Amazon Cloud Chief: AI Demand Shifting From Training to Business Application

Customer demand within Amazon's cloud division is rapidly moving from the initial, resource-intensive phase of building artificial intelligence models to the practical stage of deploying them in business operations, a process known as inference. This signals a key maturation in the adoption of AI technologies, according to the unit's chief executive.

From Training to Inference

Amazon Web Services (AWS) CEO Matt Garman stated that while companies continue to use large computing clusters for AI training, the focus is shifting. "As these models get really popular and really powerful, more and more companies are integrating that inference into their workloads," Garman said Monday on the Bloomberg Tech television show.

This marks a transition from the experimental to the operational phase of AI. Training refers to the computationally heavy process of creating an AI model, while inference is the application of that trained model to generate responses or analyze new data in real-time. This shift indicates that businesses are moving beyond development and are now embedding AI directly into their products and services.

Ramping Up Investment to Meet Demand

Ad

To support this surging demand, Amazon is significantly increasing its infrastructure spending. The company last week raised its 2026 capital expenditure forecast to $220 billion, up from a previous projection of $200 billion, citing rising costs for memory and other essential data center components.

Garman described the potential AI business for AWS as "just massive" and noted that the company is struggling to keep pace with customer needs. "Today, demand still significantly outstrips supply and we're trying to build and invest to keep up with what customers are asking for," he said. To better manage its build-out, AWS is securing five-year commitments from customers to help forecast usage for new data centers.

Market Impact

The growth in AI-driven cloud services has been a boon for Amazon's investors. The company's market value recently surpassed $3 trillion, supported by a report last week showing that its cloud-computing revenue had grown for a fifth consecutive quarter. The increasing demand for inference suggests a durable, long-term revenue stream as AI becomes a core part of enterprise technology stacks.