The Summit opens two days after the President signed a post-quantum cryptography mandate. That is not coincidence. It is a compressed version of what the public sector technology agenda looks like in 2026.

Attending this event in prior years, the sessions felt frozen in a different era. Cloud was the answer. Generative artificial intelligence was arriving. Pilots were underway. The tone was aspirational in a way that the product catalog had already outpaced. What you heard on stage and what Amazon Web Services had shipped occupied different time zones.

The 2026 Summit opens under different conditions, and the gap between the conference agenda and the policy environment has closed in the opposite direction. Three things happened before Dave Levy took the stage on June 30 that will shape every conversation in the building.

Executive Order 14409: The Quantum Clock Started Last Monday

On June 22, President Trump signed Executive Order 14409, "Securing the Nation Against Advanced Cryptographic Attacks." The order sets December 31, 2030, as the deadline for federal agencies to migrate high-value assets and high-impact systems to post-quantum cryptography for key establishment, and December 31, 2031, for digital signatures. Federal contractors face the same timeline under Federal Acquisition Regulation amendments the order directs within 180 days.

The coverage fixated on the dates. The more operationally significant line in the order is quieter: agencies must inventory their cryptographic assets and submit migration plans. Most cannot do that today. For thirty years, cryptography spread through enterprise systems the way plumbing spreads through a building that keeps getting renovated. Signing keys generated for a project that shipped in 2014 and never died. Transport Layer Security libraries bundled into vendor products with no documentation of where they live. Nobody drew a map because nobody had to.

EO 14409 put a date on a reckoning that was always coming. The quantum threat is the trigger. The underlying condition is decades of accumulated cryptographic debt that no agency was ever forced to account for.

The same day Trump signed EO 14409, he signed a companion order, EO 14411, directing federal investment in quantum computing capability. The pairing is intentional: one order accelerates the threat, the other mandates defense against it. For federal technology buyers at the Summit, quantum is no longer a future-planning session. It is a compliance deadline with contractor exposure attached.

AWS has a direct stake in how agencies respond. The AWS GovCloud regions run on Federal Information Processing Standards 140-3 validated cryptographic endpoints today. The migration question for agencies is not whether AWS can support post-quantum standards. It is whether the agency's own applications, integrations, and vendor contracts can be inventoried and migrated within four years inside a procurement structure that was not designed for that speed.

Sovereignty Is Not a Marketing Claim Anymore

The conventional read on GovCloud has been that it is a compliance tier: a more expensive, more isolated version of commercial AWS for workloads that require it. That framing misses what GovCloud has become structurally.

GovCloud is physically and logically isolated from all commercial AWS regions. Administrative access is restricted to vetted U.S. citizens. Data in GovCloud remains on U.S. soil and is not handled by foreign personnel. Connectivity is separate, with distinct account credentials and unique endpoints that have no path to commercial tenancy. For organizations subject to International Traffic in Arms Regulations, Export Administration Regulations, or Controlled Unclassified Information requirements, GovCloud provides a compliance boundary at the infrastructure level rather than a policy assertion on top of shared infrastructure.

That distinction matters more in 2026 than it did two years ago. The Clarifying Lawful Overseas Use of Data Act, the CLOUD Act, means selecting a European commercial region of any U.S. hyperscaler does not resolve jurisdictional exposure. For domestic federal workloads, GovCloud's sovereignty posture addresses a structural risk that policy overlays on commercial cloud cannot. The recent expansion of AWS Cloud Wide Area Network capabilities into both GovCloud regions extended the same policy-driven network automation available in commercial regions to classified workloads, including multi-region disaster recovery architectures across GovCloud East and West.

The customer success stories at the Summit will reflect this shift. The agencies presenting are not describing pilot programs. They are describing production workloads that required the sovereignty boundary to proceed at all.

The Infrastructure Commitment Is Breaking Ground, Not Being Announced

AWS committed up to $50 billion in November 2025 to build AI and supercomputing infrastructure for U.S. government customers. The company describes it as the first high-performance computing infrastructure purpose-built for federal workloads. Construction begins this year. The capacity, 1.3 gigawatts across GovCloud, AWS Secret, and AWS Top Secret regions, is accompanied by expanded access to Amazon SageMaker for model training, Amazon Bedrock for model and agent deployment, Anthropic Claude, Amazon Nova, and AWS Trainium chips alongside NVIDIA infrastructure.

That is a supply-side investment. It builds capacity. It does not restructure agency procurement timelines, retrain contracting officers on agentic system evaluation, or modernize data infrastructure that agents need to operate against mission-relevant information. Some of those legacy systems date to the 1960s and 1970s, maintained by a workforce approaching retirement age.

The agencies that will use what AWS is building are not the ones waiting for the Summit keynote to tell them it is ready.

What the Customer Success Stories Will Show

The 350-plus sessions at the Summit include a significant volume of customer-led content. AWS has emphasized this specifically: real solutions to real challenges, from agencies that have moved through the pilot phase. That framing is worth taking seriously, because the distance between pilot and production in the public sector is where most AI programs have stalled.

Production-level deployment in a federal environment requires data architecture that can support autonomous agents, workforce fluency to supervise agent outputs within regulatory accountability structures, and contracting vehicles that allow iteration rather than locking a configuration at time of award. Covered California's use of AWS to automate contact center operations with agentic AI while managing data pipelines for healthcare coverage determination is an example of what that looks like when it works. The U.S. Air Force's predictive maintenance deployment using AWS and C3.ai is another. Both required solving the data and workflow problems before the AI could operate.

The customer stories worth attending are the ones that name what broke first, not the ones that lead with outcomes.

The Question the Keynote Will Not Close

Dave Levy's keynote brief covers how AWS delivers the most secure cloud for AI and government workloads, with agentic AI solutions for modernization, developer enablement, and national security and defense missions. That framing assumes the viability question is settled. For the agencies in the room that have already crossed into production, it is. For agencies still at the pilot stage, the gap is not the infrastructure.

The harder conversation is one that conference formats handle poorly. EO 14409 gave agencies 30 days to designate a post-quantum cryptography migration lead, 90 days for binding guidance from the Office of Management and Budget, and 180 days for a contractor Federal Acquisition Regulation rule. Those timelines start from June 22. They do not wait for the fiscal year, the next budget cycle, or the outcome of a technology evaluation process.

Agencies that leave the Summit with a conference badge and no cryptographic inventory underway will face that timeline regardless.