AWS CEO Matt Garman says AI compute “supply” is largely booked through 2028, challenging AI hardware skeptics

Amazon Web Services CEO Matt Garman used a Bloomberg Technology appearance to address a growing investor argument about AI infrastructure. According to a report carried by Yahoo Finance, Garman said AWS’s growth is only getting started and that “supply is mostly ‘spoken for’ through 2028,” a line that counters the view that the next wave of AI spending could quickly run into an oversupply or demand slowdown.

The comments were framed as part of a larger debate between investors who see AI cloud spending as a near-term boom that will normalize, and those who worry that the industry’s buildout could be constrained by costly supply chains or that customer demand might not remain as robust once early projects mature. In the Yahoo Finance report, Garman is portrayed as pushing back on the bearish interpretation by pointing to capacity being booked far into the future.

Garman’s remarks also connect to AWS’s broader strategy of selling not just cloud storage and compute, but the full stack of AI infrastructure, including access to accelerated compute, managed AI services, and the networking and operations needed to run large-scale models. While the report focuses on the “spoken for” capacity message, the underlying point is that AWS is competing on the ability to deliver usable AI environments to customers when they need them.

The “through 2028” framing matters because it suggests AWS is planning around multi-year demand rather than assuming that AI infrastructure deployments will be short-lived. For companies and developers building and training models, lead times for specialized hardware and data center capacity can be a critical factor. A claim that supply is largely spoken for for several years implies that AWS believes it can keep meeting demand rather than being forced to pause deployments.

Amazon’s investor messaging on AI has also leaned into scale, particularly in how AWS provides the infrastructure for both foundational model workloads and enterprise AI use cases. In that context, Garman’s comments read as an attempt to shift the conversation from whether AI spending is peaking to whether AWS can keep converting demand into revenue even as competitors build similar capacity.

Still, key specifics are not provided in the Yahoo Finance write-up. The report does not lay out which type of “supply” is being referenced, how those commitments are measured, whether the figure is tied to contracted capacity versus expected demand, or what the booking assumptions are at a granular level. It also does not clarify whether the “spoken for” language reflects binding customer commitments, internal allocation targets, or a combination of both.

For AWS and the wider cloud market, the takeaway is less about a precise forecast number and more about direction. If AWS leadership believes supply constraints and demand visibility extend through 2028, it supports the argument that cloud buyers will keep prioritizing AI infrastructure investments. At the same time, investors will likely look for follow-up disclosures in earnings materials that break down AI-related growth drivers and capacity utilization more concretely.

What to watch next is whether Amazon provides additional detail in its regular reporting on AI capacity commitments, customer demand, and any changes in timing for new infrastructure build-outs. Analysts will also watch for similar statements from other cloud providers, since competition in AI acceleration and data center capacity can quickly change the balance between demand and available supply.