Federal AI Procurement under M-25-22: Navigating the 'Buy-First' Efficiency Mandate

The Shift from M-24-18 to M-25-22

In April 2025, the Office of Management and Budget (OMB) issued Memorandum M-25-22, titled "Driving Efficient Acquisition of Artificial Intelligence in Government." This memorandum effectively replaces the previous M-24-18 and signals a new era of federal AI procurement focused on speed, efficiency, and commercial-first strategies.

For contractors, the most significant takeaway is the government's renewed emphasis on the "Buy" side of the "Build vs. Buy" equation. Agencies are now directed to prioritize the acquisition of existing commercial AI systems and services rather than embarking on lengthy, custom-built development projects. This is a direct effort to leverage the rapid innovation cycles of the private sector.

The Role of the Chief AI Officer (CAIO)

Under M-25-22, the role of the Chief AI Officer (CAIO) has been elevated. The CAIO is now central to the acquisition process, responsible for ensuring that AI deployments align with the agency's strategic plan and meet minimum risk management practices.

Contractors must tailor their business development efforts to address the CAIO’s primary concerns:

• Data Rights and Vendor Lock-in: Agencies are instructed to retain sufficient data rights to avoid being tied to a single provider. Proposals that offer flexible data portability and interoperability will have a competitive advantage.
• Performance Claims Verification: M-25-22 mandates that agencies regularly evaluate contractor performance claims. Expect to see more "Show Me, Don't Tell Me" requirements in solicitations, such as technical demonstrations or sandboxed testing environments.
• FedRAMP Prioritization: The memorandum encourages coordination with FedRAMP to prioritize generative AI offerings that meet federal security standards.

Navigating AI Risk Management

Acquisition strategy is now inextricably linked to risk management. The NIST AI Risk Management Framework is the gold standard for federal agencies. Contractors should be prepared to provide detailed documentation on how their AI models were trained, how they mitigate bias, and how they ensure the security of government data.

Furthermore, FAR Part 15 principles regarding "Contracting by Negotiation" are being applied to AI with a focus on "best value" rather than lowest price. Agencies are looking for partners who can demonstrate not just a functional tool, but a secure, compliant, and scalable AI ecosystem that adheres to the latest OMB mandates.