Sam Altman Unveils Bold ‘AI Factory’ Plan to Generate a Gigawatt of Power Weekly

Sam Altman, the chief executive of OpenAI, has never been one to think small — and his latest vision proves it. On Tuesday, Altman unveiled a sweeping plan titled “Abundant Intelligence”, which outlines a factory capable of producing a gigawatt of AI infrastructure every single week.

To put that in perspective, one gigawatt is enough electricity to power nearly 8,76,000 households for an entire year. Achieving such scale on a weekly basis is nothing short of audacious. Altman himself acknowledged the difficulty, remarking that “the execution of this will be extremely difficult” and could take years of work. Still, he insisted the outcome could be “the coolest and most important infrastructure project ever.”

This announcement comes on the heels of a massive financial development. Nvidia recently disclosed a $100 billion investment in Altman’s ventures, securing access to about 10 gigawatts of AI-focused data centers. That single deal is larger than the annual GDP of several small nations, yet it represents only one piece of Altman’s larger strategy.

Stargate: The Mega-Initiative

A central pillar of Altman’s vision is Stargate, a joint $500 billion project involving Oracle and SoftBank. The initiative is focused on creating AI “super-hubs” across the United States, with the first already under construction in Abilene, Texas. OpenAI has confirmed that five additional sites have been selected, which together will deliver close to 7 gigawatts of capacity.

While Altman has emphasized that much of the infrastructure will be U.S.-based, he has hinted at forthcoming announcements detailing international partnerships and broader rollout strategies. At this stage, the plan feels equal parts moonshot and mega-construction project, blending vast sums of investment with unprecedented levels of ambition.

The Energy Question

Yet the grand plan faces a critical hurdle: energy. Running even the 10 gigawatts tied to the Nvidia deal will require staggering amounts of electricity. Industry experts are openly skeptical that existing grids can support this scale.

Brad Gastwirth, head of research at Circular Technology, cautioned that energy supply may prove the “silent bottleneck” to Silicon Valley’s AI race. “This is going to become a bigger and bigger issue as each year progresses,” he said in a recent interview.

The challenge becomes stark when considered mathematically. A gigawatt each week is the equivalent of building the energy infrastructure of a major city in mere days — and then repeating the feat, week after week. This effort extends beyond chips, cooling systems, and server racks; it hinges on securing vast supplies of clean, reliable energy without overwhelming power grids or worsening climate risks.

Looking Ahead

Despite the hurdles, Altman remains undeterred. He frames the challenge not as a roadblock but as an opportunity to rethink global infrastructure in the AI age. In his view, the foundation for this future is already in motion as generative systems become part of everyday life.

If realized, Altman’s plan would not only advance the development of ever-larger AI models but also reshape how technology and energy intersect in the 21st century. Whether the world is ready for “a gigawatt every week” remains an open question — but Altman seems determined to find out.