AWS just posted its strongest quarter in years, with revenue climbing 37% year over year to $42.2 billion for the period ended June 30. That is the fastest growth the cloud giant has seen in more than four years and its best showing in 18 quarters. Yet behind that impressive number sits a growing problem: an AWS capacity shortage that threatens to slow the momentum just as demand is peaking.
The surge in AWS cloud growth is being driven largely by artificial intelligence workloads, which require enormous amounts of compute power and storage. As more businesses rush to deploy AI tools, they are consuming infrastructure faster than it can be built out. The result is that capacity constraints are worsening even as revenue hits record levels. For you, that could mean longer lead times for new resources, tighter limits on instance availability, and a need to plan your cloud usage more carefully than before.
Why AWS Capacity Remains Insufficient Despite Record Revenue and Capex
Given those practical delays and resource limits, you might wonder how AWS can post record revenue yet still struggle to keep up. The answer lies in the sheer scale of the AWS capacity shortage. Amazon CEO Andy Jassy has been remarkably candid about the situation. He stated that even a massive $220 billion infrastructure investment plan would not be enough to satisfy all customer demand through 2026.
The Gap Between Revenue and Capacity
This directly highlights the core demand vs supply imbalance. While the revenue numbers look strong, they represent only the business AWS can currently serve. A significant portion of potential growth is being held back by the physical limits of data center construction and hardware availability. The company is spending heavily on infrastructure investment, but it is a race against the explosive growth of AI workloads.
CEO Warnings on Future Demand
Jassy warned that the capacity constraint would likely continue well into 2027. The cloud capacity backlog is a clear numerical indicator of this pressure. AWS ended the quarter with a staggering $496 billion backlog, a massive jump from $364 billion just three months earlier. This means committed customer projects are waiting in line for compute power and storage. For you, this reinforces the need to commit to long-term contracts and prioritize your most critical workloads to navigate the ongoing AWS capacity shortage.
Hardware and Resource Constraints: Memory, Power, and Grid Delays
These wait times for compute and storage are just the visible part of the problem. Beneath the surface, specific hardware and resource constraints are tightening the screws on the AWS capacity shortage. Three bottlenecks stand out: soaring memory costs, a surge in electricity demand, and painfully slow grid connections.
Memory Costs Drive Capex Increase
You might wonder why Amazon raised its 2026 capital expenditure forecast by roughly $20 billion. According to CEO Andy Jassy, higher memory costs were the primary driver. Memory, especially high-bandwidth memory used in AI servers, has become significantly more expensive. This memory cost increase forces AWS to spend more just to maintain growth, and it contributes directly to the capacity crunch. For you, this means that the price of reserving compute resources may rise, as AWS passes along some of these hardware expenses.
Electricity Demand Surge
Beyond chips, power is becoming a scarce resource. The International Energy Agency expects global electricity consumption from data centres to double between 2025 and 2030. Even more striking, electricity use by AI-focused facilities is projected to triple. This data center power consumption surge puts immense pressure on local grids. AWS and other providers must secure massive amounts of energy to run and cool their servers, and that energy isn’t always available when and where it’s needed.
Grid Connection Bottlenecks
Getting that power to new data centres is another challenge entirely. A US Department of Energy advisory report noted that grid connection delays for hyperscale data centers requiring 300 MW to 1,000 MW or more can involve lead times of one to three years. That’s a long wait for a facility that needs to come online quickly to meet AI demand. These bottlenecks slow down the entire expansion process, making the AWS capacity shortage more persistent. For you, this means that even if AWS wants to build more capacity, the physical infrastructure—power lines and substations—can’t keep up. Planning ahead for your own cloud usage becomes even more critical.
How Amazon Will Allocate the $220 Billion Capex
That’s precisely why Amazon is throwing enormous resources at the problem. The scale of its investment is staggering. The company recently increased its 2026 cash capital expenditure forecast from roughly $200 billion to $220 billion. And in the second quarter alone, Amazon recorded $53.1 billion in cash capital expenditure, with the vast majority tied to AWS and generative AI. This spending is a direct response to the aws capacity shortage that’s pressuring the entire cloud industry.
Breakdown of the $220 Billion Plan
Where exactly is all that money going? A very large chunk of this capital expenditure allocation is funneled into AWS infrastructure investment. That means building new data centers, upgrading existing facilities, and expanding the network backbone that keeps everything connected. Amazon is also heavily investing in the specialized hardware required for AI workloads—think powerful GPUs and custom chips. This generative AI spending isn’t just about buying equipment; it’s also about developing the software stack to make it all work efficiently.
Focus on AWS and Generative AI
The data center buildout is a top priority. Amazon is racing to secure land, power, and cooling systems to support the massive clusters of servers needed for AI training and inference. This includes everything from long-term power purchase agreements to modular data center designs that can be deployed faster. For you, this investment eventually means more capacity and less chance of hitting limits. But translating billions of dollars into usable cloud resources takes time—permits, construction, and equipment delivery all have their own timelines. The current aws capacity shortage won’t vanish overnight, but these moves signal a serious commitment to closing the gap.
