Federal agencies are putting more structure around how they buy artificial intelligence, and that shift could affect whether Black tech vendors compete for meaningful work. The Government Accountability Office has reviewed how agencies acquire AI and identified acquisition challenges. The Office of Management and Budget has issued governmentwide guidance for AI buying. For Black-owned and Black-led firms, the opening depends less on AI buzz than on how agencies define work, write solicitations, assign data rights and leave room for specialized vendors. GAO ↗ White House ↗

This is an access analysis, not proof that federal AI procurement has already helped or harmed Black-owned companies. Public award data can show prime contractors, contract vehicles and some recipient characteristics, but it does not provide a simple, complete measure of Black-owned AI vendor participation across primes and subcontractors. It also may not clearly show whether a task order involves AI implementation. USAspending.gov draws contract and assistance data from multiple federal systems, including prime award and subaward sources, but those records do not create a complete public count of Black-owned AI vendors. USAspending.gov ↗

The buying rules still matter. If agencies purchase AI mostly through large platform deals, smaller vendors may have to compete for the services that make those tools useful, secure, accessible and accountable inside government workflows. Depending on the mission, that work may include data cleanup, cybersecurity, testing, training, accessibility, case management and risk controls.

One public-sector technology example shows the kind of implementation work agencies may need. Fearless, a government technology firm, described work on SBA.gov, Login.gov, Search.gov and technology used by Medicare beneficiaries in prepared testimony before the Senate Committee on Small Business and Entrepreneurship. Congress.gov ↗

Fearless is relevant here as an example of mission-focused digital delivery, not as a claim about the company’s current ownership, certification or socioeconomic status. Black-owned, Black-led, small business, Small Disadvantaged Business, 8(a) and HUBZone are related but separate categories. Eligibility can change, and buyers or partners should verify status through federal registration systems, SBA certification records, GSA sources and the company before relying on any designation.

The company’s AI-assisted alt-text case study also points to a practical use case. Fearless says it built a tool for the Smithsonian’s National Museum of African American History and Culture that generated draft image descriptions, returned structured alt text for human review and preserved final review by the museum team. Because that account comes from the company, readers should treat the performance claims as company-reported unless confirmed in procurement records or by the agency. Fearless ↗

Federal AI buying is getting more formal

GAO said agencies acquire AI through new contracts, existing vehicles, products, services and other arrangements. The watchdog also identified acquisition challenges involving technical expertise, data and intellectual-property rights, requirements definition, testing, cost and acquisition timelines. GAO ↗

OMB’s memorandum M-25-22, “Driving Efficient Acquisition of Artificial Intelligence in Government,” tells agencies to pursue timely and cost-effective AI acquisition while supporting competition, performance management and risk management. The memo also tells agencies to consider portability, interoperability, sourcing and long-term dependence on a single vendor. White House ↗

Those instructions could help smaller firms if agencies carry them into solicitations. Guidance that supports market research, realistic testing, performance-based acquisition and competition can give capable firms a chance to compete on domain knowledge, user experience, accessibility, data governance and delivery. It may not open the market if agencies define AI mainly as access to large platforms.

GSA’s “Buy AI” page points agencies to purchasing resources and acquisition channels for AI and related services. Because GSA listings and vendor offerings can change, buyers and contractors should check the live page before relying on any specific offering. GSA ↗

For smaller contractors, centralized technology purchasing can help or hurt. A firm may build on commercial AI tools instead of funding its own foundation model. But if agencies centralize the platform purchase, smaller vendors may need to compete around integration, governance, training, testing, accessibility, data engineering and mission implementation.

For more context on public-sector growth strategies, see Black Biz Daily’s coverage of [Black-owned businesses in federal contracting](/black-owned-businesses-federal-contracting/) and [government procurement strategy for small firms](/government-procurement-small-business-strategy/).

Solicitation design will shape access

For Black tech vendors, the issue is not whether Washington says it wants innovation. The issue is how agencies write statements of work, score proposals, structure pilots, assign data rights and evaluate performance after award.

