Amazon bets on a world filled with AI agents
Amazon used its re:Invent conference in Las Vegas to send a clear message.
The company is not chasing a single headline partnership. It is building an ecosystem where AI agents, custom silicon and on-premises deployments become the backbone of enterprise computing.
The update kept Bank of America firmly bullish, even without a showstopping announcement. The event drew around 60,000 attendees and, according to Bank of America analysts, delivered steady product progress rather than sentiment-shifting surprises.
Still, the analysts highlighted that AWS offered “lots of AI substance,” pointing to upgrades across infrastructure and model-building tools.
AWS chief executive Matt Garman sketched the most ambitious idea. He expects enterprises to create “billions” of autonomous agents as they become more scalable and independent.
To that end, AWS introduced three frontier agents designed to work collaboratively. The Kiro Autonomous Agent maintains context across sessions. The AWS Security Agent scans code for vulnerabilities. The AWS DevOps Agent triages incidents and recommends improvements.
Bank of America said this focus aligns with rising enterprise demand for agents. The firm called AI appetite “robust” and said the roadmap positions AWS well for expanding workloads.
The hardware story was equally pointed. Trainium3 is now generally available, offering 4.4 times more compute and four times greater energy efficiency than its predecessor, according to Amazon.
Trainium4 is in development and will use Nvidia’s NVLink Fusion racks to improve speed and efficiency.
Analysts said the most meaningful shift was the deepening collaboration between AWS and Nvidia in future Trainium hardware. AWS also highlighted EC2 instances powered by Nvidia’s GB300 chips, which offer higher FP4 performance than the GB200 line.
The surprise was not silicon but geography. Amazon introduced “AI Factories,” on-premises environments that embed Trainium and Nvidia infrastructure directly inside customers’ data centres.
These sites operate as private AWS Regions, offering low-latency access to compute, storage and AI services while addressing regulatory and data-sovereignty needs. Bank of America said AI Factories could help AWS expand capacity while relying on partners to supply power.
Conversations with AWS executives and partners reinforced the same theme: AI is driving new cloud demand and intensifying competition for workloads. Even without a marquee LLM partner, Bank of America said AWS now has “growing hardware, model, and agent capabilities” that should support adoption. The analysts expect revenue acceleration and reiterated their Buy rating with a $303 target.
