Artificial intelligence is rapidly moving from the experimental stage in government to something much more consequential for the contracting community. Public officials are now buying AI, incorporating it into government operations, developing rules for its use and beginning to address the financial, security and oversight issues that follow widespread adoption.
Recent developments at the federal and state levels indicate that an entirely new contracting market is taking shape. The opportunities will extend well beyond companies with AI offerings. Systems integrators, cybersecurity firms, consultants, auditors, cloud providers, data specialists and companies capable of monitoring costs and performance will all find new demand for their offerings.
One of the clearest signals came from the U.S. General Services Administration (GSA) on Sept. 10. The agency announced a new 27-month OneGov agreement for artificial intelligence services that is scheduled to take effect Oct. 1 and continue through Dec. 31, 2028. The agreement will provide a 50 percent discount on consumption-based token usage, with no platform-access fee, minimum order or spending commitment. Eligibility will extend to federal, state, local and tribal governments, and purchases may be made directly, through resellers and/or through supported cloud marketplaces.
The numbers associated with the government’s broader AI purchasing initiative are significant. The federal General Services Agency (GSA) reports that its OneGov program has generated approximately $1.68 billion in federal cost savings since its inception. Approximately $1.4 billion of that amount is attributed to AI agreements that provided access to AI tools for about 3.5 million federal employees.
But perhaps the most interesting part of the new agreement is not the discount. It is the change in the way government will pay for AI.
Consumption-based pricing means that agencies will increasingly have to determine how much AI they expect employees and systems to use and then monitor that consumption. GSA is already warning agencies about this issue. Its current AI procurement guidance notes that usage costs can grow quickly without appropriate controls, and officials are advising agencies to establish usage limits and regularly review consumption reports.
That process will, of course, create another potential contracting opportunity. The same type of financial management discipline that developed around cloud computing will increasingly be needed for artificial intelligence. Government officials will need assistance forecasting consumption, allocating AI costs among departments and programs, establishing spending controls, tracking usage and determining whether the technology is producing enough value to justify its cost.
Another indication of how quickly the market is changing came this week from the U.S. Department of Transportation (USDOT). On Sept. 17, the department’s Intelligent Transportation Systems Joint Program Office hosted a three-hour national webinar devoted specifically to “Demystifying AI Procurement.”
The webinar was designed for state and local transportation agencies, metropolitan planning organizations, transit agencies and tribal governments. The program addressed AI-specific acquisition risks, pre-award procurement issues and collaboration among technical, information technology, legal, finance and acquisition personnel.
USDOT is doing more than discussing AI procurement. Its Intelligent Transportation Systems Program is supporting projects involving AI and automation. Proposals are currently under review for an intersection-safety prototyping initiative worth up to $17 million. The initiative includes sensor fusion, AI-based decision-making and conflict mitigation. Another solicitation offers up to $5 million for cooperative driving automation prototype development, testing and deployment in partnership with a state or local agency. Proposals for that funding are due Sept. 28.
California provides another glimpse of what the emerging marketplace may look like.
On Sept. 9, Gov. Gavin Newsom signed Senate Bill 813 and Assembly Bill 1405, and the legislation establishes a framework for independent third-party audits and assessments of AI systems. The state is describing this as the nation’s first framework of its kind. The legislation establishes independent verification organizations and creates an AI auditor registry along with standards addressing independence, transparency and integrity.
The state’s Government Operations Agency is directed to develop application requirements and designation criteria for independent verification organizations by Jan. 1, 2028. A separate AI Auditor Registry is scheduled to be established no later than Jan. 1, 2029.
That development is significant for the contracting community because government adoption of AI will require more than implementation. Public entities will also need to demonstrate that systems operate appropriately, data is protected, risks are identified and results can be independently evaluated.
An entirely new professional services market could emerge to correspond to all these requirements. Accounting and consulting firms, cybersecurity companies, technology assessors and organizations with experience performing independent compliance reviews should watch this development closely. Firms that understand existing AI risk-management and security frameworks may have an opportunity to establish credentials well before government demand reaches maturity.
State officials in California, on the same day the new audit legislation was signed, announced a pilot for AskCA, an AI-powered digital assistant designed to help residents navigate state and local government services. The system will provide a single entry point for individuals seeking assistance with such needs as employment, starting a business, family services and disaster recovery.
The state’s Office of Data and Innovation developed the AskCA knowledge base with security support from the California Department of Technology. State officials plan to put the technology into another public-facing application on Sept. 30, when the California Department of Human Resources launches an AI-powered job-skills-matching feature for state employment.
These developments illustrate how quickly the AI government market is evolving. Agencies are no longer asking whether they should experiment with artificial intelligence. They are already doing it. However, they must determine what to buy, how to procure it, how to integrate it with existing systems, how much its use will cost, how government data will be protected and how the technology’s performance will be evaluated.
Each question creates potential work for the contracting community.
Technology firms will obviously compete to provide AI products, but a much larger ecosystem will be required around those products. Opportunities are likely to emerge for data modernization, systems integration, cloud infrastructure, cybersecurity, employee training, acquisition assistance, cost management, testing, risk assessments, independent audits and ongoing compliance and performance monitoring.
For government contractors, that may be the most important development to watch. The next wave of AI spending will not just be about purchasing models or software licenses. It will involve the infrastructure, professional services and oversight required to implement and manage the technology.
The new mandates for government officials will be to manage the process of procuring AI, securing it, paying for it, monitoring its use and proving that it works as promised. The contracting marketplace forming around those needs may ultimately become much larger than the market for AI tools themselves.
