Big tech AI - A potential financial time bomb and what that means for their future AI pricing
Big tech AI - A potential financial time bomb and what that means for their future AI pricing
There were several very concerning pieces of news regarding the financial circumstances of current AI investments by US big tech companies. Here are a few key items:
- In the last three months, Google, Meta, Microsoft, and Amazon invested 112 billion dollars. But these investments are often not included in the company's official balance sheets. Companies often use bonds or loans through special-purpose credit entities or external AI centers to finance these debts. It is estimated that for all big tech AI investments, a private capital of 800 billion US dollars is needed.
- The leading AI company, Anthropic, is planning to grow its revenue from 4.7 billion US dollars this year to 70 billion US dollars in 2028. To achieve this growth, they need to increase their current gross margin from -94% to 77% over the next 3 years.
- Michael Burry, famous as the investor against the US real estate market ("The Big Short"), warns that there are significant distortions in the accounting of the Big Tech AI companies. They are slowly extending the deprecation period for GPUs, the most valuable asset for AI computing, from 3 years in 2020 to 6 years in 2025, to let their wins appear up to 20% higher than in reality. There is a reasonable doubt that GPUs can maintain their current value for that long due to technological progress and potential damage during the 6 years caused by high GPU loading.
What does that mean for AI business clients? The prices of AI services will quickly rise. If clients cannot leave the big tech company's ecosystem, it will directly harm their business. Prices will not only increase for AI services but also across the whole US big tech portfolio to reach their margin goals.
For the EU it's imperative to build up an independent AI infrastructure with the option to choose between different AI models, several server providers, and in the best case, use open source software as much as possible.
Sources:
https://www.nytimes.com/2025/11/08/business/dealbook/debt-has-entered-the-ai-boom.html
https://the-decoder.de/tech-konzerne-nehmen-fuer-ki-ausbau-rekordschulden-auf/
https://the-decoder.de/the-big-short-investor-burry-warnt-die-naechste-blase-platzt-im-ki-chipmarkt/
Comments ()