Greg Brockman, President of OpenAI, appears to believe the world is ready to think beyond chatbots. He recently articulated a thesis that, if taken seriously, could reshape the global economy: the primary resource of the future will not be barrels of oil, but teraflops. According to his vision, we are moving toward an era where access to computing power will be as, if not more, important for business competitiveness than raw materials were in the last century. Over the past six months, Brockman claims, AI has already radically accelerated software development. Forget programmers struggling with tools; now, machines adapt to humans, taking on the lion's share of routine tasks. The speed of idea implementation and the scope of solvable problems directly depend on how many GPUs you can afford.
Consequently, GPU farms are transitioning from the domain of enthusiasts to become strategic assets capable of giving new businesses an edge over giants that have not yet realized their old methods are obsolete. Brockman is essentially stating that AI is not merely a tool but an accelerator that drastically reduces "friction" in any work requiring computing power. Previously, launching complex projects demanded entire departments, years of development, and multi-million dollar budgets. Now, thanks to AI, a three-person startup, provided they have a powerful GPU fleet, can test a hypothesis in days, not months. This sounds like a dream for entrepreneurs, as it theoretically democratizes innovation, allowing resource-constrained teams to tackle tasks previously accessible only to corporations. However, as often happens, behind grand pronouncements of "lowering barriers" may lie the desire of major players like OpenAI to solidify their dominance.
Who benefits from the world believing in this new "oil" that they themselves produce? If access to computing power becomes the key to success, who will be able to afford it? The cost of training advanced AI models runs into hundreds of millions of dollars, and that's just for development. Using existing compute power also comes at a significant price. If access to this "new oil" remains a privilege of the few—large corporations, countries with developed IT infrastructure—then instead of the promised "renaissance," we risk seeing an even greater imbalance in the distribution of wealth and opportunities. A successful startup today requires not only a brilliant idea but also access to computing clusters.
Those unable to secure this access risk being left behind in the technological revolution. Claims that "small teams can tackle tasks that previously required armies of specialists" sound appealing until those teams face the necessity of competing with entities that have access to thousands of A100s. This is important because for CEOs, it signals that it's time to stop viewing AI as just another automation tool. It is a fundamental competitive factor, comparable to access to markets or capital. Your strategy must include a clear plan for securing access to necessary computational resources—through partnerships, investments in cloud services, or your own infrastructure. Ignoring this trend is akin to ignoring electrification at the beginning of the 20th century.
Your competitors, even those who seem weaker, may surpass you if they can test hypotheses and launch new products faster and cheaper due to their access to AI computing power.