Something’s shifting. For twenty years, the story was simple: software was eating the world. Marc Andreessen made it a headline in Software Is Eating the World, and for a while, it felt like gospel. Apps were everywhere. Platforms exploded. Investors couldn’t get enough. But then you start noticing cracks. Software is powerful, sure. But it’s messy, expensive, and brittle. It piles complexity on top of complexity. AI dazzles in small corners but forgets almost everything from yesterday. It’s clever, but it’s not a foundation. Not yet.
Fred Wilson gives us a way to look past the hype. In The Next Big Thing Will Start Out Looking Like a Toy, he points out something obvious only after the fact: the real breakthroughs start small, weird, and easily overlooked. They begin as toys, novelties dismissed until they quietly reshape the world. Spot them early? Advantage. Miss them? You’re watching someone else build the future while you scroll Twitter.
So where is the next GPT (General Purpose Technology) actually coming from? Not from software alone. In The Empire Runs on GPUs, I argue that progress moves in stacks. Energy at the base. Hardware. Software. Models. Intelligence. Society. Get the base right, and the rest gets amplified. History isn’t subtle. Coal, steel, and telegraphs built the British Empire. Oil, electricity, and silicon chips made the American era tick. Today, it’s datacenter electricity, GPUs and TPUs, AI frameworks, foundation models, and society trying to figure out how to deploy all of it.
Control matters. But domination alone doesn’t guarantee anything. Sometimes orbiting works better. Sometimes resisting is smarter. Sometimes you thrive quietly in a niche no one cares about. It isn’t a philosophical question; it’s power. Moore’s Law accelerates hardware; Page’s Law drags software. These gaps decide winners and losers.
AI itself isn’t the full story yet. In From Hype to Utility: What AI Needs to Become Truly Transformative, I laid out what’s missing: generality, agency, and usefulness. Can it handle multiple tasks? Act on its own? Actually create value? Some LLMs hit parts of this. Agentic systems hit other parts. None hit all three. Recommenders and chatbots do small bits well, but they’re narrow. They talk, but mostly they don’t do.
Now, here’s where it gets interesting: energy and advanced materials. They check the boxes naturally. Universal. Scalable. Constantly improving. Fusion labs at Helion, solid-state battery teams at QuantumScape, and nanotech factories refining lighter composites are all pushing quiet, physical revolutions. These touch everything from hospitals to farms to factories, often quietly. Incremental improvements compound. Denser batteries, lighter composites, smarter grids. Like Moore’s Law in the real, physical world.
Look at the markets, and you start to see where reality is pointing. Investors aren’t just chasing the latest AI buzz anymore. The money is quietly flowing to the things that actually touch the world: energy, semiconductors, materials, and defense (in the context of indian markets). These are the areas where progress is real, measurable, and sticks. Software swings on hype; these sectors build on physics. The next big leaps won’t come from a viral app or another chat interface. They’ll come from kilowatt-hours, denser batteries, lighter composites, and smarter grids. This pushes the limits of what’s physically possible. Even the AI boom runs on this physical foundation.
And it’s not just tech. Geopolitics matters. The US excels at chips, frameworks, and models. China has energy scale and coordination. Access to AI weights, rare materials, and reliable energy are the strategic levers now. Deployment matters as much as invention. History reminds us: oil without railways doesn’t make empires. Early AI hype in the 1980s fizzled because the stacks weren’t ready. You need invention and the ability to use it.
Think of it like a body. Energy is blood. Materials are skeleton. AI and GPUs are the nervous system. Software is skin: flashy, visible, but ultimately thin. Institutions and markets? Muscles. Move the body. Ignore one layer and the system falters. Focus only on the skin, and you get hype without impact.
And the human side? Critical. Access shapes societies, economies, and opportunity. AI alone won’t transform the world. Transformation happens when you see the layers beneath. When you invest in energy, materials, compute. When you spot toys before they grow into infrastructure.
The future? Messy. Uneven. Full of surprises. Maybe a small lab in Sweden perfects a new battery chemistry while no one notices. Maybe a group of engineers in a Chinese city finds a cheaper way to make composites. Those who pay attention, who position themselves thoughtfully, who integrate technologies in a human, careful way, will define the era. Maybe the next superpower won’t have the flashiest apps. Perhaps it will control energy and materials, the complex intangible forces that actually drive the world.
References
From Hype to Utility: What AI Needs to Become Truly Transformative