Amazon Web Services has expanded the availability of third-party large language models within its AWS GovCloud (US) Region, giving government agencies access to models from Anthropic, Meta, OpenAI, xAI, and NVIDIA through the Amazon Bedrock platform. The move also includes Amazon's proprietary Nova series models, all operating within isolated cloud environments built to meet federal regulatory requirements.
AWS said in a statement: "We are pleased to announce that AWS GovCloud (US) now offers multiple AI model families through Amazon Bedrock, providing government customers, their industrial base, and technology partners with the AI model choices necessary to meet their mission requirements."
The expansion arrives as AWS posts its strongest growth in years. In the second quarter of 2026, AWS reported revenue of $42.2 billion, up 36.7% year-over-year - the fastest growth rate in 18 quarters. Operating income reached roughly $16.6 billion, a 63% jump, with an operating margin of 39.4%. AWS now contributes about 21% of Amazon's total revenue but accounts for more than 60% of its operating income.
AI business momentum
Both AWS's AI business and its custom chip business have surpassed $25 billion in annualized revenue run rates, each recording triple-digit percentage growth. The Bedrock model hosting platform serves hundreds of thousands of customers, and single-quarter spending on Bedrock in Q2 2026 exceeded the cumulative total of all previous quarters combined.
AWS's backlog reached $496 billion in the second quarter, a new record. Company management expects AWS's future revenue scale to reach the trillion-dollar level, noting that 85% of global IT spending still sits in on-premises data centers.
Amazon has committed to investing up to $50 billion in 2026 to add roughly 1.3 gigawatts of AI and high-performance computing capacity across secure cloud regions in the United States, including AWS GovCloud, AWS Secret, and AWS Top Secret. For government professionals working with Generative AI and LLM tools, this means the models they may already use in commercial settings are now available in compliant federal environments.
Fifteen years of GovCloud
This August marked the 15th anniversary of the launch of AWS GovCloud (US). Built as the world's first cloud region designed specifically for U.S. government security and compliance needs, GovCloud has grown since 2011 into an operational cornerstone for missions in national security, energy, public health, and financial services. Its workloads include processing Mars exploration data for NASA's Jet Propulsion Laboratory's Curiosity rover and providing real-time orbital analysis for the Artemis II lunar flyby mission.
The addition of major third-party models to GovCloud gives agencies more choice in how they deploy AI, rather than locking them into a single vendor's approach. For those exploring AI for Government applications, the practical effect is straightforward: the same frontier models used in the private sector can now be applied to federal workloads without leaving a compliance-controlled environment.
Why this matters for government professionals
For government employees and contractors, the practical takeaway is that AI procurement options just widened. Agencies can now select from multiple model families - Anthropic, Meta, OpenAI, xAI, NVIDIA, and Amazon's own Nova series - within a cloud environment built specifically for federal security requirements. That means mission teams can match a model to a task rather than working around a single mandated option. The $50 billion infrastructure investment also signals that AWS expects government AI demand to keep growing, not taper off.
Your membership also unlocks:
