Skip to main content
Back to IntelligenceQuantum

Quantum Computing: Where We Actually Are in 2025

Stripping away the hype from quantum computing requires understanding what error rates actually mean, why fault tolerance is hard, and the honest timeline to practical quantum advantage.

F
Firedeus
4 min read

The gap between announcements and reality

Quantum computing headlines follow a predictable pattern: major lab announces milestone, financial media interprets as breakthrough, skeptics point out error rates, everyone moves on. Repeat every six months.

The difficulty is that quantum computing progress is real but the timelines are genuinely uncertain, and most people covering it lack the background to distinguish a meaningful advancement from a press release.

This is an attempt at a grounded reading of the current state.

What a qubit actually is

A classical bit is 0 or 1. A qubit is, until measured, in a superposition of both states simultaneously. Two qubits can be entangled, meaning their states are correlated regardless of distance.

This allows quantum algorithms to explore a much larger solution space in parallel. For specific problem classes — factoring large numbers (Shor's algorithm), simulating molecular chemistry, certain optimization problems — this is a genuine computational advantage.

The problem: real qubits are fragile. Any interaction with the environment — vibration, heat, electromagnetic interference — causes decoherence, which collapses the quantum state to classical noise. This is why quantum computers operate near absolute zero.

Physical qubits vs. logical qubits

Current machines have "physical qubits." These have error rates around 0.1–1%. For any useful computation, you need error rates below 10^-10 or so.

The solution is quantum error correction: using many physical qubits to encode one reliable "logical qubit." Current estimates suggest you need 1,000–10,000 physical qubits per logical qubit depending on the error correction code and the physical qubit quality.

Google's Willow chip (late 2024) demonstrated error rates that decrease as they scale up the surface code — a meaningful milestone. IBM's Condor reached 1,000+ physical qubits. But 1,000 physical qubits with current error rates corresponds to perhaps one decent logical qubit.

The machines that will demonstrate practical quantum advantage will need millions of physical qubits.

The honest timeline

The quantum computing field has a history of overpromising. In 2019, major institutions were suggesting "quantum advantage in 5 years." In 2025, serious estimates have shifted to the mid-2030s for fault-tolerant computation on practically useful problem sizes.

There are three phases:

NISQ (Noisy Intermediate-Scale Quantum) — where we are now. 100–1,000 physical qubits, no error correction. Useful for quantum chemistry simulations of small molecules. Not practically useful for cryptography, optimization at scale, or most of the applications marketed as "quantum."

Early fault-tolerant — perhaps late 2020s. Logical qubits with error correction. Small-scale useful computation. Valuable for specific pharmaceutical simulations, materials science.

Full fault-tolerant — 2030s+. The machines that run Shor's algorithm against RSA-2048 at useful speed, or run Grover's algorithm for meaningful search acceleration.

The cryptography threat is real but not imminent

Shor's algorithm, running on a fault-tolerant quantum computer, can break RSA and elliptic curve cryptography. This is the threat that has driven NIST's post-quantum cryptography standardization process.

NIST finalized its first post-quantum cryptographic standards in 2024: CRYSTALS-Kyber for key exchange, CRYSTALS-Dilithium for digital signatures. The migration to these standards needs to begin now, not when the quantum threat is imminent, because cryptographic migrations take 10–15 years.

The phrase "harvest now, decrypt later" captures the risk: adversaries collecting encrypted traffic today can decrypt it once quantum computers arrive. For data with a 20-year sensitivity horizon, the quantum threat is already relevant.

What to actually watch

The meaningful signals in quantum computing:

Physical qubit error rates below 0.01% — threshold that makes error correction dramatically more tractable.

First logical qubit demonstrations — Google, IBM, Quantinuum, and PsiQuantum are all racing here.

Quantum utility demonstrations — applications where quantum simulation produces a result that classical hardware cannot match. Chemistry and materials science are the most likely first domains.

Post-quantum migration timelines — less exciting but arguably more practically important than hardware progress.

The technology is advancing. The question is whether it advances fast enough to matter for the specific applications most frequently associated with quantum computing in public discourse.

More Intelligence

Space

The Orbital Economy: Who Controls the Pipes in Space

Launch costs have collapsed 95% since 2010. A new orbital economy is emerging around the infrastructure layer — satellites, spectrum, and the companies building the highways of space.

4 min
Legal Tech

AI Liability: Who's Responsible When the Model Is Wrong

As AI systems make consequential decisions in medicine, law, and finance, the legal frameworks for assigning responsibility are years behind the technology. Here's where the first major collisions are happening.

5 min
Robotics

Humanoid Robots: What the Hardware Actually Allows in 2025

Figure, Agility, Boston Dynamics, and Tesla are all racing toward general-purpose humanoid robots. A grounded look at where the bottlenecks actually are and what capabilities are real versus theatrical.

5 min