The AI boom in numbers: from ChatGPT to the AI Act
How generative AI has changed markets, companies and regulation since late 2022 – a detailed analysis based on company reports, EU law and research data.
On 30 November 2022 OpenAI released ChatGPT to the public. What began as a research preview triggered one of the fastest technology waves in economic history. This analysis traces how the boom has played out in capital markets, companies and regulation.
The infrastructure winners
The boom is most visible among chipmakers. Nvidia, whose graphics processors are used to train large AI models, first passed a market value of $1 trillion in May 2023 and $3 trillion in June 2024. Its data-centre revenue multiplied within a few years, according to its quarterly reports.
Large cloud providers also massively raised spending on data centres. Microsoft, Alphabet, Amazon and Meta report sharply rising capital expenditure in their annual reports, largely attributed to expanding AI capacity.
Research and investment
Stanford University's AI Index Report tracks developments annually. It shows, among other things, that private investment in generative AI has surged since 2022 and that most leading models are now developed by companies rather than universities, while training costs rise sharply.
Europe's answer: the AI Act
With the AI Act (Regulation (EU) 2024/1689), the EU adopted the world's first comprehensive AI law. It entered into force on 1 August 2024 and applies in stages. It takes a risk-based approach: certain practices have been banned since February 2025, high-risk uses face strict requirements, and general-purpose AI models have had specific transparency obligations since August 2025.
What it means for companies
Firms using or offering AI face documentation, transparency and governance duties. The OECD notes that the productivity effects of generative AI look promising but are not yet clearly measurable economy-wide. Many companies remain in the pilot phase.
Risks of the boom
Analysts and central banks have repeatedly warned of possible overvaluation of AI companies. Added to that are the high energy needs of data centres, copyright questions around model training and geopolitical tensions over export controls on high-end chips.
Assessment
Generative AI has moved quickly from research topic to central investment field. Whether the huge spending translates into matching productivity gains is the key open question. All figures are documented in the sources linked below.