Quantum Computers Have a Dirty Secret

December 29, 2025·7 min read

Every few months, a headline announces that quantum computing has achieved another "breakthrough." And every few months, your laptop continues to be better at basically everything you actually need a computer to do. That is not a coincidence. Quantum computers have fundamental limitations that the hype cycle consistently glosses over. Here are five of them.

Check out the full video on my YouTube channel Divide and Quantum.

Limitation 1: Qubits Are Incredibly Fragile

Classical bits are robust. A bit stored in a silicon transistor will hold its value for years. You can drop your laptop, spill coffee on it, and the bits on your SSD remain intact. Qubits are the opposite.

A qubit's quantum state, the superposition of 0 and 1 that gives quantum computing its theoretical power, is destroyed by almost any interaction with the environment. Heat, electromagnetic radiation, vibrations, even a stray photon can cause decoherence, collapsing the qubit into a classical state and ruining the computation.

This is why most quantum computers operate at temperatures near absolute zero, around 15 millikelvins (-273.135 degrees Celsius). That is colder than outer space. The dilution refrigerators required to maintain these temperatures are the size of a room, cost millions of dollars, and consume enormous amounts of energy.

IBM's 1,121-qubit Condor processor looks impressive on paper. But the qubits only maintain coherence for about 100 microseconds. Your computation must start, execute, and finish within that window, or the result is garbage. Imagine if your laptop's RAM erased itself every 100 microseconds. That is what quantum engineers are working with.

Limitation 2: Error Rates Are Still Terrible

Classical computers make errors too, but at rates so low they are essentially negligible. A modern CPU has a bit error rate around 10^-18 per operation. Quantum computers? Current state-of-the-art qubits have error rates around 10^-3 per gate operation, which is a trillion times worse.

This means roughly 1 in every 1,000 quantum operations produces a wrong result. For a computation that requires millions of operations, the accumulated errors make the final answer meaningless without correction.

The solution is quantum error correction (QEC), but it comes at a staggering cost. The most promising error correction schemes require between 1,000 and 10,000 physical qubits to create a single logical qubit that is reliable enough for real computation. So that 1,121-qubit processor? After error correction, it gives you maybe 1 to 2 logical qubits. Not exactly enough to break encryption or simulate complex molecules.

The industry estimates we need around 1 million physical qubits to have enough logical qubits (on the order of a few thousand) for the breakthrough applications that quantum computing promises. We are at roughly 1,000 physical qubits today. That is three orders of magnitude away.

Limitation 3: Quantum Speedup Is Not Universal

This is perhaps the most misunderstood aspect of quantum computing. Quantum computers are not faster at everything. They are faster at a very specific class of problems.

Quantum advantage exists for:

Factoring large numbers (Shor's algorithm: exponential speedup)
Searching unsorted databases (Grover's algorithm: quadratic speedup)
Simulating quantum systems (natural fit: exponential speedup)
Certain optimization problems (potential speedup, still debated)

Quantum computers provide no advantage for:

Web browsing
Word processing
Video rendering
Most machine learning training
Database queries
Virtually all everyday computing tasks

For the vast majority of software, a quantum computer would actually be slower than your laptop, even if it had perfect qubits. This is because quantum algorithms require encoding the problem into quantum states, executing quantum gates, and then measuring the result, a process with enormous overhead compared to classical computation.

Grover's algorithm is often cited as a "quantum search" breakthrough, but it only provides a quadratic speedup. If a classical computer needs N operations, Grover's needs sqrt(N). For a database with 1 million entries, that is 1,000,000 vs 1,000 operations. Sounds great, but a classical computer running at 4 GHz executes those 1 million operations in 0.25 milliseconds. The overhead of running Grover's on actual quantum hardware makes it slower in practice for any database size we care about today.

Limitation 4: You Cannot Just Read the Answer

In classical computing, you compute a result and read it. Simple. Quantum computing has a fundamental constraint called the measurement problem.

A qubit in superposition holds both 0 and 1 simultaneously. But the moment you measure it, the superposition collapses to either 0 or 1 probabilistically. You do not get to see the superposition. You get a single classical bit.

This means quantum algorithms must be carefully designed so that the correct answer has a high probability of appearing when measured. For many algorithms, you need to run the computation multiple times and take a statistical sample. Shor's algorithm, for example, requires classical post-processing to extract the factors from the measured output.

This also means you cannot "debug" a quantum computation by inspecting intermediate states. Measuring a qubit mid-computation destroys the quantum state and ruins everything downstream. Quantum debugging is, in a very real sense, impossible through direct observation.

Limitation 5: The Software Ecosystem Barely Exists

Even if we had perfect quantum hardware tomorrow, we would not have much to run on it. The quantum software ecosystem is in its infancy.

Classical computing has 70+ years of programming languages, compilers, operating systems, libraries, and frameworks. Billions of lines of battle-tested code. Quantum computing has a handful of research-grade SDKs and a few dozen algorithms.

The languages that exist, like Qiskit, Cirq, and Q#, are primarily tools for physicists and researchers. There is no quantum equivalent of Python's standard library. No package manager with 400,000 modules. No Stack Overflow with millions of answers.

More fundamentally, thinking about problems in quantum terms requires a completely different mental model. Classical programming is built on deterministic logic: if-then-else, loops, variables. Quantum programming requires reasoning about probability amplitudes, entanglement, and interference. The gap between a software engineer and a quantum programmer is not a weekend bootcamp. It is closer to a graduate degree.

So When Will Quantum Computers Actually Matter?

Quantum computing is not a fraud. The physics works. The algorithms are mathematically proven. The engineering challenges are real but being actively worked on. The question is timeline.

Near Term (2025-2030)

We are in the NISQ (Noisy Intermediate-Scale Quantum) era. Current devices are useful for research and proof-of-concept demonstrations but not for practical applications that outperform classical computers. Expect continued progress in qubit counts and error rates, but no commercially relevant quantum advantage.

Medium Term (2030-2040)

If error correction advances continue, we may see the first demonstrations of genuine quantum advantage on problems of practical interest, most likely in pharmaceutical molecule simulation and materials science. This is where quantum computing could start delivering real value.

Long Term (2040+)

Large-scale fault-tolerant quantum computers could transform cryptography, drug discovery, logistics optimization, and climate modeling. But "could" is doing a lot of heavy lifting in that sentence.

The Honest Engineering Perspective

Quantum computing is a technology with extraordinary potential and extraordinary challenges. The hype cycle does a disservice to both the public and the researchers doing the hard work. Declaring that quantum computers will "change everything" is as inaccurate as declaring they will never work.

The honest position is this: quantum computers solve a narrow class of problems exponentially faster than classical computers, but the hardware is not yet reliable enough to demonstrate that advantage at scale. Your laptop is not going to be replaced by a quantum computer. Ever. They are fundamentally different tools for fundamentally different problems.

If you are a software engineer, the most useful thing you can do right now is understand which problems quantum computers are actually good at, so you can recognize the genuine breakthroughs when they arrive and ignore the noise in between.