Beginning with the FY2020 budget request, NITRD has annually reported consolidated Federal nondefense investments in AI R&D. This reporting began in response to Executive Order 13859, “Maintaining American Leadership in Artificial Intelligence”, which required a standardized approach to accurately account for AI R&D investments across the Federal government. At that time, NITRD worked with Federal agencies to create a new Program Component Area (PCA) for AI and also account for investments in AI R&D that are captured under other PCAs (refer to Table 1, footnote 4, for examples of distinctions between AI and other PCAs). This accounting aligns AI R&D investments with the key strategies in The National Artificial Intelligence Research and Development Strategic Plan: 2019 Update and enables tracking of nondefense AI R&D investments from year to year. The National AI Initiative Act of 2020 also requires consistent reporting of AI R&D investments, to include investment in National AI Research Institutes, which are included for the first time in the FY 2022 report.
Table 1 shows two types of AI R&D investments: (1) those with a primary emphasis on AI R&D, which are reported under the AI PCA, and (2) those with primary emphases in areas other than AI R&D, which are reported in other PCAs. The total FY2022 requested funding level for nondefense AI R&D, $1,670.2 million, indicates the total of the nondefense Federal programs directly related to AI ($1,108.5 million) and the nondefense AI-related efforts reported in the other PCAs ($561.7 million). This total Federal FY2022 nondefense budget request in AI is an 8.8 percent increase over the FY2021 nondefense enacted investments of $1,534.4 million and an 11.2 percent increase over the FY2021 nondefense budget request of $1,502.5 million. An additional, partial accounting of defense investments in AI R&D brings the FY2022 budget request up to $1,789.5 million. In addition to what’s counted in the NITRD PCAs for AI R&D, the National Science Foundation also funds $73.6 million in AI education, which is not otherwise included in the charts on this page.
For more information on Federal efforts in AI R&D, please refer to the innovation strategic pillar webpage on AI.gov.
Chart 1: Federal budget for nondefense AI R&D, FYs 2018-2022
The Federal budget for nondefense AI R&D accounts for investments in both core AI and crosscutting AI that is captured under other NITRD PCAs. The first accounting of core AI R&D investments occurred in FY18. The AI crosscutting budget data was reported for the first time in FY19 in the “NITRD Supplement to the President’s FY2020 Budget”, and is not available for FY18.
Chart 2: Nondefense AI R&D FY22 Budget Request by Agency
Table 1. Agency Budgets by NITRD PCA for AI R&D,¹ FYs 2019–2022
|Nondefense TOTALS||FY20 Actual||797.5||110.6||38.7||75.4||16.3||5.4||71.8||77.1||53.2||121.4||27.2||31.4||1,426.0|
|Nondefense TOTALS||FY20 Supplemental||5.0||4.0||0.2||0.5||0.0||0.0||0.0||0.0||0.1||0.7||0.0||0.0||10.6|
|Nondefense TOTALS||FY21 Enacted||954.9||102.3||30.8||75.9||13.6||3.6||57.7||78.8||55.8||109.5||19.7||31.7||1,534.4|
|Nondefense TOTALS||FY21 Supplemental||0.6||4.2||0.0||0.0||0.0||0.0||0.0||0.0||0.0||0.0||0.0||0.0||4.8|
|Nondefense TOTALS||FY22 Request||1,108.5||107.8||27.0||79.2||12.1||2.9||59.6||80.4||39.6||107.6||12.9||32.5||1,670.2|
AI Budget Table 1 Notes
1. DOD and DARPA AI R&D investments are not publicly available.
2. FY20 = FY2020 actual budgets; FY21 = FY2021 enacted budgets; FY22 = FY2022 proposed budgets; FY21-S = FY2020 Budget Supplemental (these have not been used to calculate differences between the FY2020 and 2022 budget numbers.
3. For full names of agency subcomponents (departments and offices), see the Abbreviations list https://www.nitrd.gov/pubs/FY2022-NITRD-NAIIO-Supplement.pdf#page=75
4. (1) Examples of AI R&D investments reported under the AI PCA are R&D on general methods for machine vision; (primarily) machine learning; cybersecurity challenges unique to AI, e.g., ability to exploit flaws in an AI system’s goals; algorithms for computational linguistics; and neuromorphic computing architectures or chips optimized for neural nets. (2) Examples of AI investments reported in other PCAs are R&D on robots (reported for IRAS); the data analysis and management ecosystem (reported for LSDMA); broad issues of human-machine interaction (reported for CHuman) and cybersecurity research (reported for CSP); and general neuromorphic computing (reported for EHCS).
5. The Electronics for Networking and Information Technology (ENIT) PCA is new for FY2022.
6. Totals might not sum exactly due to rounding.
Table 2. Agency Budgets for National AI Research Institutes, FYs 2020–2022
FY2020 Budget Actuals, FY2021 Budgets Enacted, and FY2022 Budget Proposed ($ in millions)
|Institute Name||Institute Theme||Fiscal Year||NSF||DHS S&T||DOT||USDA NIFA||Totals||Year Launched|