Amazon AWS cloud capacity largely sold out through 2027 on AI demand
Amazon is facing mounting pressure to expand data center capacity as demand for artificial intelligence computing continues to outstrip supply. The company says much of its Amazon Web Services capacity is already committed through 2027, with sizable bookings also extending into 2028.
Highlights
- Amazon CEO Andy Jassy states AWS cloud capacity for 2026 is already sold out, with 2027 largely committed and some 2028 capacity reserved.
- AWS continues to reserve some on-demand, Bedrock inference, and EC2 availability despite the surge in AI-driven capacity commitments from labs and enterprises.
- Amazon plans about $220 billion in capital expenditure for 2026 and may invest more, reflecting high AI infrastructure demand ahead of Q3 results on October 29.
AWS capacity commitments stretch into 2028
As reported by Reuters, Amazon Chief Executive Andy Jassy says Amazon Web Services is struggling to keep pace with heavy demand from AI labs and enterprise customers, leaving cloud computing capacity for 2026 already sold out.Speaking in a video podcast with Andreessen Horowitz co-founder Ben Horowitz and general partner Erik Torenberg on Friday, Jassy says AWS capacity for 2027 is largely sold out as well, while the company still reserves some availability for on-demand services, Bedrock inference offerings and EC2. He adds that a good portion of 2028 capacity is also committed.
EC2 is AWS's cloud server service for running applications, while Bedrock gives customers access to AI models for building generative AI products.
Spending plans and investor focus
Earlier this year, Amazon said it would spend about $220 billion in capital expenditure for 2026. Jassy says the company could invest significantly more and that additional spending would still represent sound capital allocation given the level of demand.The comments highlight how the rush to secure AI infrastructure is shaping investment plans across the cloud sector, as providers race to add computing power for model training and inference workloads. Amazon is scheduled to report third-quarter results on October 29, when investors are likely to focus on spending levels, capacity expansion and AWS demand trends.
In our earlier article on the mounting challenges for U.S. AI data center expansion, we examined how higher financing costs, construction delays and power limitations are starting to slow the buildout of new computing capacity. We also noted that weakening investor confidence and local pushback could further constrain how quickly developers can deliver on ambitious infrastructure plans.