The AI Revolution Runs on Silicon. That's a Bigger Deal Than It Sounds.

It’s easy to talk about artificial intelligence as if it lives entirely in the cloud. Models improve, chatbots get smarter, and new applications seem to arrive every day.

But “the cloud” has an inconvenient habit of needing very real things: chips, factories, electricity, water, supply chains, and geopolitical stability.

In the October 13 episode of FUTUREPROOF., I spoke with Rakesh Kumar, a professor of computer engineering at the University of Illinois and author of THE CHIP AGE: HOW SEMICONDUCTORS BECOME THE WORLD’S MOST POWERFUL RESOURCE, about why semiconductors have become such a consequential resource.

Software gets the attention. Hardware sets the limits.

We tend to evaluate AI by what we can see: an answer on a screen, a generated video, a task completed in seconds. Underneath those experiences sits a complicated industrial system that can’t be scaled by simply adding a few lines of code.

Advanced chip manufacturing takes specialized equipment, deep expertise, capital, and time. That means AI competition isn’t just a race between software companies. It’s also a race among the countries and businesses capable of supporting the underlying infrastructure.

Why geopolitics is now a product concern

The conversation explores Taiwan’s central role in semiconductor supply chains and the difficult questions raised by export controls. Attempts to protect strategic technology can reshape incentives across the global market, sometimes in unintended ways.

For business leaders, this is not abstract foreign-policy trivia. The availability and cost of computing power affect what companies can build, how fast they can build it, and whether their economics make sense.

What if the bottleneck is electricity?

One of the most interesting questions in the episode is whether power availability, rather than algorithms, could become a defining constraint on AI growth. It’s a useful reminder that digital transformation ultimately collides with the physical world.

There’s an irony here. We often describe AI as the ultimate productivity technology, yet the industry’s own growth depends on solving some decidedly old-school infrastructure challenges.

My takeaway: if you want to understand where AI is going, don’t just watch the model launches. Watch the chip supply chain, energy capacity, and international policy decisions. Those may tell us more about the pace of change than the next impressive demo.

Listen to FUTUREPROOF., including “Why AI Chips Are the New Oil.”