The Red Queen and the AI Race

Read the market analysis and fund positioning

" Now, here, you see, it takes all the running you can do, to keep in the same place." – Lewis Carroll’s Red Queen may live in a fairy tale, but her words aptly describe the essence of today’s AI industry. An unforgiving race in which companies are investing on an unprecedented scale – not primarily to win, but simply to avoid falling behind. In this race, leading tech corporations are pouring in billions just to maintain the pace.

The capital hamster wheel of AI giants

The numbers are both impressive and sobering: global corporate investments in AI reached a record high of USD 252.3 billion in 2024 – an increase of 25.5% over the previous year. But what looks like a boom at first glance turns out, on closer inspection, to be an attempt to keep up the pace – not to win, but to avoid falling behind.

The burn rate of the leading AI companies has reached unprecedented levels. Elon Musk’s AI startup xAI is reportedly losing around USD 1 billion per month – with expected annual expenses of USD 13 billion and revenue of only USD 0.5 billion in 2025. OpenAI is said to have spent about USD 9 billion in 2024 to generate around USD 4 billion in revenue. Sam Altman even admitted in January 2025 that OpenAI is currently losing money with its paid ChatGPT Pro plan due to unexpectedly high usage. And Anthropic posted losses of USD 5.6 billion on revenue of less than USD 1 billion.

There is no end in sight to this trend. According to media reports, American AI startups raised USD 104.3 billion in the first half of this year alone – nearly the total amount for 2024. Meta, for instance, invested USD 14.3 billion in Scale AI in June, poaching CEO Alexandr Wang and several other top employees. A wonderland, or just a façade?

The MIT Sloan Management Review puts it succinctly: “If everyone has access to the same AI technology… it may move the market as a whole but will not uniquely advantage anyone.” The Red Queen hypothesis from evolutionary biology offers the perfect analogy: in a system that is constantly evolving, it’s not enough to be good – you have to stay better than everyone else.

The cost of “running”: training and infrastructure

One fundamental problem: the investments follow a pattern that appears economically unsustainable. Training costs for large language models alone have exploded – from a few hundred dollars (Transformer, 2017) to several million (GPT-4) to an estimated USD 170 million for Llama 3.1 in 2024. Anthropic CEO Dario Amodei has hinted that model training runs costing USD 1 billion already exist.

The cost increases follow a clear pattern: the computing power required to train major AI models doubles roughly every five months. Companies face a stark choice: keep up or fall behind. There seems to be no middle ground.

But the benefits are not growing at the same pace. The technical progress achieved per dollar invested is diminishing – the classic “diminishing returns.” Increasing computing power now yields only marginal performance gains. And the paradigm “more data = better models” is starting to crack. Google DeepMind has stated that “results from scaling pretraining are plateauing,” and both Dario Amodei of Anthropic and Satya Nadella of Microsoft have acknowledged scaling limitations.

The search for profitability

Investors are currently valuing AI companies primarily based on user numbers and growth, less so on profitability.

ChatGPT recently had 800 million users, yet the company remains unprofitable. A similar dilemma faces 71% of tech CFOs: they are struggling to generate returns from AI, and only 29% of these companies have a clear AI revenue model. Barclays put it this way: the industry invests USD 60 billion annually but expects only USD 20 billion in revenue by 2026.

Conclusion: running to survive

The AI industry is in a classic Red Queen dynamic. Billions are being invested just to stay in the race. But for most AI companies, the road to profitability remains rocky. It is also questionable whether today’s multibillion-dollar investments will ever yield sustainable returns. Margins are often razor-thin, while companies must continue pouring in billions just to remain in the game.

As is often the case, those already leading are the infrastructure suppliers – chipmakers, cloud providers, and data centers – as well as end consumers who benefit from efficiency gains.

For us as investors, this means: beware of alluring growth stories. Because, in the words of the Red Queen: “… If you want to get somewhere else, you must run at least twice as fast as that!”