Are quantum computers useful yet? (An honest answer)
The hype says yes. The truth is more nuanced. Here's what quantum computers can and can't do today, who's using them, and when they'll actually matter.
You’ve read about qubits, interference, and why quantum computers aren’t just faster classical ones. The obvious question: can they do anything useful today?
The honest answer: barely. But the “barely” part is more interesting than it sounds.
What “useful” means
Let’s be precise. A quantum computer is “useful” when it solves a problem that matters to someone, better than a classical computer could, at a cost that makes sense.
By this definition, quantum computers are not yet useful for practical problems. No company is making better drugs, optimising real supply chains, or breaking real encryption with a quantum computer in 2026.
But that’s not the full story.
What they can do today
Run small demonstrations of future applications
Several teams have run quantum simulations of simple molecules (hydrogen, lithium hydride, small drug fragments) on real quantum hardware. The results match what classical computers can already calculate — so they don’t provide new information. But they demonstrate that the method works.
Why this matters: these are proof-of-concept runs. When hardware improves enough to simulate molecules that classical computers can’t handle (roughly 50-100 qubits with very low error rates), the same methods should scale up.
Example: In 2026, the $5M Healthcare Quantum competition has six teams using quantum hardware to tackle cancer diagnostics and drug discovery on today’s NISQ machines. They’re not solving these problems yet — they’re proving the approach is viable.
Beat classical computers at artificial tasks
Google’s 2019 “quantum supremacy” experiment showed their 53-qubit processor could sample from a specific random circuit distribution in 200 seconds — a task they estimated would take a classical supercomputer 10,000 years (later revised to days by IBM, and further reduced by others).
The task itself has no known practical application. But it demonstrated that quantum hardware can outperform classical hardware on something. Similar demonstrations have followed.
Provide quantum-safe security (already deployed)
Quantum Key Distribution (QKD) — using entanglement to distribute encryption keys with provable security — is the most commercially deployed quantum technology. It doesn’t need a quantum computer, just quantum communication hardware.
Companies like QCi, Toshiba, and ID Quantique sell QKD systems. Banks, governments, and telecom providers are deploying them. This is real, in production, today.
Enable quantum sensing
Quantum sensors — using quantum properties to make ultra-precise measurements of magnetic fields, gravity, and time — are already commercially useful. They’re used in medical imaging (magnetoencephalography), mineral exploration, and navigation.
This isn’t quantum computing, but it’s quantum technology providing real-world value right now.
What they can’t do yet
Simulate useful molecules
The molecules that would transform drug discovery (proteins, complex drug interactions) require thousands of logical qubits with very low error rates. Today’s best machines have ~1,100 physical qubits with error rates that limit useful computation to very simple systems.
The gap: ~1,000 physical qubits today → millions needed for transformative chemistry.
Break encryption
Shor’s algorithm could factor the numbers that protect internet encryption. But it requires millions of physical qubits with sustained error correction. We’re at least a decade away, probably more.
Governments and companies are already migrating to “post-quantum” cryptography — encryption algorithms that quantum computers can’t break. By the time quantum computers are powerful enough, most systems will have upgraded.
Optimise real-world problems better than classical
Quantum optimisation algorithms (like QAOA) have been demonstrated on small problems. But for real-world-sized problems, the best classical algorithms still win. The quantum approaches need more qubits and lower error rates to overtake classical methods.
Replace any classical computing task
As we discussed in why they’re not faster — quantum computers are the wrong tool for everyday computing, and more qubits won’t change that.
The timeline (honest estimates)
Now (2024-2026): Demonstrations, proof-of-concept, quantum sensing, QKD. No practical quantum computing advantage for real problems.
Near-term (2027-2030): First practical quantum advantages likely in molecular simulation and specific optimisation problems. Still limited to narrow applications. Error correction becomes viable at small scale.
Medium-term (2030-2035): Quantum cloud services become genuinely useful for chemistry, materials science, and specific financial applications. Still not general-purpose.
Long-term (2035+): Fault-tolerant quantum computers tackling problems that reshape industries. Quantum-as-a-service becomes mainstream for specific applications.
The caveat: These timelines assume continued progress at current rates. Breakthroughs could accelerate them. Unexpected obstacles could delay them. Anyone who gives you exact dates is guessing.
Why it matters now (even though they’re not useful yet)
Three reasons:
1. The trajectory is clear. Error rates are improving ~10× every few years. Qubit counts are scaling. The path from “lab demo” to “useful tool” is visible, even if the timeline is uncertain.
2. The infrastructure is being built. Companies, governments, and universities are investing billions. Cloud quantum services exist (IBM Quantum, Amazon Braket, Azure Quantum). The ecosystem is real.
3. Whoever understands it early wins. When quantum services become useful for specific domains, the competitive advantage goes to people and organisations who already understand what quantum computing can and can’t do. Learning now is like learning about the internet in 1995 — the technology isn’t ready for mass use, but the people who understand it will be positioned when it is.
The honest summary
- Quantum computers are not practically useful for real-world problems in 2026
- They can run small demonstrations, beat classical at artificial tasks, and enable quantum sensing/security
- The gap between now and useful is primarily engineering (more qubits, lower errors), not physics
- First practical advantages will likely come in molecular simulation (late 2020s)
- Quantum-as-a-service (cloud APIs) is the likely delivery model
- Understanding the field now is a genuine competitive advantage
- Anyone who says quantum computers are already transforming industries is selling something
What’s next?
You now have the complete foundations: why quantum is different, how qubits work, interference, entanglement, error correction, what they’re good at, and where we actually stand (this article). Ready for Level 1? Start with five ways to build a quantum computer.