AWS Summit DC 2026: From Possibility to Production for the Public Sector

The keynote: If that's possible, what else is?

Dave Levy, AWS Vice President of Worldwide Public Sector, Healthcare and Life Sciences

I was there for the keynote, and Levy opened with a number instead of a slogan: most enterprise AI projects still never make it into production. That single stat reframed everything AWS announced for the rest of the day.

What the keynote was really about

Levy tied the moment to the country's own story, timed to next year's 250th anniversary. He talked about the early founders who saw one breakthrough and immediately asked, "if that is possible, what else is?" It's the same instinct AWS is betting public sector leaders have right now, not a shortage of ideas, but a need for the infrastructure and funding to turn those ideas into systems that actually run.
The keynote organized around three practical goals: accelerate modernization, empower developers and secure organizations. It's the same three-part framework Deepak Singh used to open the LA Summit a few weeks earlier: modernize, empower developers, secure. That's not a coincidence. It's AWS's core narrative for 2026, public sector included.

Classified cloud, opened up

The biggest security announcement was AWS Secret Cloud for Industry (ASCI). For the first time, cleared defense contractors can run contractor-owned classified workloads directly on AWS infrastructure, inside an environment built to the same standards the Department of War already trusts. Northrop Grumman is the first partner live on it, and AWS is backing it with up to $20 million in credits over the next three years. Levy's point was blunt: infrastructure built to run email was never going to be enough to run AI on sensitive data. Agencies need a cloud designed for that job from the start.

Paying agencies to modernize

AWS also launched the IC Accelerated Modernization Framework (ICAMF): up to $1 billion in credits, available through October 2030, tied directly to intelligence community agencies migrating qualified workloads to AWS. It builds on the $50 billion infrastructure expansion AWS announced last fall. CIA Director John Ratcliffe joined the keynote for his first public remarks since taking office, confirming the agency will use ICAMF as part of what he called a technology renaissance inside the CIA: acquisition timelines cut from a year or two down to under six months, and cyber elevated to its own Mission Center.

AI for the next big science push

Secretary of Energy Chris Wright joined Levy on stage to talk about the Genesis Mission, the Department of Energy's push to bring AI to bear on energy, science and national security. AWS is working with Idaho National Laboratory on a digital twin of a small modular reactor, and the National Nuclear Security Administration announced the first cloud environment authorized to process Secret/Restricted Data.

Engineers, not just endpoints

The other headline investment was AWS Forward Deployed Engineering (FDE), backed by another $1 billion. It's a new global team that embeds AWS engineers directly with customers to build and ship production AI systems, using what AWS calls an AI-Driven Development Lifecycle. AWS points to real numbers behind the model already: Lyft cut driver support resolution time by 87% working this way, and the Allen Institute, Cox Automotive, the NBA, Ricoh, Southwest Airlines and the NFL are already working with the team. The model is explicitly not traditional consulting; the goal is to leave customers self-sufficient, with new systems, skills and knowledge graphs of their own.

Proof it's already working

The clearest public sector example came from outside the US. UK Government CTO Sonia Patel joined the keynote to share how the UK is deploying AI at national scale, across hundreds of departments and 67 million citizens. HMRC, the UK's tax authority, is investing more than £450 million to migrate three legacy data centers to AWS, aiming to help close a £47 billion tax gap while improving things for citizens filing returns. She also pointed to GOV.UK Chat, an AI assistant already answering citizen questions directly inside the government's app, tested with more than 10,000 users before launch. On the research side, the Fleming Initiative, a partnership between Imperial College London and Imperial College Healthcare NHS Trust, is using AWS to build a global intelligence platform against antimicrobial resistance, a public health threat that doesn't get nearly enough attention.

A practical win for citizen call centers

Not every announcement this week needed a security clearance to matter. AWS also expanded Amazon Connect, its contact center platform, with a new identity verification integration from CLEAR: callers can opt in to verify themselves through a secure text link before they ever reach an agent, and the agent sees that verification the moment the call connects. No more working through a script of security questions for a password reset or account change. For any citizen-facing call center, DMV, benefits, 311, that's less fraud exposure and shorter calls, and it's one of the few things from this Summit that state and local agencies can put to use immediately, no classified cloud or intelligence community contract required.

The honest takeaway

When we covered this Summit last year, the message was that it was time to build. This year, the message was about what happens after you build: can it run, at scale, inside real security and budget constraints. That's a meaningful shift, and it showed up in almost every announcement: not new demos, but production commitments with dollar figures and delivery dates attached.
Worth saying plainly: this year's headline moments leaned further into federal, defense and intelligence community stories than state and local government, K-12 or higher ed, the corners of public sector our team spends most of its time in. But the infrastructure being proven out here doesn't stay contained. The same modernization funding models, secure landing zones and embedded-engineering patterns AWS is testing with the IC and DOE tend to show up in what's available, and expected, everywhere else in public sector within a year or two.

Where WWT fits in

WWT is an AWS Premier Tier Services Partner, and our Advanced Technology Center lets us validate AWS capabilities against real-world environments before a customer has to commit to them. We stay in it from strategy through migration, security and ongoing support, which matters more this year than most, since production was the whole point of the Summit.

What we're helping customers do right now

Modernization and migration funding

Programs like ICAMF are new, but the underlying question isn't; it's how you build a migration plan a budget office will actually approve. We help agencies put that business case together.

Security and compliance for sensitive workloads

Classified cloud access is opening up fast. Getting the architecture, governance and compliance right before day one is still the hard part, and it's where we spend most of our time with defense and intelligence customers.

Moving AI from pilot to production

The gap between a working demo and a system an organization can actually run on hasn't closed by itself. We help agencies line up proof-of-concept funding to properly evaluate an AI pilot first, then carry that momentum into AWS's Migration Acceleration Program (MAP) funding once they're ready for production, one partner covering both ends of that journey, data architecture, governance and security included.

Public sector and SLED

Federal announcements set the tone this year, but state, local and higher ed customers are asking the same questions about funding, security and production-readiness. Our SLED team was there for the Summit, and that momentum carries straight into the second half of the year.

Final thought

The infrastructure argument is settled. AWS spent this Summit proving it has the classified cloud, the funding and the engineering talent to back the shift to production. The harder question left standing, for a defense contractor, an intelligence agency or a state IT office alike, is whether the organization on the other end is ready to run with it. That's the conversation we're built for. If you're ready to have it, we are too.