The AI Circular Economy: How Every Layer of AI Is Connected

AI is moving beyond today’s tools, with increasingly powerful systems capable of tackling far more complex problems. But the bigger story is that AI is not a standalone technology, it is a circular, interconnected economy. The five layers - energy, chips, cloud infrastructure, AI models and applications, depend on one another. Energy powers chips; chips power data centres; data centres run models; models power applications; and applications create demand for more compute, chips and energy. Think of it as a loop, not a ladder. If AI makes applications dramatically more powerful, demand for models and computing rises. That requires more data centres, more advanced chips and ultimately more energy. At the same time, better chips and cheaper, more efficient infrastructure can make increasingly powerful AI systems economically viable. A weakness in any one layer-whether energy availability, chip supply or computing capacity-can therefore constrain the entire ecosystem. This is why the AI economy is ultimately about much more than building the smartest model. ASI will sit at the top of a much larger economic machine, with every layer feeding the next and creating demand for the layer below it. Understanding this interconnection helps explain why countries and companies are competing across the entire stack-from electricity and semiconductors to cloud infrastructure, models and applications-not just in AI software itself.