You don't need physics or maths to get the big idea. Play with the demos below and see for yourself.
Everything on your phone or laptop (photos, messages, videos) is stored as a long string of tiny switches. Each switch is either 0 or 1. That's called a bit. Think of it as a coin lying flat on a table: it's heads or it's tails, never anything else.
A qubit is the quantum version. Think of a coin that's spinning in the air. Until it lands, it isn't just heads or just tails. It's in a blend of both.
Try it: press Look at it. The moment you check a qubit, the spinning stops and you get a plain 0 or 1. Scientists call this measuring.
The spinning state has a name: superposition. A qubit can lean one way or the other. Set the lean below, then measure the qubit many times and see what happens.
Superposition gives a quantum computer many possibilities at once, but on its own that would just produce random answers. The real trick is interference.
Qubits behave a bit like waves, and waves can combine. When two waves line up, they add together and get bigger. When one wave's peak meets the other's dip, they cancel out. Drag the slider to shift wave B and watch the combined result.
A quantum program is a carefully designed sequence of steps that makes wrong answers cancel out and the right answer add up. Here's a simplified version. There are 8 possible answers and one is correct (green). Run rounds of interference and watch the odds shift.
Each extra bit in a regular computer adds one more switch. But each extra qubit doubles the number of possibilities a quantum computer can juggle at once. Drag the slider and watch the numbers explode.
Qubits can be entangled, meaning they're linked so that their results are connected, no matter how far apart they are. Einstein famously called it "spooky". Try it:
Matches: 0 of 0
Without the link, the coins ignore each other and match only about half the time, by chance. With the link on, they match every time, even though each individual result is still random.
They won't replace your laptop. They're specialist tools for a few kinds of very hard problems. Here are the most talked-about examples, each with an honest status.
Molecules follow quantum rules, so simulating them is a natural job for a quantum machine. Regular computers struggle once the molecules get large.
Example: predicting how a drug molecule will stick to a disease-related protein, before expensive lab testing begins.
Making fertiliser industrially takes a huge amount of energy, yet some bacteria do it gently at room temperature using a molecule too complex to simulate exactly today.
Example: understanding that molecule could lead to greener fertiliser, and the same tools could help design better batteries and materials that capture carbon.
A large enough quantum computer could break some of the encryption that protects banking and messaging. None is that large yet, but organisations are already switching to "quantum-safe" encryption, and the first official standards were published in 2024.
Example: "save now, crack later": scrambled data copied today could be unlocked years from now.
Finding the best arrangement among astronomically many options. Whether quantum beats the best regular methods here is still being tested.
Example: a courier with 100 stops has more possible orderings than there are atoms in the universe. Airline crew rosters and factory timetables are similar puzzles.
Banks run huge "what if" simulations to measure risk and price complicated products. Quantum methods may speed up some of them.
Example: testing how an investment portfolio would hold up across millions of possible market scenarios.
Some researchers hope quantum methods could speed up parts of machine learning. So far this is largely unproven.
Example: searching through huge sets of options when tuning an AI model.
Tap a task to see which kind of machine suits it best.
No. For everyday tasks like email, streaming and games, regular computers are better and far cheaper. Quantum machines are delicate, need extreme cold or careful isolation, and are used for specific jobs.
Not fully. Today's machines are still small and error-prone, and building reliable large ones is one of the biggest challenges in science and engineering. Progress is real, but "useful at scale" is still being worked toward.