Generative AI may be less valuable than big tech expects

Generative AI’s commercial future looks less certain than the investment boom suggests. Current estimates suggest big AI firms face a US$800 billion dollar revenue shortfall, while productivity gains remain limited and concentrated among programmers and copywriters. Cheap or free services such as ChatGPT and Gemini are costly to operate, and OpenAI has acknowledged that even paid pro accounts can lose money because of computing costs.

The path to revenue may be difficult. OpenAI is considering ads in ChatGPT, but advertising may not cover the infrastructure spending needed to sustain genAI. Copyright disputes add another liability, with firms facing lawsuits over scraped content or paying for licences. One model can recall “from memory” 42 per cent of the first Harry Potter novel, and a proposed author settlement involving US$3,000 dollars per book and US$1.5 billion dollars was rejected by courts.

Open models could further undermine valuations. Meta’s Llama lets users run a local version for free, while DeepSeek briefly shook AI stocks by releasing an open model comparable to commercial systems. If open or cheap alternatives become good enough, investors may become more skeptical, commercial services may struggle to justify their prices, and genAI could remain accessible without becoming the dominant profit engine big tech has promised.