We Need a Bigger Quantum Computer to Tell Us What the Smaller One Is For

(Or: The Quantum Craze That Keeps Asking for One More Qubit… Forever)

Remember when artificial intelligence was still the slightly awkward cousin at the tech party? Before ChatGPT turned every PowerPoint into a manifesto about “the new industrial revolution,” there was another miracle technology that was definitely about to change everything. Quantum computing. The one that would crack cryptography, design room-temperature superconductors over lunch, optimize global supply chains while you waited for coffee, and generally make classical computers look like abacuses with impostor syndrome.

For a glorious stretch of years, quantum was the designated “before AI” hype vehicle. Investors, consultants, and press releases treated NISQ (Noisy Intermediate-Scale Quantum) devices like they were just one more press release away from usefulness. Then reality—and a lot of classical algorithms that kept getting better—showed up with a clipboard.

As philosopher Amit Hagar dryly observed in a much older paper (and the universe has a sense of humor), it increasingly looks like we need a quantum computer to tell us whether a large-scale, fault-tolerant, computationally superior quantum computer is even possible. Which brings us, inevitably, to the greatest computer in science fiction.

Deep Thought Approves This Message

In The Hitchhiker’s Guide to the Galaxy, a race of hyper-intelligent pan-dimensional beings builds Deep Thought—the second-greatest computer in the universe—to answer the Ultimate Question of Life, the Universe, and Everything. After 7.5 million years it produces the answer: 42. The beings are disappointed. Deep Thought explains that they never really understood the question, so it will design an even bigger computer (the planet Earth) to figure out what the question actually was.

The quantum version writes itself:
“We need a bigger quantum computer to tell us what the smaller quantum computer can be used for.”

(Imagine the scene: a vast, glowing, slightly judgmental supercomputer the size of a small city, surrounded by hooded technicians, delivering that line in the calm, slightly bored voice of Helen Mirren or Valentine Dyall. Yes, that scene.)

The Laughable “Quantum Advantage”

Enter Amit Hagar’s July 2026 preprint, The NISQ Trap: Eight Years of Demonstrations the Hardware Was Built to Lose (arXiv:2607.07530). The abstract is a masterpiece of polite academic shade:

With a single clear exception, every NISQ-era flagship demonstration of “quantum advantage” has, within eighteen months of its announcement, been classically reproduced, shown to rest on classically tractable structure, or closed by a simulability theorem.

In other words: the quantum process does things faster than binary… useless things. The hardware can only run the circuits that classical algorithms happen to be able to compress efficiently, because the same features that keep the noise manageable (low effective depth, algebraic structure, locality) are exactly the features classical methods exploit. The regions of circuit-space NISQ devices can actually reach with decent fidelity are the same regions that turn out to be classically simulable. It’s not a conspiracy. It’s just physics being a comedian.

Eight years and more than thirty (or forty, counting Chinese efforts) advantage-class announcements later, the pattern is hard to ignore. The exit from the loop, as Hagar notes, is where the threshold theorems always said it was: fault tolerance. Until then, we’re mostly running very expensive random-number generators that classical computers keep catching up to.

A Greatest Hits of Predictions That Aged Like Milk

The optimism was not subtle.

• Google, 2017: Researchers from Google’s Quantum AI Lab published in Nature a piece titled “Commercialize Quantum Technologies in Five Years.” The vibe was clear—useful devices, including for AI-related tasks, were around the corner. Quantum supremacy itself was expected imminently (and partially delivered in 2019 on a carefully chosen sampling task that later got classical competition). Consumer-market or near-term practical devices within a five-year horizon were part of the broader conversation.
• IBM Institute for Business Value / Think 2018: IBM’s “5 in 5” predictions declared that within five years quantum computing would move from researchers’ playground to mainstream. It would be used extensively by new categories of professionals and developers, show up in university classrooms, and even reach high-school level to some degree. Quantum would solve problems once considered unsolvable.
• BCG (Boston Consulting Group): In their 2019 analysis they projected value for end users of $2–5 billion by 2024. Later updates kept the long-term (2040) numbers lofty ($450–850 billion in economic value) while quietly revising the near-term NISQ-era numbers downward—sometimes dramatically. The 2021–2024 forecasts kept getting more cautious about when the money would actually show up.
• IonQ around its 2021 public listing: The company and its materials projected rapid progress toward broad quantum advantage on relatively near-term timelines, with algorithmic-qubit milestones and commercial traction presented as imminent. Roadmaps and investor materials painted a picture of useful machines arriving far sooner than the more cautious “decade-plus” estimates circulating elsewhere.
• Kipu Quantum and similar algorithm-focused startups: Claims of near-term returns and quantum advantage on existing or near-term hardware appeared in talks and posts (sometimes “months” rather than years). The pattern has been familiar: exciting demos, followed by classical competition or limited practical uptake.

Other greatest hits include repeated “quantum supremacy / advantage” announcements (Google 2019, various IBM, D-Wave, Quantinuum, etc.) that were either classically simulated shortly afterward, restricted to contrived tasks, or quietly walked back in usefulness. D-Wave’s materials-simulation claims have been challenged almost immediately by classical methods. Microsoft’s Majorana announcements have drawn heavy skepticism. The field has a reliable cycle: announce → hype → classical paper or simulability theorem → “well, the next chip…”

Other Fun Failures (Because the Universe Is a Comedian)

• Random-circuit sampling tasks that were supposed to take thousands of years on classical hardware later took seconds or minutes after better algorithms and hardware appeared.
• Error rates that improve, but never quite fast enough to outrun the overhead of error correction for anything commercially interesting.
• Public roadmaps that quietly drop the most aggressive qubit-count targets (IBM’s earlier exponential scaling plans being a notable example) in favor of “focus on quality and error correction.”
• The perpetual “five years away” horizon that has been sliding forward since at least 2017–2018.

None of this means quantum computing is impossible. Fault-tolerant machines with enough logical qubits could still deliver genuine advantages in quantum simulation, certain optimization problems, and cryptography-breaking (the latter being the one application everyone actually understands). Progress on error correction (Google’s Willow below-threshold results, various ion-trap and neutral-atom advances) is real. The physics is not a scam.

What is fizzling is the NISQ-era fantasy that noisy, intermediate-scale devices were about to deliver broad, near-term commercial miracles. They mostly deliver papers, press releases, and the occasional sampling task that classical computers eventually catch. The industry is slowly pivoting language toward “fault-tolerant” and “logical qubits” and longer timelines—exactly the move you make when the previous story stops landing.

So here we are in 2026: still waiting for the computer that can tell us what the current computers are actually good for. Deep Thought would understand. It would also probably charge us for the answer in a currency that doesn’t exist yet, and then design an even bigger machine to explain the invoice.

But our new investments – going ALL IN on AI – surely those are not misguided at all. Because we learned from past mistakes…

Until then, the quantum craze joins a long and honorable tradition of technologies that were five years away… for a decade. Share and enjoy. And maybe keep your classical optimizers handy.