Nvidia and Amazon Web Services (AWS) are planning a massive expansion of AI computing capacity. According to Nvidia's chief financial officer, the two companies intend to add 2 million more Nvidia graphics processing units (GPUs) to AWS's global data centers during 2027 and 2028. That includes building US government data centers equipped with 100,000 of those chips on what the company calls "secure AWS infrastructure."
What's behind the big number?
The headline figure is eye-catching, but the timing is just as important. The 2027-2028 window is far enough out that this announcement functions partly as a reservation system for scarce AI chips. On a recent call reported by Reuters, Nvidia's CFO said the company expects supply to remain tight through the end of fiscal 2028, even as demand expands from hyperscalers like AWS to newer "neocloud" providers and AI labs building their own infrastructure.
GPUs are the specialized processors that power most AI training and inference workloads. They have become the most sought-after component in the tech industry, with companies like Microsoft, Google, and Meta competing for limited supply. AWS, as the largest cloud provider, is a key customer for Nvidia, and this deal locks in a significant portion of future capacity.
The US government component is notable because it signals growing federal interest in AI infrastructure. Government data centers with 100,000 GPUs would be among the largest AI computing facilities in the world, likely used for defense, intelligence, and scientific research applications.
Why this matters for investors
For everyday investors, this announcement is a strong signal about the durability of the AI spending boom. When two of the biggest names in tech commit to multi-year GPU purchases, it suggests they see sustained demand for AI services well into the late 2020s. That's reassuring for those who worry that the AI trade might be a bubble that bursts quickly.
It also reinforces the dominant position of Nvidia in the AI chip market. Despite competition from AMD and custom chips from Google and Amazon, Nvidia continues to secure massive orders years in advance. For Nvidia shareholders, this kind of forward visibility is valuable because it reduces uncertainty about future revenue.
For AWS and its parent Amazon, the deal means committing billions of dollars to AI infrastructure. That's a bet that customers will keep renting these chips at high margins. Amazon's cloud division has been a major profit driver for the company, and doubling down on AI capacity suggests management sees this as a long-term growth opportunity.
What to watch next
Investors should keep an eye on a few things. First, whether Nvidia can actually deliver on these commitments. The company has faced supply chain constraints in the past, and meeting a 2 million GPU order will require continued manufacturing expansion.
Second, watch for signs of demand softening. If AI spending slows, these long-term commitments could become a burden rather than a benefit. So far, however, the trend is in the opposite direction. Nvidia's recent $108 billion forecast suggested Big Tech's AI spending remains strong, and this AWS deal reinforces that picture.
Finally, the government angle could open a new market. If the US government becomes a major buyer of AI computing, that could create additional demand for both Nvidia and cloud providers. It also raises questions about how these facilities will be used and what security measures will be in place.
The bottom line
This announcement is a vote of confidence in the long-term future of AI. For investors, it's a reminder that the AI infrastructure buildout is still in its early stages, with major commitments stretching years into the future. While no one knows exactly how the AI market will evolve, the fact that Nvidia and AWS are planning this far ahead suggests they expect the boom to last.
As always, it's wise to remember that these are plans, not guarantees. Supply chains can break, demand can shift, and technology can change. But for now, the message from two of the industry's biggest players is clear: AI computing is going to be a very big business for years to come.
