At AWS re:Invent 2025, Amazon CEO Matt Garman delivered a sweeping vision for the future of artificial intelligence, arguing that AI agents will ultimately be more transformative than both the Internet and cloud computing.

Speaking during his opening keynote at the annual conference in Las Vegas from December 1–5, Garman framed AI agents as the next foundational layer for businesses of every size and sector.

“This change is going to have as much impact on your business as the Internet or the cloud,” Garman said. “I believe that in the future, there’s going to be billions of agents inside of every company, across every industry.”

Brian Cole Jr & Pipe Bombs: Virginia Man Appears in Court - USA Herald

A Defining Moment for AI at AWS re:Invent 2025

Opening his keynote—streamed in an unexpected twist on Fortnite—Garman emphasized that the pace of technological change is only accelerating.

“The world is not slowing down,” he said. “More change is coming—and one of the biggest opportunities that is going to change everyone’s business is agents.”

According to Garman, nearly every industry is already being reinvented by AI, with customer experiences, workflows, and decision-making fundamentally reshaped by intelligent agents that can reason, act, and collaborate with humans.

Matt Garman Reveals AWS by the Numbers

Garman began with a recap of AWS’s scale and momentum. AWS generated approximately $132 billion over the past year, while more than 500 trillion objects are now stored in Amazon S3. Platforms like Amazon Bedrock also saw significant growth as enterprises adopted generative AI at scale.

AWS’s global infrastructure continues to expand, now spanning 38 regions, with three additional regions planned. Still, Garman underscored that growth alone isn’t the point.

“At Amazon, everything starts with the customer,” he said, adding that enterprises across virtually every industry now run core parts of their businesses on AWS.

AWS re:Invent 2025: Amazon Connect Gets More Human

One of the most tangible demonstrations of AI agents in action came with major upgrades to Amazon Connect. First launched in 2019, Amazon Connect already supports neural text-to-speech in more than 30 languages and speech recognition in over 25.

At AWS re:Invent 2025, AWS announced new capabilities designed to give AI agents more human-like reasoning and conversational skills across voice and messaging channels. Powered by advanced speech models from Nova Sonic, agents can now respond with more natural pacing, tone, and contextual understanding—even across different languages and accents.

AWS describes this as “true collaboration between humans and AI,” where agents analyze customer sentiment, manage routine tasks, and suggest next steps, freeing human workers to focus on complex problems and relationship-building.

“Instead of waiting for customers to ask, businesses can anticipate needs based on real-time behavior,” AWS said, pointing to new revenue opportunities driven by personalization.

Matt Garman Says An All-New Tech Stack is Needed

To support this future, Garman said an entirely new technology stack is required—starting with hardware. AWS highlighted its long-standing partnership with Nvidia, noting that the two companies have worked together for more than 15 years.

A new P63 GPU offering featuring Nvidia’s GB200 and GB300 chips promises major performance and efficiency gains. Garman revealed that OpenAI is already using this infrastructure to train and improve its models.

AI Factories, Trainium, and the Future of Infrastructure

AWS also unveiled AWS AI Factories, which function like private AWS regions deployed directly within customer data centers. These allow organizations to run highly customized AI infrastructure while maintaining control over sensitive data.

Custom silicon remains central to AWS’s strategy. Garman highlighted increasing adoption of Trainium chips for inference workloads, including use by AI services like Perplexity’s Claude. He also detailed Project Rainier, where more than 500,000 Trainium 2 chips are being used to scale Claude’s development.

Next up is Trainium 3, which offers 4.4x more compute than the previous generation, with ultra servers now available. AWS is already working on Trainium 4, promising even greater performance, bandwidth, and memory—though no release date was announced.

Bedrock, Nova 2, and Model Choice

On the software side, Garman stressed that model choice matters. Amazon Bedrock already supports models from DeepSeek, Meta, OpenAI, and Writer, with new additions from Google, Nvidia, Kimi, Minimax, and Mistral announced at re:Invent.

AWS also introduced Amazon Nova 2, the latest evolution of its foundation models. Available in three tiers, Nova 2 spans everyday workloads to complex reasoning tasks and now supports speech-to-speech, human-like conversations.

Garman said Nova 2 delivers “fast, cost-effective reasoning for everyday workloads,” positioning it as a strong competitor in a crowded AI market.

Why AI Agents Will Be “Bigger Than the Internet”

Garman closed by reiterating his boldest claim: AI agents represent a platform shift on par with—or greater than—the Internet itself. With agents embedded across applications, infrastructure, and customer interactions, AWS believes this technology will reshape how businesses operate at every level.

As AWS re:Invent 2025 made clear, the age of AI agents is no longer theoretical—it’s already arriving at scale.