Electrifying growth for AI
Generate or have not.
Industrialization and economic advancement have pretty much always been energy dependent from the invention of the wheel to figuring out fire to whale blubber, and coal. Energy is the enabling leverage. Now we are not in a different situation.
There is a new industrialization economic race happening around the world. In some parts of the world, to be more precise as the divergence is significant. That race is who can store, manipulate, and move more bits. The energy demand here will be greater than ever.
The computational advancements of artificial intelligence are remarkable. It currently doubles the length of tasks it can do every seven months. To continue this exponential growth, and overall industry growth, will require massive amounts of resources and including power generation infrastructure.
As perhaps polarizing Donald Trump’s presidency may be, one of the best attributes of his presidency is his accessibility to innovators, builders, and disruptors. He gets this constraint.
It may make sense now why he wants power to be as cheap as possible in the US, and his goal to get chip manufacturing into the US. It’s not just economics but also national security and geopolitical strategy.
Hardware innovation will yield better performance per watt, quantization and algorithm optimization will reduce energy per computation, but the advancement of artificial intelligence will still be training cluster-based and its computational intensity.
A trillion-dollar cluster is a matter of time, and the time to build capacity to power these training clusters takes longer. You want these data centers to be built in your own backyard for security reasons. The industry is at significant risk of a power bottleneck.
AI’s Energy demand will grow 300% in the US in the next few years. By the end of 2026, it is estimated that data centers will consume twice the energy of Japan’s total consumption. Global energy demand from clusters will increase by almost 5x by 2030.
The question is which country can compete in this innovation, and to compete, you need to have infrastructure for power generation. The unfortunate part for many European countries is that they fell into the carbon tax - Green Agenda, as we shall call it, and started demolishing reliable generation and underbuilt.
China, however, saw the writing on the wall that energy is key to economic growth and future industries. In the last 20 years, China has not only surpassed America in electricity production but is now doubling it. With coal being its main reliable source, which Trump has also said the US will be restarting.
AI energy demand isn’t only about its development, it’s also about its use. The utilization and interaction with AI models as a search engine, generating photos and other casual interaction, and all the way to research will also be responsible for immense growth in energy requirements.
This is a field that will be a cornerstone of modern technology and economic production competitiveness. AI applications will proliferate into every industry and will be their base function. This transformation wave of optimization will reshape industries.
This means that countries that enable this AI growth and hold the innovative companies will have the greatest strategic advantage and participation in the over $15T that will be added to the global economy from this field over the next 5 years.
Grid and reliable, uninterrupted power supply will be the difference between the economic and military superpowers of tomorrow.