GSA Releases AI Acquisition Guide in Line with Anticipated GenAI Spending

Published: May 01, 2024

Federal Market AnalysisArtificial Intelligence/Machine LearningInformation TechnologyPolicy and LegislationSpending Trends

The guidance provides an acquisition framework for generative AI technologies, supporting the growth in federal AI spending seen in the last three fiscal years.

The General Services Administration (GSA) released an AI guidance earlier this week to serve acquisition professionals in the procurement of generative AI (GenAI) at civilian agencies. The Generative AI and Specialized Computing Infrastructure Acquisition Resource Guide is in response to Section 10.1 of the White House AI EO. The guide release comes at the heels of the EO’s six-month anniversary.

The newly issued document distinguishes GenAI acquisitions from other IT acquisitions, with additional practices acquisition officials should consider with such transactions. Considerations include:

The guide also promotes several acquisition methods and vehicles for GenAI procurement. Among the listed acquisition methodologies are Cooperative Research and Development Agreements (CRADAs), Interagency Agreements, Other Transaction Authorities (OTA), Commercial Solutions Opening (CSO) and Prize Competitions. The guide also refers to GSA Multiple Award Schedules (MAS) IT, in which vendors have achieved a series of compliance check and pre-negotiated pricing. Additional procurement vehicles the document promotes as most useful for GenAI purchases include:

The issued document also provides a guide to acquisition officials should agencies decide to create specialized computing infrastructure to support GenAI tools. Specialized computing infrastructure is defined by GSA as, “high-performance computers, powerful chips, software, networks and resources made specifically for building, training, fine-tuning, testing, using and maintaining artificial intelligence applications.” The guide provides federal procurement personnel with a roadmap to conduct needs assessments, cost estimates, and compliance when moving forward with the acquisition of specialized computing infrastructure for GenAI.

Increase in GenAI Acquisitions Set to Expand Federal Spending

Each year, Deltek analyzes federal obligations and identifies those transactions and contracts related to AI services and products. Deltek’s preliminary analysis of FY2021-2023 AI spending saw the debut of some GenAI-related transactions, with more expected in the following years as agencies grow in GenAI usage. Sample GenAI-related obligations include:

  • FY23 Generative AI Proof of Concept for the Office of the Chief Information Officer at USDA
  • Microsoft Azure Cloud Hosting for Virtualized Machines and Generative AI Engineering Research Support at Commerce
  • Generative AI for Rapid Scene Understanding and Target Recognition at the Air Force

Initial analysis of FY 2021-2023 AI spending reveals a $4.6B federal AI market in prime contract obligations, growing 31% from FY 2021 to 2023. Specifically, prime obligations grew 31% from FY 2021 to 2023.

Top departments in AI prime obligations from FY 21-23 include DOD ($2.0B), Air Force ($845M), NASA ($392M) and Army ($325M). Separate analysis of OTA spending from FY 2021-2023 revealed $2.2B market in AI-related transactions, led by DOD ($1.1B), Army ($779M) and the Air Force ($215M).

A breakdown of FY 2021-2023 AI obligations among the top ten contract vehicles by dollar value reveals many of the Government Wide Acquisition Vehicles that GSA recommends for GenAI procurements in the latest guidance.

Though the onset of GenAI’s expansion in the market caused some agencies to initially exclude the tool, while others took cautious steps to explore it, GSA’s GenAI acquisition framework opens the doors for agencies to procure GenAI without large reservations. With that, the increase in GenAI-related acquisitions throughout the federal space will contribute to the ongoing growth in agency AI spending seen in the last several years.