Who will be out of work and how much money is in AI: six questions from the Stanford report

Stanford University has published the AI Index Report, a summary of how AI is changing the economy, the labor market, and everyday life. In 423 pages, the authors analyze data for the year 2025 and conclude that artificial intelligence has rapidly become a mass technology, and money, infrastructure, and influence are concentrating even faster around a narrow group of players.

This year's report centers on the nearly disappearing gap between the U.S. and China in terms of model quality and the rapid growth of the AI economy. The report's authors write that since 2013, investment in AI has grown roughly 40-fold - if M&A, private equity and public offerings are included - and global corporate investment in the sector more than doubles in 2025.

In 2025, more money than ever came into AI: $581.69 billion - twice as much as in 2024, according to the report. 60% of that flow came from private investment - $344.7 billion, more than twice as much as a year earlier. The number of deals worth more than $1 billion almost doubled - from 15 to 28 for the year.

These investments are highly concentrated by country: the main center of capital attraction remains the US, which leads by a huge margin. Almost 83% of all private investment - or $285.9 billion - was attracted by US AI companies last year. China and the UK, also in the top 3, had investments of $12.4 billion and $5.9 billion respectively.

Among the companies, a narrow circle of leaders is scaling fastest: according to the report's authors, OpenAI and Anthropic, whose annual revenues are approximately $25 billion and $19 billion, respectively, have taken the lead. By comparison, xA and Mistral AI have revenues of less than $500 million.

The gap between the U.S. and China in the quality of AI models has almost disappeared, according to the report. Since the beginning of 2025, U.S. and Chinese systems have repeatedly swapped places in the performance rankings: in February 2025, DeepSeek-R1 took the lead, and by March 2026, Anthropic had taken over the lead.

But the US is still ahead of China in terms of the number of releases: in 2025, US companies will release 50 AI models versus 30 Chinese ones. In terms of private investment, the gap also remains wide - $285.9 billion versus $12.4 billion - but the authors stipulate that the statistics may not be accurate. They estimate that since 2000, state development funds have channeled about $184 billion into Chinese AI companies.

Another area where the US leadership is undeniable is data centers. There are more than 5,000 of them in America - ten times more than in any other country. China has 449. In addition, the report mentions the project of OpenAI, SoftBank, Oracle and MGX, which involves investments from $100 billion to $500 billion in the construction of data centers in the U.S. by 2029.

Productivity growth in the U.S. reached 2.7% in 2025, but AI's real contribution to this, according to one estimate, was only 0.01 percentage points. The report's authors write that in tasks that require more complex reasoning, AI doesn't always speed things up. For example, AI assistants slowed down developers of open-source projects - by 19% on average. But at the level of individual tasks, the AI effect looks much more noticeable. For example, support specialists began to close 14-15% more requests per hour, and AI helps GitHub Copilot developers fulfill 26% more requests.

The impact of AI on the labor market is uneven and is strongest in hiring and among younger workers. Employment among software developers aged 22 to 25 has fallen by almost 20% compared to 2024. Employer surveys show that the situation could get worse, with a third of respondents expecting staff reductions over the next year. The most likely reductions are predicted in service operations, supply chain management and software engineering.

Polls show that the public and experts have different views on AI's impact on employment. 64% of Americans expect AI to reduce the number of jobs over the next 20 years, and only 5% expect it to increase. Among experts, the assessments are less gloomy: 39% predict a reduction in employment, 19% - its growth. At the same time, it is the experts who expect a much faster implementation of the technology: according to their estimates, by 2030, generative AI will help in 80% of working hours in the United States, while the public estimates this figure at only 10%.

Generative AI is spreading faster than personal computers or the Internet did in its time: in three years, it has reached a penetration rate of 53%. But adoption rates vary markedly across countries and are closely linked to GDP per capita: in Singapore, the figure has reached 61%, in the UAE 54%, while the US ranks only 24th with 28.3%.

By early 2026, the estimated consumer benefit to Americans from generative AI has reached $172 billion a year, up from $112 billion a year earlier, and the median value per user has tripled. Most tools remain free or nearly free for now. Against this background, investor interest has also grown dramatically: investments in generative AI have increased by more than 200% in 2025.

Most still believe the benefits of the technology outweigh its drawbacks, but anxiety about the implications of AI is growing, the report said. The gap between experts and the general public is particularly pronounced. Positive impact of AI on work is expected by 73% of experts and only 23% of the public. The authors note similar discrepancies in assessments of AI's impact on the economy and healthcare.

Opinion on who should regulate this technology is also unevenly distributed. The highest optimism and the greatest trust in their governments are found in Southeast Asian countries. In North America and Europe, expectations are noticeably more subdued. The lowest level of trust in their own government when it comes to AI regulation was in the US - 31%.

This article was AI-translated and verified by a human editor