The race for artificial intelligence has become a literal race for power, and OpenAI’s deal with AMD is fundamentally about securing the joules needed to create intelligence. The agreement’s headline figure is not in dollars, but in watts—a colossal six gigawatts of computational energy.
This focus on energy as a primary strategic asset marks a new phase in the AI industry. The limiting factor for progress is no longer just clever algorithms but the sheer physical capacity to power the processors that run them. This deal is an energy acquisition strategy as much as it is a technology one.
The six-gigawatt infrastructure, powered by hundreds of thousands of AMD chips, will consume an amount of electricity comparable to a major city. This energy will be transformed into the mathematical calculations that are the basis of AI’s learning processes. Every joule will be a unit of potential intelligence.
This energy-centric view explains the immense scale of the investment. Building AI is now an industrial process that requires an industrial supply of power. OpenAI is ensuring its “factory of intelligence” will not be idled due to energy shortages.
The partnership with AMD is also about efficiency—getting the most intelligence per watt. The choice of AMD’s upcoming MI450 chips reflects a belief in their ability to perform calculations with minimal energy waste. In the new era of AI, the race for intelligence is won by those who can most effectively and efficiently command the power grid.