In a few years, the same quantum computing concepts that gave Albert Einstein the heebie-jeebies could help Amazon deliver your toothpaste faster.

That's because IBM, the company that surprised the world in 1989 by spelling its name with 35 xenon atoms, is launching a business built on the weird science of quantum computing. Thirty-five years of research into the physics of the freakishly tiny is about to start paying its first dividends with real-world customers.

"We will be providing access to quantum systems for selected industry partners starting this year," said Scott Crowder, who's leading the handoff of the quantum computing work from IBM Research to the IBM Systems product team.

A lot is riding on quantum computing. It offers fundamental breakthroughs that could help bring back the good old days of steadier computing progress. Why is today's smartwatch as powerful as last century's refrigerator-size mainframe? Chalk it up to Moore's Law, which describes the reliable pace of chip improvements across the decades. But some computing progress has stalled, which is why a 2017 laptop doesn't get work done much faster than one from 2012.

"Moore's Law is struggling," Crowder said. But quantum computers will complement traditional computers, not replace them. "It'll do the pieces of the problem the classical computer can't."

Quantum computers, which take advantage of the peculiar behavior and properties of atoms, are notoriously hard for even physicists to comprehend. But quantum computing is bubbling up at university and government labs, startups such as Rigetti and D-Wave, and the research arms of Microsoft, Intel and Google.

The science is still in its infancy, but even as it matures, you shouldn't expect a quantum-powered iPhone. IBM's quantum computer must be cooled to a fraction of a degree away from absolute zero, a temperature colder than outer space, so its innermost niobium and aluminum components aren't perturbed by outside influences. The cooling alone takes days. That's why IBM customers will tap into quantum computers over the internet, not tuck them under their desks or plug them into the company data center.

### Cheap fertilizer, better security

What kinds of work are quantum computers good for? Early efforts will figure out how to use quantum computers effectively and reliably -- kicking the tires, in effect.

After that, though, should come gains like these: quantum chemistry work that could predict how molecules like new medicines interact; logistics projects to figure out the most efficient way to ship packages during the holiday shopping season; and new forms of security that rely on quantum physics instead of today's prevailing approach using math problems too hard to solve in a given period of time.

One specific quantum chemistry example: Factories need lots of expensive energy to make fertilizer, but microscopic bacteria do the same thing much more efficiently somehow. "We don't understand how that reaction occurs," said Jerry Chow, manager of IBM's experimental quantum computing team.

A quantum computer helps people understand what's really going on at the molecular level instead of fumbling around with trial-and-error experiments, Chow said.

IBM believes quantum computers also will let investors better understand financial data and risks. And the budding field of artificial intelligence could get a boost in situations where computers must make decisions without much information.

The common thread for quantum computing tasks is rapid analysis of a huge number of possible scenarios. That strength will, for one thing, be able to crack today's encryption by testing a colossal list of possible numbers to find which ones are mathematical keys that'll unlock private data.

Don't hold your breath, though. That ability is still "really far away," Crowder said. Meanwhile, governments and businesses are developing new quantum-proof algorithms.

### Just plain weird

The quantum era will add a thicket of new jargon to computing vocabulary. Brace yourself for cryogenic isolators, Josephson junctions and decoherence. For processing data, "and" and "or" logic gates from classical computing are joined by Hadamard gates and Pauli-X gates from quantum computing.

At their core, quantum computers store data with "qubits" -- quantum bits. Classical computers work by manipulating conventional bits, small units of data recorded as either a 0 or 1. A single qubit, by contrast, can store both 0 and 1 overlaid through a quantum peculiarity called superposition.

Superposition, combined with another quantum weirdness called entanglement, means that multiple qubits can be ganged together with exponential benefits to how much data they can store and process. A single qubit can store two states of information -- 0 and 1 at the same time -- while two qubits can store four states, three can store eight states, four can store sixteen and so on.

All that overlapping data stored in the same qubits lets quantum computers explore many possible solutions to a problem much faster than conventional computers -- finding which two integers multiply together into a huge number in encryption, say, or the fastest way to deliver a lot of packages.

But even then there are many practical difficulties. For example, the answer to a computing problem can be tucked away in one particular combination of 1s and 0s among many stored through superposition in entangled qubits. But the act of reading data from qubits "collapses" all the qubit states into a single collection of 1s and 0s -- and not necessarily the right one that holds the answer. So yes, it's complicated.

### Learning the ropes

It'll be a long time before programmers learn the ropes for quantum computing. That's why IBM, Microsoft and Google offer simulated quantum computers. More ambitiously, IBM in 2016 opened access to a website called Quantum Experience that lets outsiders noodle with a real 5-qubit quantum computer.

A laptop can't simulate more than about 30 qubits, Crowder said. In the next few years, the IBM Q quantum computers will move to 50 qubits -- each a patch of niobium atoms hooked to its comrades with specialized aluminum wiring.

The real business breakthrough comes when companies can build quantum computing into their operations. That'll require about a thousand qubits, Crowder said. Another big threshold could arrive with about a million qubits, enough to overcome problems with errors undermining calculations. Those errors are more of a problem with quantum computing than classical computing.

"Fault tolerance is hard in a quantum system. You're going to need a lot of qubits," Crowder said. "That's probably at least a decade-plus away."

So maybe your toothpaste will arrive faster. But not anytime soon.

**CNET Magazine:** Check out a sampling of the stories you'll find in CNET's newsstand edition.

**Crowd Control:** A crowdsourced science fiction novel written by CNET readers.