Sam Altman just announced OpenAI’s plan to build a gigawatt AI factory every week. Here’s why this changes everything about the next decade…

1/ Abundant intelligence as infrastructure, superintelligence as a service.

“We’re not merely building faster computers, we’re now creating the substrate for a post-scarcity world, where every grand challenge will be solved.”

When Altman describes 10 gigawatts to cure cancer or provide global tutoring, he’s outlining the last era when humanity must choose between transforming one grand challenge over another. Soon, increasing compute capacity will allow us to solve ALL of our hardest problems. Intelligence transitions from scarcity to abundance, and problem solving becomes something every human can take on, enabling all of us to be entrepreneurs and Moonshot engineers.

2/ The “impossible” is becoming a capacity problem, not a capability problem. As one observer notes: “I still can’t fathom that in just 15 years we’ll have AI models capable of solving any conceivable problem, scientific, economic, medical, or creative, within minutes. The entire concept of ‘impossible’ will collapse.”

The below chart reveals the trajectory: AI is learning to maintain 80% success rates on increasingly long-horizon tasks. Today’s models successfully complete 15-minute tasks. By 2030, they’ll handle projects spanning 2 days. By 2035, initiatives that require 2 years to execute.

And by 2050, undertakings equivalent in scope to 113 million years of continuous work.

Think multi-year drug discovery programs, decade-long infrastructure projects, century-spanning climate interventions. All completed by AI systems.

We’re getting exponentially more capable at tackling harder and harder problems. The only remaining question is: how fast can the Western world build the computing infrastructure to make this future real?

3/ Infrastructure is the new geopolitical battleground. Altman’s admission that “other countries are building chip fabs and energy production much faster than we are” reveals the actual stakes. This is about building the factory that builds the intelligence that then builds everything else.

The country that controls AI infrastructure controls economic advantage for the next century. China understands this completely. The question is whether America does. We’re in a battle to create free and abundant resource capability to enable a hopeful and compelling future.

4/ Compute scarcity is the final bottleneck before solving everything. While everyone debates AI safety and regulation, Altman is addressing the real constraint: we have the capability to cure cancer, provide personalized education to every child, and extend human healthspan by many decades… but only if we have sufficient compute capacity to run these solutions at scale.

His gigawatt-per-week factory is the first move in determining who gets to participate in the age of abundant intelligence. The winners will be nations and organizations that treat AI infrastructure like the Manhattan Project. The losers will be those still debating whether AI is overhyped.

5/ OpenAI is not alone. xAI, Google, Meta are “ALL-IN” and competing as well. On top of OpenAI’s superclusters, trillions of dollars are also being invested by the likes of xAI, Google, Meta, Oracle, Microsoft, Nvidia, and others. We are methodically tiling the surface of the Earth in compute!

Here’s what this means:

By 2035, every person will have access to personalized swarms of AI agents: handling research, education, health optimization, and creative work. Scientific discovery will become autonomous and continuous, with AI running millions of experiments daily. AGI tutors provide one-on-one adaptive instruction to every child in any language.

This is the abundance inflection point: when exponential technologies solve humanity’s greatest challenges faster than new problems emerge.

Intelligence becomes substrate: the fundamental layer upon which civilization operates.

Act accordingly.

Until next time,
Peter