Why AI companies are burning billions to build the future

The AI industry is currently navigating three distinct bottlenecks. First, startups like Kog are trying to make AI run much faster on the standard computer chips—known as Graphics Processing Units, or GPUs—that data centers already own. Second, large technology companies like Google and Amazon are betting their future on natural gas, building massive power plants to keep their data centers running. Third, high-growth companies like Databricks are raising billions of dollars in private funding to pay for the massive computing costs required to develop and run their AI systems.

Think of an AI model like a complex recipe. To get an answer from the AI, the computer must read through the entire recipe and calculate the result, a process called inference. Normally, this happens on a GPU—a specialized chip designed to handle thousands of small math problems simultaneously. Most software is written for general tasks, but startups are now manually tweaking the internal code of these chips to squeeze out extra speed, effectively teaching the hardware to work more efficiently for specific AI tasks. Simultaneously, the physical machines running this software require constant, reliable electricity. Since wind and solar energy can be intermittent, these companies are turning to natural gas to guarantee a steady, 24/7 supply of power, essentially becoming energy companies themselves to avoid the delays of the public power grid.

The massive fundraising rounds and aggressive energy projects we are seeing signal that AI has moved past its experimental phase. These are no longer just software projects; they are industrial-scale operations. For the user, this means the AI assistant you use is becoming faster and more capable, but it also creates a hidden dependency on fossil fuel markets and massive capital investments. We are reaching a point where the cost of running AI is so high that the companies building these tools have to fundamentally reshape their business models, their infrastructure, and their relationship with global energy markets just to keep the service running.