In an unprecedented show of financial strength, OpenAI is signalling that it is prepared to spend almost any money to ensure its dominance in the field of artificial intelligence. The company plans to spend an additional $100 billion over the next five years on leasing back-up servers. This move dramatically increases its already gigantic cloud infrastructure commitments.
The decision, which was first reported by The Information, citing management’s discussions with shareholders, comes on top of previously projected spending of $350 billion on server rentals by 2030.
In total, including standby servers, OpenAI’s infrastructure spending is expected to average around $85 billion per year over the next five years. This is an amount that dwarfs the IT spending of most global corporations and some countries.
This astronomical investment in computing power underlines a fundamental truth of the current AI era: advances in advanced modelling are inextricably linked to access to powerful and, crucially, limited infrastructure.
Technology companies are in a fierce battle for every available gigawatt of power, driving up prices and securing huge profits for cloud service providers and chipmakers such as Nvidia.
Internally, OpenAI executives see these back-up servers not as a mere security cost, but as a ‘monetisable’ asset. The company hopes that the extra computing power can be used to fuel unexpected research breakthroughs or to handle surges in the popularity of its products, thereby generating revenue that is not yet included in official forecasts.
However, this strategy comes with a huge financial risk. According to earlier reports, the ChatGPT developer expects to ‘burn’ around $115 billion in cash by 2029. The gigantic expenditure on servers is a bet that OpenAI’s future capabilities and products will not only dominate the market, but also generate revenue on a scale to justify such a colossal investment. For now, it is the most expensive bet in the history of the technology industry.
