Philippe Fauchet
Philippe Fauchet

Often, AI dominates conversations in meetings, news cycles, and boardrooms. The money proves it. According to PwC's Global Data Centre Outlook, global investment in AI infrastructure will hit a record US$31.6 trillion through to 2050. Goldman Sachs Research forecasts US data center power demand will more than double to 66 GW in 2027 from 31 GW in 2025. The subject of all this attention stays the same: AI. Yet, the technology that makes it possible rarely gets mentioned.

The conversations about AI differ. It is the innovation it brings to business, or the damage it does to the environment, or the energy it demands. The subject does not. People pay very little attention to the things that make AI possible in the first place, from the chips it runs on to the data centers that house them. Those facilities are power-hungry and water-hungry. Demand keeps climbing, and so does the pressure to build more, bigger, and faster.

The scale is hard to grasp. Data centers are expected to more than double their power draw within two years. According to Goldman Sachs Research, the share of peak summer power they take is projected to rise from 4.1% in 2025 to 8.5% in 2027. In the background, there may be an energy shortage, a water shortage, and a shortage of the expertise needed to build and run these centers at the scale required. The default answer has been simple. Build more. This reflects a tendency to search for more rather than better.

Human nature explains some of it. People see what is in front of them and miss what is under the hood. The invisible layers, like the air itself, often go unnoticed, yet they are no less important. Nanotechnology may be the clearest example. The same technology that allowed AI to blossom in the first place may be what allows it to keep growing in the years ahead.

Philippe Fauchet has had a career inside that under-the-hood world. He is a professor at Vanderbilt University, where he served as dean of the School of Engineering for 11 years, after faculty posts at various reputed universities. He has published numerous scientific papers, moving his research into nanoscience in the early 1990s.

His framing of the field is simple. He says, "Nanotechnology is everywhere in the physical world, like AI is everywhere in the software world. That comparison runs deeper than language. AI could not exist without everything that nanotechnology has done." The chips, the memory, and the hardware inside data centers all depend on advances made at the nanoscale. "Nanotechnology is, by and large, invisible," he notes. Everybody uses it, and nobody talks about it. In his words, it is what happens under the hood.

Invisibility shapes investment too. "AI is sexy," Fauchet says. Companies raising money for software point to fast returns and quick iteration. Nanotechnology is a different game. "Once upon a time Nanotechnology was that sexy symbol of the tech industry. But, as the technology advanced, it became taken for granted," he says. Building a fab line, where Nanotechnology reigns dominant, takes years and many tens of billions of dollars. The returns come slowly. Yet the physical infrastructure that AI depends on may not improve without exactly this kind of patient, long-term investment.

But advancements are happening. In June, IBM unveiled a 0.7-nanometer chip design, reported by CBS News, that it projects will offer up to 50 % more performance or 70 % greater energy efficiency than its current 2-nanometer chips. Less power-hungry chips mean less heat, less water, and less electricity across the data centers around them. Fauchet stays measured about the timeline.

"These advancements are very important, but they are still playing the same game. The scale of growth we see within data centers changes this conversation completely. We can't afford to play the same game anymore; we need a breakthrough. Without this, growth will come to a standstill, and this isn't a conversation for 10 years from now; it's happening now," he says.

The resistance to the status quo is growing too. A Gallup survey published in May 2026 found 7 in 10 Americans oppose constructing data centers for artificial intelligence in their local area, including nearly half who are strongly opposed. Two forces are colliding. The public wants AI, but not the data centers that run it. There is only so much water, so much space, and only so much patience in the communities asked to host the buildout.

The way forward may rest on a realization. Nanotechnology is not just the premise that made AI possible. It can also be the solution to the problems AI has created, in chips that use less power and cooling that needs less water. The question now is whether the physical world behind AI's rise gets the attention and the investment that the software world has enjoyed. The future of AI may depend on what happens under the hood.