A Decade of AI Investment: The Crash Everyone Forgot

Overview

Global private AI investment is almost always described as exponential growth, and the actual series is considerably more interesting than that. It rose, crashed, and then surged.

Funding grew steadily from $15.3 billion in 2015 to a first peak of $145.4 billion in 2021. It then fell for two consecutive years — down 28% to $104.6 billion in 2022, and down again to $92.8 billion in 2023, leaving it 36% below the peak.

Recovery began in 2024 at $130.9 billion, still short of the 2021 record. In 2025 investment more than doubled again to $290.1 billion, roughly twice the previous high.

The two-year decline is the part routinely erased from charts of this data, and it happened to span the period in which generative AI reached the public. Investment was falling while the technology was becoming a household subject.

The Shape Nobody Draws

Charts of AI investment are usually drawn as a smooth upward curve, because that is what the story of AI is assumed to look like. The measured series does not have that shape.

It has a peak in 2021, a two-year trough, and a steep recovery — the profile of a technology going through a normal capital cycle rather than uninterrupted ascent.

That distinction matters for anyone reasoning about what happens next. A smooth exponential implies a process immune to ordinary market conditions. A boom-bust-boom pattern implies a sector that responds to interest rates, valuations and sentiment like every other, and one that has already demonstrated it can fall by more than a third.

2021: The First Peak at $145.4 Billion

Investment nearly doubled between 2020 and 2021, jumping from $77.3 billion to $145.4 billion in a single year.

The surge was not primarily about AI capability. It coincided with a broad technology funding boom driven by near-zero interest rates, abundant liquidity and elevated valuations across the entire venture market. AI captured a large share of that flow along with everything else.

The models available in 2021 were meaningfully less capable than what followed. Capital was arriving ahead of the technology rather than in response to it, which is precisely why it proved able to leave again.

The Two-Year Decline

From that peak, private AI investment fell in both of the next two years. It dropped 28% to $104.6 billion in 2022, then a further 11% to $92.8 billion in 2023.

Cumulatively that is a 36% decline from peak to trough — roughly $52 billion of annual investment withdrawn from the sector over two years.

This was not an AI-specific failure. Rising interest rates repriced risk assets across the board, technology valuations corrected sharply, and venture funding contracted globally. AI declined because it is part of the venture market, not because anyone concluded the technology had stopped working.

ChatGPT Launched Into a Falling Market

The timing here is the most counter-intuitive fact in the entire dataset. Generative AI reached mainstream public awareness at the end of 2022, in the middle of this decline — and investment fell again in 2023.

The year in which AI became the dominant subject in technology, business and politics was a year in which private investment in it dropped 11%.

The explanation is that capital allocation responds to macroeconomic conditions with more force, and more speed, than to product demonstrations. Interest rates in 2023 made every risk asset less attractive regardless of how impressive the underlying technology had become. Public excitement and investment flows are related, and they are not the same signal — a useful corrective to any argument that reads funding levels as a verdict on capability.

2024's Partial Recovery

Investment turned in 2024, rising roughly 41% to $130.9 billion as monetary conditions eased and generative AI moved from demonstration into commercial deployment.

It is worth noting what that recovery did not achieve. At $130.9 billion, 2024 remained below the $145.4 billion recorded in 2021 — three years after the peak, and two years after the technology's public breakthrough, annual investment had still not returned to its previous high.

Stanford's AI Index recorded corporate AI investment reaching $252.3 billion in 2024 on a broader definition that includes more than private venture flows. Different measures of the same phenomenon produce different totals, which is why comparisons should always specify which series is being cited.

2025 Doubled Again to $290.1 Billion

The following year broke the pattern entirely. Private AI investment reached $290.1 billion in 2025, more than double the 2024 figure and roughly twice the 2021 peak.

A single year adding around $159 billion of annual investment — more than the entire 2021 record — is without precedent in this dataset. The increase alone exceeds total AI investment in any year before 2021.

What changed is the nature of what is being funded. Frontier model development requires data centres, specialised chips and energy contracts, and those are capital expenditures on an industrial scale rather than software investments. The sums grew because the thing being built changed from writing code to constructing infrastructure.

Where the Money Concentrates

The geographic distribution is extraordinarily lopsided. In 2024, US private AI investment reached $109.1 billion — nearly 12 times China's $9.3 billion and 24 times the United Kingdom's $4.5 billion.

Those ratios are far wider than the difference in research output, talent or published capability between these countries. What they measure is access to capital markets willing to fund unprofitable companies at scale over long horizons, which the US has and most others do not.

Within the total, generative AI accounted for $33.9 billion of private investment in 2024, up 18.7% on the previous year and more than 8.5 times the 2022 level. It now represents over 20% of all AI-related private investment — a substantial share, and a reminder that most AI investment still goes to things other than chatbots.

Why Inflation Adjustment Matters

The figures in this series are adjusted for inflation, and that choice materially changes the picture over a decade this long.

Between 2015 and 2025 the purchasing power of a dollar fell substantially. Nominal comparisons across that span overstate real growth, because part of the apparent increase reflects the currency rather than the investment.

Adjusted figures make the cycle clearer too. In real terms the 2022-23 decline was deeper than nominal figures suggest, since money was losing value while investment was falling — a period of unusually high inflation. Any AI funding chart should specify whether it is nominal or real, and most do not.

What the Series Is Actually Measuring

The scope of this data is narrower than the headline suggests, and the exclusions are large.

It captures external funding for privately held AI companies raising above $1.5 million — venture capital and private equity deals. It does not include internal corporate research and development, capital expenditure, or public sector funding.

Those omissions are substantial. The AI research conducted inside large technology companies from their own revenues does not appear here, nor does the enormous capital expenditure on data centres by firms building their own infrastructure, nor government programmes. The real total flowing into AI development is considerably larger than $290.1 billion — this series measures the portion that moves through private investment markets.

Conclusion

The decade of AI investment was not a smooth ascent. It ran from $15.3 billion in 2015 to a peak of $145.4 billion in 2021, fell 36% across two consecutive years to $92.8 billion in 2023, recovered to $130.9 billion in 2024, and then more than doubled to $290.1 billion in 2025.

The most instructive stretch is the decline, because it coincided with generative AI's arrival in public consciousness. Investment fell 11% in 2023, the year AI dominated every technology conversation — evidence that capital responds to interest rates and market conditions faster than to demonstrated capability.

The 2025 figure reflects a change in what is being funded rather than simply more enthusiasm. Frontier development now requires data centres, chips and power at industrial scale, which is why a single year added more investment than the entire previous record. It also means the sector's capital requirements now resemble heavy industry — and heavy industry cycles.

Data Source and Attribution

Our World in DataStanford AI Index 2025Stanford HAI

The annual investment series comes from the Stanford AI Index as published by Our World in Data, measuring external funding for privately held AI companies raising above $1.5 million, expressed in inflation-adjusted US dollars. The series covers venture capital and private equity deals and excludes internal corporate research and development, capital expenditure and public-sector funding. Country-level breakdowns and generative AI investment figures come from the Stanford AI Index 2025 report.

FactsFigs reviews, cleans, and cross-checks every source dataset before shaping it into a data story. Each visualization is created and designed in FactsFigs Design Studio — an internal tool developed and owned by FactsFigs — and is the original work of a FactsFigs author, not an AI-generated copy of any existing graphic. Individual assets within a visual may or may not be produced with AI tools, but the design of the visual itself is solely FactsFigs' own.

Figures are estimates at the time of publication, provided for information only — nothing here is financial advice or a guarantee of accuracy.

2026-07-20