Editorial illustration for Amazon triples Nvidia chip order amid surging demand for AI
Amazon Triples Nvidia GPU Order for AI Expansion
Amazon just agreed to buy 2 million more Nvidia GPUs, and the ink on the last order barely dried. The two companies announced an expanded partnership Wednesday, during Nvidia's quarterly earnings call, that will send Blackwell Ultra, Rubin, and Rubin Ultra chips to Amazon Web Services data centers in 2027 and 2028. Five months ago, Amazon had committed to deploying more than 1 million Nvidia GPUs starting this year.
Nvidia says that order is no longer big enough. "Demand has exceeded those expectations," the company said in a statement, without disclosing financial terms.
Nobody's saying exactly what Nvidia stands to make off this, but at current GPU unit costs, the math points to tens of billions of dollars. What makes the deal stand out isn't just the size or the turnaround time. Nvidia's networking hardware, open models, CPUs, data processing tools, and robotics platform are all getting folded into AWS infrastructure too.
That's a deeper entanglement between the two companies, and it's happening at the same time Amazon is pouring money into its own AI chips, ones that could eventually compete with Nvidia's. Startups, enterprises, AI labs, and governments are all named as sources of the demand driving this expansion.
Neither company shared financial terms. It’s unclear what the exact return will be for Nvidia. But considering GPU unit costs, the deal is worth tens of billions of dollars.
Why this matters
Amazon building its own Trainium chips while tripling its Nvidia order in the same breath tells us something about where AI infrastructure actually stands right now: nobody, not even the hyperscalers with the deepest pockets and the clearest incentive to cut Nvidia out, can walk away from the GPU stack yet. That's the real signal here, more than the 2 million unit figure or the 2027-2028 delivery window. For developers and founders building on AWS, this locks in access to Blackwell Ultra and Rubin-class hardware for years, which matters if your roadmap depends on training runs that don't exist yet.
For researchers watching the chip market, it's a data point against the "AI capex bubble" narrative, at least in the near term, since this is a multi-year commitment from a company with every reason to hedge. Worth watching whether Amazon's own silicon ever narrows the gap enough to change these numbers, or whether Nvidia's grip on frontier-scale training just keeps tightening regardless of who else is in the room.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
Related Topic NVIDIA and Google Cloud let developers scale AI from prototype to production
Market Trends
Related Topic NVIDIA NeMo powers telco reasoning model for autonomous network workflows
Industry Applications
From Archives AI Agents Will Define Future Customer Engagement
From Archives OpenAI's Astra AI model family fuels Sam Altman's 2026 AGI timeline
From Archives Nvidia Claims Groq 3 LPX Cuts Coding Tasks From Hours to Minutes
Research & Benchmarks
From Archives NVIDIA Blackwell Leads AI Agent Performance in SemiAnalysis Benchmark
Common Questions Answered
Why did Amazon triple its Nvidia GPU order just five months after committing to 1 million GPUs?
Amazon tripled its order because demand for AI chips has significantly exceeded initial expectations. The company's expanded partnership with Nvidia reflects the surging demand for GPU capacity needed to power AI infrastructure and services across AWS data centers.
What specific Nvidia chip models will Amazon receive in the expanded order?
Amazon will receive Blackwell Ultra, Rubin, and Rubin Ultra chips as part of the expanded partnership. These chips are scheduled for delivery to AWS data centers in 2027 and 2028, representing Nvidia's latest generation of GPU technology.
What is the estimated financial value of Amazon's 2 million GPU order from Nvidia?
While neither company disclosed specific financial terms, the deal is estimated to be worth tens of billions of dollars based on typical GPU unit costs. The exact return for Nvidia remains unclear due to the lack of publicly shared pricing information.
Why does Amazon's continued reliance on Nvidia chips matter despite developing its own Trainium chips?
Amazon's decision to triple its Nvidia order while simultaneously building proprietary Trainium chips signals that even hyperscalers with significant resources and incentives cannot yet replace Nvidia's GPU stack. This demonstrates that Nvidia's technology remains essential to current AI infrastructure, and no company has successfully walked away from GPU dependency yet.
Further Reading
- Amazon just tripled its order of Nvidia chips over 'surging demand' | TechCrunch - TechCrunch
- Nvidia Announces Expanded Amazon Partnership on Earnings Call - The Wall Street Journal
- AWS and NVIDIA expand partnership for next-gen AI infrastructure - Amazon About Amazon
- Nvidia forecasts quarterly revenue above estimates, shares rise - Reuters
- AWS and NVIDIA deepen strategic collaboration to accelerate AI from pilot to production - AWS Machine Learning Blog