Timeline for New AWS Capacity: When Will the Shortage Ease?
If you’re wondering when the aws capacity shortage will finally let up, the honest answer is: not soon. Building the kind of infrastructure that powers cloud computing is a marathon, not a sprint. New data centers take years to plan, permit, and power up, and much of the capacity coming online in the near future is already spoken for. Here’s a look at the key stages of the timeline and what they mean for you.
Two-Year Construction Cycle
Amazon doesn’t start writing checks when a data center is nearly finished. The company begins spending on new facilities about two years before they actually open. That means the data center construction timeline is already baked into AWS’s road map. For you, this capacity lead time explains why the shortage feels stubborn—even as Amazon pours billions into expansion, the physical buildings and equipment simply aren’t ready yet. The projects breaking ground today won’t contribute a single server until at least 2026 or 2027.
On a similar note, Apple TV’s Neuromancer Explores Eerily Timely Future explores this topic with concrete examples.
Grid Connection Delays
Even after the concrete is poured, there’s another bottleneck: getting enough electricity. A US Department of Energy advisory report noted that grid-connection requests for hyperscale facilities requiring 300 MW to 1,000 MW or more can involve lead times of one to three years. These hyperscale data center delays are a major reason why new capacity takes so long to come online. Power utilities have to upgrade substations, run new transmission lines, and navigate permitting—all of which adds years to the clock. So even if construction finishes on schedule, the lights might not turn on until the grid catches up.
Reservations Stretch to 2028
Perhaps the most sobering detail for anyone hoping for a quick fix: the “lion’s share” of AWS computing capacity planned for 2027 had already been reserved by customers, with some 2028 capacity also committed. That 2027 capacity reservation means that even as new data centers come online, a huge portion is already locked in by existing clients. For you, this reinforces that the aws capacity shortage isn’t a temporary blip—it’s a structural squeeze that will persist for years. If you haven’t secured long-term reservations yet, acting sooner rather than later is wise.
Customer Prioritization, Pricing, and Competitor Comparison
This means that as the aws capacity shortage continues, the way AWS allocates its limited resources becomes crucial. The company has clear priorities, and they heavily favor long-term commitments. If you rely on flexible, on-demand cloud pricing, you need to understand how this shift affects you.
Long-Term Contracts Dominate AI Capacity
Most of Amazon’s current AI capacity is contracted for terms of at least five years. These enterprise cloud contracts guarantee access for large customers, but they also lock up significant resources for extended periods. The latest AWS backlog, which reached $496 billion at the end of the quarter, up from $364 billion three months earlier, shows that customers are committing capacity far in advance. This trend is a clear signal of the cloud infrastructure shortage driving demand for capacity reservation.
Impact on On-Demand Customers
For you, if you rely on on-demand pricing, this shift is significant. When most capacity is pre-committed, on-demand customers face higher prices or may find themselves unable to secure the resources they need. AWS prioritizes its long-term contract holders, leaving less flexibility for shorter-term usage. This could mean your typical pay-as-you-go costs rise, or you experience delays in provisioning instances. The aws capacity shortage makes it harder to spin up resources on the fly.
Competitors Face Similar Constraints
This isn’t unique to AWS. Competitors like Microsoft Azure and Google Cloud are also dealing with similar capacity constraints. The entire cloud industry is feeling the pressure from AI demand. When comparing AWS vs Azure vs Google Cloud, you’ll find that all three are pushing long-term contracts and prioritizing enterprise accounts. The cloud infrastructure shortage is widespread, so switching providers might not be a quick fix. You’ll need to evaluate each platform’s capacity reservation options and plan accordingly.
Frequently Asked Questions
How will the AWS capacity shortage affect you if you have not reserved capacity?
If you haven’t reserved capacity, you may face longer wait times for new instances or see certain instance types become temporarily unavailable. AWS prioritizes reserved and committed-use instances, so on-demand users could experience slower scaling during peak demand. To mitigate this, consider using a mix of instance types or regions and enabling capacity reservations for critical workloads.
How does the AWS capacity shortage compare to competitors like Microsoft Azure and Google Cloud?
AWS faces unique challenges due to its massive scale, but competitors like Azure and Google Cloud also report capacity strains from AI demand. The key difference is that AWS’s shortage is more visible because it serves the broadest customer base. Each provider handles bottlenecks differently, yet all are investing heavily in new data centers and hardware upgrades to keep pace.
Will the AWS capacity shortage lead to price increases?
AWS has not announced any general price increases due to the shortage, but it may adjust pricing for specific high-demand services or instance types. You can lock in current rates by using reserved instances or savings plans. For now, the focus is on expanding capacity rather than raising prices broadly, though ongoing demand could shift this strategy.
