Quantum computing is drawing renewed attention owing to a recent advance from Google and talk of the U.S. government taking stakes in companies working on the technology. But the industry has a way to go before the rewards for investors outweigh some glaring risks.

Most U.S.-listed quantum stocks have risen sharply in recent months, including IONQ Inc., which has nearly doubled in the past six months, and D-Wave Quantum, which has more than quadrupled.

That enthusiasm reflects budding interest in an industry that, like artificial intelligence, may become a crux of geopolitical competition that the U.S. seeks to dominate. But it also reflects quantum computing's significant underlying promise.

Quantum computers make calculations differently than conventional computers. Instead of representing data with bits that can be either one or zero, quantum computers harness the quantum-mechanical properties of so-called qubits that can be a combination of the two at the same time.

That tweak allows quantum computers to juggle more possibilities at once, opening the way for solutions to problems that would take conventional computers a near eternity. A powerful quantum computer, for example, could test complex molecular combinations quickly, potentially leading to the rapid discovery of new drugs. Already, scientists have used quantum computers to identify materials that could make solar cells more efficient, simulate Airbus's aircraft performance and optimize power grids.

But performance is uneven. Even today's most advanced quantum computers for the most part don't outperform regular computers in arenas where quantum computing should excel. That is mostly because today's quantum computers don't have electronic brains that are large enough and don't fix calculation errors reliably enough.

And it has proven extremely difficult to build large and error-free quantum computers. Many of them have components that need to be cooled to near absolute zero for quantum effects to become usable. They are often physically large and delicate. International Business Machines has been at it for about a decade and produces some of the most powerful quantum computers, but its most advanced system has just 156 qubits.

Analysts say quantum computers will need much larger numbers of qubits to tackle many problems ordinary computers can't. IBM released a road map this year that lays out a path to 2,000 qubits in 2033. Google, the other Big Tech company considered a leader in the arena, has a quantum chip with 105 qubits and is aiming to hit a milestone of 1,000 qubits, although its timeline is less clear.

Google last month said its chip could make certain computations 13,000 times as fast as an ordinary computer, providing a taste of the advances that quantum computing could bring.

"Scalability is the primary question as you look from now to the next five years or the end of the decade," said Wamsi Mohan, an analyst at Bank of America who covers the quantum computing industry. "If you can make them scalable then the usefulness of this technology really becomes quite significant."

Who will win the scaling-up battle is far from clear. While IBM and Google have poured in money, so have their Big Tech rivals Amazon and Microsoft. One of the smaller listed quantum-computing players could also come in from below and prevail in the market. Or it could be a startup like PsiQuantum, which is building large-scale quantum computers in Australia and in Chicago, where it broke ground in September.

Quantum computing is so nascent that it isn't even clear which basic technical approach to it will scale up best. Some companies, like IBM and Google, use materials cooled to near absolute zero. Others -- including IonQ -- use charged particles trapped and suspended in space. PsiQuantum uses the quantum properties of light.

For potential investors -- including the Trump administration, which has denied The Wall Street Journal's reporting late last month that it is considering taking stakes in companies including IonQ and D-Wave -- there is therefore no lane for exposure to the quantum-computing phenomenon that doesn't entail a big dose of risk. Any of today's approaches could easily fail, just as Betamax lost out to VHS in the videotape format war decades ago. Early-stage government backing of one approach over another could also have the perverse impact of holding back the industry if the state bets on the wrong horse.

How long it will take to shake out is also uncertain. BNP Paribas analyst David O'Connor said in a recent note that quantum computing was now less of a science experiment than an engineering problem involving how to make computers bigger. That could take three or four years to work out, he estimates.

Whether that holds is hard to predict. But it seems likely that quantum computing will grow fast and generate significant returns for investors if those challenges get worked out. Mohan estimates quantum-computing revenue could reach $4.25 billion by 2030. That isn't an incredible amount, but also nothing to laugh at: It is about what Nvidia was pulling in about a decade ago.

The question now is more one of when, not if, quantum computing becomes a technology worth investing in. And that could be a while.