Amazon’s $5 billion Nvidia AI servers sit idle in warehouses amid data center power shortages

# Amazon’s $5 Billion Nvidia Server Deliveries Face Delays as Data Centers Lack Power

Amazon is investing heavily in artificial intelligence infrastructure, but some of its newly purchased Nvidia server systems are reportedly sitting in warehouses because its data centers do not have enough electricity to operate them. The situation highlights the growing infrastructure challenges facing technology companies racing to expand their AI computing capacity.

According to a Business Insider report, internal emails indicated that Amazon expected to receive approximately $5 billion worth of Nvidia AI server systems in August, followed by another $3 billion worth each month. The emails did not establish how long the deliveries would continue.

Under a process called Early Rack Receipt, Amazon Web Services (AWS) planned to receive about $5 billion worth of Nvidia GB300 server racks at U.S. facilities before they were ready for final deployment. One August email projected a further $3 billion in GB300 racks each month “for the foreseeable future.”

### Electricity Shortages Delay Server Deployment

Although some AWS locations reportedly had sufficient physical space to accommodate additional server racks, they lacked the electrical capacity needed to power them. As a result, completed equipment had to remain in intermediate warehouses until the relevant data centers were ready.

One internal email indicated that at least one AWS team did not have “a clear forecasted path” for moving incoming racks into operational facilities at the time.

The problem reflects a broader challenge in the AI industry: building data centers and securing electricity can take longer than purchasing advanced computing hardware. As companies place large orders for AI chips and servers, delays in power availability can prevent those investments from being put to work.

### The Growing Problem of “Dark GPUs”

The situation is an example of what the industry calls “dark GPUs,” referring to AI processors and related server equipment that remain unused because of power shortages or other deployment obstacles.

These delays can create substantial financial pressure. Companies often commit billions of dollars to secure Nvidia GPUs and other specialized hardware well before their data centers are fully operational. When facilities cannot begin running the equipment, capital remains tied up without generating the expected computing capacity.

Financing can help companies cover the gap between hardware purchases and operational readiness. However, the report noted that lenders have recently become less willing to finance certain data center projects, potentially making it harder for developers to manage the costs of delayed deployments.

### AI Expansion Adds to Amazon’s Cash-Flow Pressure

Amazon’s infrastructure spending comes as the company commits enormous resources to expanding its AI capabilities. AWS infrastructure chief Prashant Kalyanaraman reportedly described Amazon’s AI business as “hugely free cash flow negative,” adding that he expected the company’s free cash flow to remain negative for another couple of years, according to Benzinga.

Amazon plans to spend $220 billion on capital investments across the company this year. At the same time, its free cash flow fell to negative $7.6 billion for the 12 months ended June 30, compared with positive $18.2 billion during the preceding year.

The reported warehouse delays illustrate the difficulties of coordinating hardware deliveries, electricity supply and data center construction at the scale required for AI. For Amazon and other major technology companies, securing advanced chips is only one part of the challenge; ensuring that facilities have the power and infrastructure to use them is equally important.