Andy Jassy unleashed 8-minute defense of Amazon’s AI Playbook on Carnings Call

“I have two (questions) for you. They’re a little difficult, but I’ll throw them at you,” Nowak said. Amazon CEO Andy Jassy. “There’s a Wall Street Finance story that AWS is late with concerns about losing share to its peers. What’s your counterargument, would you talk about the most important focus of you and your team to ensure that AWS stays on the edge of knife innovation and hyperschool peers?”

Nowak also pushed Jassy why it’s not fair to assume that AWS revenue growth should not accelerate by half a year, as AWS revenue growth would cash in on this transformational technology, as well as widespread demand from AWS’ generative AI delivery and businesses of all sizes.

Jassy responded by emphasizing that this is an early stage of technological transformation that extends into the future. Some of the frontier’s top model providers use AWS in one capacity, but non-AI AWS customers rushing to build generic, agent AI services using AWS “is pretty early on, many of which are just a little bit in terms of usage compared to some of the above-mentioned top heavy applications.” That must change.

So, following Jassy’s ideas, as more companies understand what they want to build and how they want to build, they start to have different needs. For the largest model makers, such as Open AI and Anthropic, Jassy foresees the costs of vomiting inference costs primarily from the mix between the training of the model and the costs associated with “inference” or the customer-oriented part where the model spits out predictions, answers, or actions. And Jassy argues that AWS is well positioned for this transition, due to low-cost AI chip line training.

“Currently, we have around 30% and 40% better price performance than other GPU providers, and we are already working on the third version,” he said.

For others who want to create their own generation AI applications using models from other companies, Jassy claimed that Amazon Bedrock, which offers a wide range of enterprise models, has become its destination and is “very substantially growing.”

Jassy realized that companies were just beginning to think about deploying AI agents, and continued on the thread to inning this first. With recent announcements of Agent AIAWS is good for capitalizing.

The CEO of Amazon and former AWS chief added that AWS Cloud leadership position also offers lock-in as AI’s “inference” becomes another component of the company’s cloud services stack.

“(P)EOPLE wants to run (AI) applications near where other applications are actually running and where the data is located,” says Jassy. “There are more applications and data running on AWS than anywhere else.”

Jassy doesn’t directly answer Nowak’s question about the possibility that AWS’ growth rate could accelerate later in the year, but he emphasized his optimism.

Early in the call, Jassy defended AWS’s 18% revenue growth rate Microsoft Recently reporting annual revenue growth for Azure Cloud Units and Alphabets, it has reported quarterly growth of 32% Google cloud. Azure generates about 2/3 of AWS revenue, while Google Cloud registers less than half of Amazon’s Cloud Behemoth annual revenue.

“You’re looking at the business. It’s a revenue implementation business of $123 billion a year, and it’s still too early,” he said. “An opportunity to say that annual revenue occupancy rates are $123 billion, and it’s still too early? That’s a very unusual opportunity for us to be very bullish.”