Is your government ready to buy AI?

Governments are procuring artificial intelligence at a growing pace. In 2025, the UK spent twice as much on AI as the previous year, awarding 521 public contracts worth a total of £1.17 billion. Between 2022 and 2024 alone, U.S. federal agencies committed $5.6 billion to AI projects. How those dollars are spent will determine whether AI truly improves public service delivery, or deepens inequities, opacity, and vendor dependency. 

People grappling with these challenges are asking us more often how governments might spend their money in ways that have a positive influence on markets and reflect their values, while mitigating risks and delivering the best results. These questions come not a moment too soon: Money is at stake, and so is public trust.

Over the past few weeks, I’ve had the opportunity to present Open Contracting Partnership’s new Buying AI guide, hear about the on-the-ground challenges in procuring AI from many governments, and learn from others at several events around procuring AI for the public sector. 

For government leaders looking for best practices in AI procurement, these are the five priorities we recommend focusing on: 

• Think through your AI organizational strategy now. A good organizational strategy can help you develop a coherent position on critical questions like governance mechanisms, policies around data use, and transparency and accountability. These decisions should be made long before any one procurement process takes place. 
• Build your staff’s baseline understanding of AI. A lot of the opportunities and risks around AI are only realized during contract implementation. This means that it’s especially important for the procurement staff who are involved in the procurement process, as well as any eventual users of the technology, to have some understanding of how the technology works. Don’t outsource your product technology strategy to vendors! (This knowledge and responsibility should also not rest with your technology staff alone.) To get you started, we’ve summarized the basic AI concepts on one page.
• Bring your technology, procurement, and program staff together early and often. Take a look at your processes and practices for technology procurements. Right now, technology and procurement typically work in silos and are not pulled into planning processes as early as they should be. Buying AI well requires the best of everyone’s expertise and asking the right questions. 
• Get your data management house in order. Too often leaders want to skip this step and jump straight to using the flashy new tools at their disposal. This is a mistake. As with any technology tool, when it comes to AI, it’s garbage in, garbage out. Meanwhile, the governments that have good data management practices, including promoting interoperability through the use of APIs and building up critical internal capacity to manage the AI contracts, are in a position to take real advantage of AI technology. We’ll have a post up soon that explores in detail the case for better data management, particularly for AI projects. 
• Support agile, innovative, and outcomes-driven procurement practices through open competition. Good AI procurement looks like great tech procurement. But many governments aren’t yet able to pursue the kind of agile and innovative procurement methods that can open the door to working with new vendors, support learning and innovation, and fight vendor lock-in. 

Deciding whether to build or buy the infrastructure is another important consideration. While there are no hard and fast answers here, much depends on your specific goals and internal capacity. Even for large, high-capacity governments, it’s realistically going to make the most sense to pursue a blend of building, buying, and customizing AI (see this review for more insights and our handy overview table of use cases on ways to buy AI).

Given the pace at which AI is changing, what really matters is that governments procure for outcomes rather than a specific technology. 

How OCP can help 

We recognize that tackling these challenges is easier said than done. That’s why we’ve developed the guide, Buying AI: tips and tools for the public sector. 

We will continue to share our insights in this blog series on AI in procurement, covering topics like  the value of structured data for AI, and case studies from our partners in places like Brazil and Ukraine, who are leveraging AI to strengthen procurement. We also plan to build out our guidance with even more specific resources. 

Want to learn more in the meantime? You can check out the recordings of our latest sessions, including our conversation around emerging trends with George Washington University Government Procurement Law Studies’ Jessica Tillipman and U.S. Digital Response’s Michael Boyce, and a webinar training hosted by InnovateUS with OCP and the State of Maryland’s Former Senior AI Advisor, Nishant Shah.

Reach out to us for one-on-one-technical assistance on your procurement reforms and anything digital, data and AI! We’d be happy to help you think through your next steps.