Amazon’s AI Bet Points Toward a Smaller Model Future

Amazon CEO Andy Jassy sees an AI future shaped by a surprisingly small group of powerful models. He predicts that 6 to 8 AI models will become “good enough” for almost everything, while Amazon’s own cloud business races to support the companies building and deploying them.

That prediction arrives as Amazon expands its AI infrastructure, reports major AWS growth, and prepares to spend even more on the systems behind the AI boom. Amazon reported its third-quarter results on Thursday, with Bedrock usage, AI agents, and data center investment all pointing toward a business entering a larger phase of expansion.

Bedrock Is Becoming AWS’s AI Growth Engine

Amazon continues to expand the capabilities of Amazon Bedrock, its inference service, and Jassy said that expansion is driving “remarkable growth.” Fortune 500 companies are spending nearly seven times more on Bedrock year over year, while total Bedrock token volume has grown nearly 17 times over the same period.

Those numbers place Bedrock at the center of Amazon’s AI strategy. Jassy described it as the fastest growing multi-billion dollar service in AWS’s history, a milestone that gives the cloud division a powerful new growth engine alongside its existing infrastructure business.

Amazon has also added OpenAI models to Bedrock, and usage of those models is growing. That broader model selection gives customers another path into Bedrock as they choose the systems that fit their applications, workloads, and AI agents.

Jassy also pointed to customer interest in security, saying it “exceeds anything we’ve ever seen for a new security service.” The statement places AI infrastructure and security in the same growth story, as businesses build systems that need both model access and protection.

AI Agents Are Moving From Experiments Into Deployment

The Bedrock expansion is not limited to model access. Agents are being deployed at accelerating rates, with many built on Bedrock agent core building blocks. Tens of thousands of customers currently use agent core, and those customers deployed two times more agents with it in Q3 than during the prior three quarters combined.

That shift gives Amazon a second major AI measure beyond token volume. Models answer requests, but agents can carry out tasks through connected systems, making their adoption a key sign of how businesses are putting AI into operation.

The scale of agent core usage also connects with Jassy’s prediction about the AI model market. If 6 to 8 models become good enough for almost everything, businesses may focus less on finding endless new models and more on deploying those models through cloud services, agents, and enterprise tools.

Amazon’s figures show Bedrock serving both sides of that transition: customers can access models, and they can build agents on top of the infrastructure that supports them. The company’s AI strategy is therefore expanding across models, inference, security, and agent deployment.

A Bigger Spending Push Is Coming

Jassy said, “Our AWS investment in functionality and infrastructure is working,” pointing to over 39% year-over-year revenue growth in Q3 and a $184 billion revenue run rate. He also gave projected revenue growth rates of over 40% in both 2027 and 2028.

Amazon has already committed to $220 billion in capital expenditure for 2026, and the company expects “to spend meaningfully more in 2027.” Amazon is already spending a record amount on AI data centers, so the next stage of investment will place even more attention on the infrastructure required to serve growing model and agent workloads.

The spending outlook arrives alongside a reported $42.4 billion hole for the trailing twelve months in Amazon’s third-quarter results. At the same time, the company is presenting AWS growth, retail gains, and newer businesses as parts of a much larger expansion plan.

Jassy said broader selection and speedier delivery are helping Amazon capture more retail market segment share while improving margins. He also said newer businesses, ranging from Grocery to Pharmacy to Amazon Leo to Zoox, are progressing encouragingly.

Amazon Sees Room For Two Trillion-Dollar Businesses

Jassy framed Amazon’s outlook on a scale that reaches beyond AI infrastructure. “There has never been a trillion dollar revenue business, and we may have two of them in the next 5-10 years,” he said.

That forecast ties AWS’s AI momentum to Amazon’s broader ambitions. Bedrock is growing at a pace measured in multiples, AWS revenue growth is projected above 40% in 2027 and 2028, and Amazon expects its capital spending to rise after an already massive 2026 plan.

The central question now is how the model market develops. Jassy’s 6-to-8 prediction suggests a future where a limited set of capable models powers a huge range of products, while Amazon competes to provide the cloud infrastructure, model access, security, and agents that turn those systems into working businesses.

Amazon is betting that the real opportunity will not come from having the largest number of models. It will come from making the most useful models available everywhere customers need them, then giving those customers the infrastructure to build what comes next.

Stay connected via Google News

Follow us for the latest AI News