Tech Giant Amazon Abandons Expansive AI Data Center Projects, Mirroring Microsoft's Decision
The number of tech giants holding back on their AI data center expansion plans has risen to two, according to banks Wells Fargo and TD Cowen. Reports suggest Amazon has paused negotiations on some co-location data center deals, primarily in Europe, following similar actions by Microsoft.
A Wells Fargo report noted the scale of the pause in Amazon's negotiations is unclear but likened it to the positioning previously heard from Microsoft. Despite the pause, the report emphasized that Amazon seems intent on sticking with deals already signed. Co-location refers to the practice of sharing enormous infrastructure costs by constructing data centers in collaboration with other companies that require it.
It's worth noting that companies like Meta and xAI continue to aggressively build out data centers to fuel their AI models. Building large-scale data centers necessitates significant amounts of power, which grids have struggled to provide. Amazon may need more time to open data centers already under construction, as it currently holds 9 GWs (gigawatts) of active power capacity in its existing data center infrastructure.
Concerns have arisen that the demand for AI infrastructure might be cooling, as businesses search for ways to effectively apply new technology to save time and money. These concerns are compounded by President Trump's ongoing trade war, which has caused stocks to fall. Amazon has dropped 24% this year, and the company is exposed to tariffs on China, with estimates suggesting more than 70% of goods on its namesake marketplace come from China.
On Monday, Kevin Miller, a vice president of global data centers at Amazon Web Services, wrote a LinkedIn post in response to the Wells Fargo report. He stated that Amazon has begun to "consider multiple options in parallel," and that it frequently adjusts its plans on where to build out new server infrastructure based on evolving needs.
There is apprehension amongst economists that the trade war and potential recession could slow down the AI boom, as major players, including Nvidia, find themselves caught in the crossfire. Nvidia receives a substantial amount of its business from China and is under scrutiny for potentially turning a blind eye to high-end chips evading sanctions and landing in China. If Amazon scaling back on new data centers, it could further hurt Nvidia's sales of chips.
Amazon is set to release its next earnings on May 1st, and attention will be focused on how AI demand looks. Tech companies, including Amazon, have invested heavily into AI infrastructure and have an incentive to maintain an optimistic outlook. However, Microsoft recently pulled back on an ambitious $1 billion data center project in Ohio, surprising officials who offered generous tax incentives to secure the deal despite concerns about limited employment and energy use. It also passed on leasing more capacity from CoreWeave, which recently went public as a provider of data centers tailored for AI. CEO Satya Nadella has attempted to temper expectations regarding the AI revolution, stating in a recent interview that the technology has not yet turned into a meaningful lift for the U.S. economy. Despite this, Microsoft has reiterated plans to spend $80 billion on infrastructure in the next few years.
OpenAI has claimed ChatGPT has over 400 million weekly active users, but many AI products today fall short of generating the hype they generate online. Their user or revenue numbers may not be sustainable if the products fail to meet expectations. Microsoft Copilot, marketed towards the enterprise, has been criticized for offering little added value for the additional cost and resources required to implement it on internal systems.
If there are any silver linings, locals will not have to pay as much in taxes to support the canceled upgrades. On the flip side, they do employ a lot of construction workers during the initial rollout, and there was some hope that the demands of data centers would push local municipalities to upgrade fraying infrastructure and build out clean energy. There was a sliver of hope that AI would provide something of immediate practical value, but disappointments like chatbots that get things wrong all the time and Palantir-based police state systems have dampened these hopes.
- The pause in Amazon's AI data center expansion, as reported by Wells Fargo and TD Cowen, may signal a potential slowdown in the demand for AI infrastructure, raising concerns among economists about the future of the AI boom.
- On the tech front, reports suggest major tech players like Amazon and Microsoft are cautiously approaching data center expansions, while companies like Meta and xAI continue to aggressively build out their infrastructure to fuel AI models.
- Despite the ongoing trade war and potential recession, technology companies, including Amazon, have invested heavily into AI infrastructure and are incentivized to maintain an optimistic outlook, such as Microsoft's reiterated plans to spend $80 billion on infrastructure in the next few years.
