What is a quantum computer actually good for?
Absolutely nothing. Not yet.
To this day, these machines haven’t performed a single useful task convincingly. They’re tiny. Error-ridden. Useless for any commercially relevant problem.
That hasn’t stopped anyone.
Donald Trump’s science adviser promised a “quantum computer powerful for scientific discovery by 2027” in June. On the same day, June 22, Trump issued an executive order to rush US development, framing it as a race against China. 🇺🇸🇨🇳
The hype is coming from both sides of the aisle.
In June, Microsoft announced the Majorana 2 chip. They claimed it pushes their timeline for a practical machine to 2029.
Henry Legg, a physicist at St Andrews who hates Microsoft’s take, called it “complete codswallop.”
“This is not a technology yet.”
Legg dropped a paper in Nature on June 24 dismantling Microsoft’s claims from the previous year. He pointed out massive discrepancies between their peer-reviewed work and their press releases. Nature included Microsoft’s rebuttal, of course.
The pattern is messy. Companies hype. Academics smackdown. Governments set impossible goals. Rinse, repeat.
Incremental gains, massive prices
Is there any progress?
Sure. Just don’t expect it to break the internet. The gains are small, expensive, and incredibly boring.
For a decade, Google, IBM, and Amazon have burned billions. Startups followed suit. The promise? New medicines. Better materials. Superior AI.
Or so they say.
National security experts treat it like Cold War II. They frame qubit supremacy as survival.
The theory sounds nice on paper.
Classical computers use bits. 1 or 0.
Quantum computers use qubits.
Qubits represent probabilities. Imagine a coin flipping in the air. Before it lands, it is both heads and tails simultaneously. That ambiguity is perfect for simulating molecules or photosynthesis, which are naturally probabilistic.
Bad news for email users? Quantum computers suck at classical tasks. They won’t speed up your word processor. 🐢
Everyone is guessing what physical hardware will work.
– Neutral atoms : Used by startup QuEra.
– Ions : Quantinuum uses barium ions.
– Superconducting circuits : The choice for Google and IBM.
– Majorana particles : Microsoft’s pick. Which might not even exist.
The encryption myth
You’ve probably heard the doom scenario: Quantum computers break the bank.
In 1994, Peter Shor designed an algorithm to factor prime numbers. It would crack RSA encryption. That’s the stuff securing your email and bank transfers.
Sounds terrifying.
Except… it probably doesn’t matter.
Cryptographers are already building “post-quantum” protocols. Systems designed to survive the apocalypse. The government ordered a migration to these protocols by 2031.
Plus, the cryptographers didn’t need a real quantum computer to design them. So arguing that quantum computers are essential because they force new security standards is… well. It’s circular.
Current chips like Google’s Willow are too primitive to crack RSA anyway.
“It’s a computer with a very specific purpose.” — Dries Sels
Aim isn’t a gadget in your pocket. It’s a specialized server in a cloud data center. You log in. You run a job. You leave.
But nobody really knows what that job should be yet.
The $12 Billion contradiction
IBM announced last June: $10 billion over five years.
The Trump administration added $2 billion, giving IBM half that money directly.
Money flows. Questions pile up.
In 2019, Google claimed “quantum advantage.” Their machine did a random number generation task faster than the world’s best supercomputer.
Spokespeople called it “supremacy.”
Experts called it a parlor trick.
Useful? No.
Invested anyway? Yes.
In October of last year, Google did it again. They simulated 15 and 29 atom molecules to check magnetic behavior. Their press release bragged that they surpassed classical supercomputers.
Jason Freidenfelds, Google’s spokesperson, doubled down in 2025. “First demonstration of verifiable quantum advantage… relevant for NMR.”
Sels disagrees. “They simulated nothing interesting,” he says.
Why try to refute it with classical computers? It’s contrived.
“If someone gave me billions of dollars, I might,” Sels admitted. Otherwise, why waste time debunking a fake milestone?
Qubits fail. Fixing them costs billions.
Here’s the actual problem.
Qubits lie.
They lose information. Errors compound as algorithms get longer. Any real application requires a long chain of math. The chain breaks. The result is garbage. 🗑️
Researchers are fighting back in two ways.
First, better materials. Andrew Houck at Princeton made superconducting qubits that hold info three times longer last November. How? By making the substrate layer cleaner. More careful temperature control.
“It’s all very subtle tweaks,” Houck said.
Second: Error Correction.
This is where the real excitement lives.
Instead of trusting one physical qubit, you encode one logical piece of info across many physical qubits. The system votes on what the truth is.
Who can build logical qubits cheapest?
– 2024 Google : 105 physical bits for 1 logical.
– 2025 IBM : 12 to 1.
– 2025 Amazon : 9 to 1.
– End of 2024 Quantinuum : 2 to 1.
Getting that ratio down means scalability is actually possible.
Enter Microsoft again.
Microsoft claims they found a shortcut using Majorana particles. Theoretical oddballs that naturally resist errors. They say their thin-wire superconductors make these particles dance.
Legg says they are full of it.
“If you repeatedly try and find God in your toast… you’ll eventually find God. One piece doesn’t prove an epiphany.”
Legg argues their evidence could just be quantum dots. Noise. Garbage.
He notes Nature retracted a major Microsoft Majorana paper back in 2021.
Chetan Nayak, Microsoft’s lead, pushes back hard.
“Our data maintains the strength of our roadmap. We look forward to sharing the energy of our achievements.”
The Road Ahead?
Microsoft says they’ll deliver a full quantum machine.
Others play the longer game.
By 2029 or 2030, IBM and Quantinuum aim for data centers packed with hundreds of error-corrected qubits.
If Legg is right? Microsoft is decades, maybe centuries, off base.
If Microsoft is right?
We’ll see.
Though history suggests we’ll still be paying for hype long after the math catches up. 📉
