Looking for a technology that will smash the limits of physics? It isn’t AI
Big Tech execs are behaving less like innovators and more like hucksters — pricing risk into the economy
AI isn’t transcending physical limits — it’s colliding with them at scale
Today’s Big Tech leaders don’t just describe AI as useful—they present it as a force capable of dissolving constraints once thought fundamental. A miracle remedy, basically—to match their messianic complexes.
When Jensen Huang of Nvidia was asked how the world would meet the enormous energy demands of AI, his answer was telling: AI itself would "invent new ways to generate energy," he said.
Which is interesting because, while AI is certainly a force multiplier in doing the work to solve problems in physics, medicine, chemistry, and so on, the AI itself isn’t “inventing” shit.
LLM-based AI systems, including tools like ChatGPT, are powerful because they recognize patterns and compress information at scale. They can accelerate workflows, generate ideas (including bad ideas), and improve decision-making. But they operate within the constraints of the physical world—compute, energy, materials, and time.
Which creates a tidy paradox: not only are these systems not a miracle cure for science’s hardest problems, but they are also the hungriest consumers of the very resources they are supposed to help make more abundant.
Huang’s bloviation, and that of Demis Hassabis of Google DeepMind, who boasted AI could cure all disease in a decade, rest on a widespread Big Tech assumption—that intelligence, multiplied at scale, can wish away the world’s stubborn constraints. But as we all know, assumption makes an ass out of you, and umption.
The flaw in the “if you scale it, the miracles will come” argument is already evident in the supercomputers paired with AI to train large language models, run global recommendation engines, and simulate everything from weather to markets.
These industrial-scale systems cost hundreds of millions of dollars, fill warehouse-sized facilities, and deliver exascale compute—quintillions of calculations per second—yet even they tap out when asked to fully simulate complex chemistry, model real-world turbulence, or solve the hardest optimization and cryptographic problems.
Quantum leap of faith
AI and classical supercomputing won’t solve the world’s existential challenges, but quantum computing might—if we can get it to work at scale.
What separates quantum computing from AI and supercomputing is not scale, but how it works. AI and supercomputing extract more performance from classical machines—more data, more processors, more energy—to approximate solutions ever faster. Quantum computing, by contrast, is an attempt to compute differently altogether, exploiting quantum states to explore solution spaces that classical systems can only approximate or brute-force. It’s like the technology in a Marvel multiverse movie—if Marvel movies were still good.
Topline: Where the classical stack gets stronger by getting bigger, quantum aims to get stronger by changing the rules.
Significantly, the same hyperscale companies driving the AI boom are quietly investing in quantum computing as well. Google, Microsoft and Amazon are all building quantum programs — not with the urgency or spending of AI, but with long-term intent. They are investing heavily on AI because it scales today, and investing in quantum because they know classical scaling, even at supercomputing levels, has limits; a classic hedge.
But unlike AI, quantum computing is not a settled race dominated by a few giants. It remains fragmented, experimental — and open to surprise. Smaller companies like PsiQuantum, Quantinuum and Atom Computing are pursuing radically different approaches, from photonics to trapped ions to neutral atoms. With no clear winning architecture, a breakthrough could come from a focused lab or a hyperscale data center.
That doesn’t mean those companies will dominate the market. More likely, any breakthrough will still need the infrastructure, capital and distribution that larger players provide. But it does mean the future of quantum computing may not follow the same concentrated path as AI.
Quantum efforts by companies like IBM and IonQ are also not about scaling existing systems further. They are about changing what can be computed efficiently in the first place. Instead of relying on ever-larger classical machines, quantum systems exploit different physical principles to solve specific classes of problems that would otherwise be intractable.
Slow your roll
This doesn’t mean quantum computing will magically solve energy production or supercharge the economy. It won’t. Quantum computing will likely become useful in narrow domains within the next decade — but broad, transformative impact remains at least another decade beyond that, if it arrives at all.
For that to happen, solutions must be found to problems, including scalable error correction and stable, high-fidelity systems where the qubits don’t wobble like a kindergartener's milk tooth when you ask them a tough question. AI can help with that. So that’s nice. But it won’t be quick — think 10 years for systems that start to scale usefully, and mid-century for the miracle-working variety of quantum (if ever).
In the meantime, hyperscaler hyperbole will continue unabated, fueled by a seemingly endless torrent of sucking up from media and analysts. The consequence is massive, reckless overinvestment—both in company shares and in the data centers and compute needed to build a “made in America” AI machine that won’t ever do what it says on the label—all while the rest of the world leans into more rational (literally) strategies, using deterministic machine learning to build digital industrial powerhouses.
Sundar Pichai is particularly fond of the “AI is all-powerful” trope, describing it as “…more profound than fire.” He’s right about one thing: you can get burned by both.
For more of this kind of thing, click here
Or just click on any of the related content below. It’s from FNTV, so you know it’s good.
Stephen M. Saunders MBE is a communications analyst and USPTO-registered inventor examining how digital infrastructure — 5G, cloud, and AI — is reshaping industry, power and society, as well as underpinning the emerging, ubiquitous global digital economy. As anchor of FNTV and a longtime industry insider, he focuses less on growth narratives and more on execution, risk and how hyperscale technology is distorting markets, governance and society at scale.
Opinion pieces from industry experts, analysts or our editorial staff do not necessarily represent the opinions of Fierce Network.