Follow the money: how venture capital wrote the story of generative AI

Before ChatGPT became a household name, before "foundation model" entered the corporate lexicon, a small number of venture capitalists were placing bets on something most people could not spell. The OECD's data on VC flows into generative AI tells a story not just about money, but about how an entire technological paradigm found its shape.

AI as the new frontier

The United States dominated the investor side from the beginning focussing mostly on business processes and support services, digital security and IT infrastructure. AI Investments were made in the field of automation, trying to make existing enterprise workflows smarter and selling data analytics tools to corporations looking for a competitive edge.

As early as 2013, the EU27 appeared as a meaningful investor, alongside Canada and the United Kingdom with healthcare and education appearing only as minor destination sectors.

Building the plumbing

Around 2017, something shifts. IT infrastructure and hosting began to rise from background noise to become the dominant destination sector. By 2020 and 2021 they are unmistakably at the top of the chart. In retrospect, the logic is simple: before you can deploy AI on a large scale, you need the computing power, storage and cloud architecture to support it.

The investor base also diversifies during this period. Singapore and Australia emerge as significant sources of capital together with the UAE. With 2019 media, social platforms, and marketing begin their ascent. Healthcare and biotechnology attract a sustained thread of investment together with Robots, sensors, and IT hardware, yet they both remain in the background.

The Application Duopoly: Media and Infrastructure

In the 2022–2025 frame, the flows reorganise around two sectors: IT infrastructure and hosting; and media, social platforms, and marketing. By 2023-2024, the two destinations account for the overwhelming majority of VC investment in generative AI. The UAE's emergence at the top of the 2024 chart, reflects sovereign and institutional capital from the Gulf entering the field at scale. The US still dominates, but the source map has genuinely globalised.

This duopoly reveals the actual commercial logic of generative AI as it has been deployed thus far. Writing, image generation, advertising copy, social platform algorithms, the technology found its early product-market in producing text and images faster and cheaper than humans could.

The other sectors are not absent; they are lagging. Whether healthcare, education, or industrial robotics eventually absorbs a comparable share of investment remains an open question for the next decade.

Source: OECD.AI (2026), data from Preqin, last updated 2026-01-05, accessed on 2026-03-10, https://oecd.ai/

Draft mode