’s chip business, not the value of the NVIDIA expansion. Amazon’s Q1 2026 results
Amazon CEO Andy Jassy said the company would still not have enough capacity to meet all demand in 2026, even with the $220 billion capital-spending outlook. That statement underscores the central financial tension: spending more does not instantly produce usable compute capacity.
Does the NVIDIA expansion replace Amazon’s own AI chips?
No. Amazon’s announcements point to a mix of outside GPUs and custom silicon rather than an either-or choice. NVIDIA GPUs add capacity for workloads that use NVIDIA’s platform, while Amazon continues developing and deploying its own chips, including Trainium, Graviton and Nitro. The Q1 figure of more than 2.1 million AI chips landed in the prior 12 months included more than half Trainium, according to Amazon.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteRank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5070
- Integrated with 12GB GDDR7 192bit memory interface
- PCIe 5.0
- NVIDIA SFF ready
For AWS, the combination offers customers different hardware options and lets the company build its own chip business alongside its NVIDIA infrastructure. For investors and customers, the announced GPU count alone does not reveal how much AWS will spend, how quickly each data-center site will come online, or what share of demand will be served by custom chips.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why is building AI capacity so expensive and difficult?
Buying processors is only one part of a data-center buildout. NVIDIA’s July 2026 SEC filing identifies land, power, data-center capacity and capital as critical to AI infrastructure deployment. It describes expansion as a complex, multi-year process involving regulatory, technical and construction challenges. The filing discusses general industry risks; it does not say AWS’s announced plan has been delayed. NVIDIA’s SEC filings
Rank #4
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
Those constraints help explain why headline spending totals and GPU counts do not translate directly into immediate computing supply. Facilities, electricity and supporting systems must be available, and construction and approvals can take time. NVIDIA also says its infrastructure commitments and guarantees depend on customers and partners meeting their own performance obligations.
Quick Recap
What the announcement does—and does not—tell us
- Established: AWS plans to deploy two million additional NVIDIA GPUs across its global infrastructure in 2027–2028.
- Not established: The announcement does not disclose a purchase price or confirm an $8 billion value.
- Broader scope: The collaboration includes CPUs, networking, memory, AI models, data processing, vector indexing and robotics—not GPUs alone.
- Business context: Amazon is expanding NVIDIA capacity while continuing its own chip effort, against a backdrop of $220 billion in expected 2026 capital spending across the company.
- Execution context: Power, land, facilities, capital and construction affect how quickly announced capacity can become available.
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.
Free tools Windows power users keep installed
One-click scans. No signup required.