A procurement process that favors modular systems, measurable outcomes, interoperability and human oversight may give specialized firms more room to compete. An agency might buy a major AI platform through a governmentwide agreement, then hire another contractor to design an accessible workflow, test outputs with real users, document risks and train staff.

A procurement process that bundles too much into one enterprise deal can produce the opposite result. If a platform vendor or incumbent prime controls the full stack, smaller firms may have fewer chances to win prime contracts and build direct past performance.

OMB’s memo addresses part of that risk by telling agencies to consider competition, portability and vendor lock-in. It also supports acquisition planning that includes relevant technology, cybersecurity, data, privacy, legal and mission officials.

NIST’s AI Risk Management Framework gives agencies and vendors a voluntary structure for discussing trustworthy AI design, development, use and evaluation. The framework does not guarantee fair procurement outcomes. It does give buyers and vendors common language for risk, testing, governance and monitoring. NIST ↗

That shared language may help firms that can document their methods. It may also raise the compliance burden for companies that lack large proposal, legal and policy teams.

Small-business tools still matter

Federal small-business programs remain important, even as AI changes the technology market. SBA says certification programs can help eligible firms compete for federal contracts through set-asides, sole-source awards and other contracting tools, depending on the program and procurement. Certification can support access, but it does not guarantee contracts. SBA ↗

The legal environment has shifted, especially for 8(a). In a 2023 federal district court ruling in *Ultima Services Corp. v. USDA*, the court enjoined SBA from using a rebuttable presumption of social disadvantage in administering the 8(a) program. The ruling did not eliminate 8(a), but it pushed SBA away from race-based presumptions and toward individualized showings of disadvantage. Justia ↗

That distinction matters for Black business owners. Certification can support market access, but it cannot replace capability, compliance, teaming relationships, past performance and a clear technical approach. In AI procurement, vendors should expect buyers to scrutinize data practices, human review, error correction, cybersecurity, testing and performance.

Some Black-owned firms may seek prime work through Multiple Award Schedules, governmentwide acquisition contracts, blanket purchase agreements, 8(a), HUBZone or other vehicles where eligible. Others may subcontract to larger integrators or platform providers. Subcontracting can help firms enter the market, but companies should examine workshare, customer access and past-performance credit before signing teaming arrangements.

For related context, see Black Biz Daily’s coverage of [SBA 8(a) changes for Black business owners](/sba-8a-program-black-business-owners/) and [Black tech founders pursuing AI opportunities](/black-tech-founders-ai-opportunities/).

Proof will matter more than buzzwords

Black-owned and Black-led technology firms may gain more traction by showing how they make AI work safely in an agency context than by simply saying they use AI.

That means vendors need more than a demo. They may need documentation on what data a system uses, how outputs are tested, where humans review decisions, how errors get corrected, how costs scale, who controls data and how the system avoids locking the agency into one ecosystem.

Fearless’s alt-text case study suggests one possible model: a specific tool that supports human workers instead of replacing judgment. If agencies independently validate similar outcomes in procurement records and performance reviews, that approach would align with federal guidance that stresses defined use cases, human oversight, testing, accessibility, cost control and integration with existing work.

GAO also identified a learning gap. The watchdog said agencies had not systematically collected lessons learned from AI acquisitions, while noting an OMB expectation for agencies to share acquisition knowledge through a GSA-managed repository. GAO emphasized best practices involving contract clauses, data rights, testing requirements and lessons learned. GAO ↗

That gap may affect business access. If the government cannot show which AI acquisition strategies produced stronger competition, safer systems or broader vendor participation, efficient acquisition may not translate into wider access.

Federal AI guidance will not automatically diversify the technology market. It can open doors if agencies buy AI as modular, testable and interoperable work with room for specialized firms close to the mission. It can narrow the market if speed mainly means centralized access to the biggest platforms and incumbent integrators.

For Black tech vendors, the next federal AI cycle will require more than technical fluency. Firms will need to document trust, manage data rights, build around human review, partner strategically and prove outcomes in public-sector settings. Whether Washington makes enough room for those firms remains an open procurement question.