It doesn’t quite have the buzz of artificial intelligence, but quantum computing is having a moment of its own.

Some of the most powerful institutions in the world, including Google, Microsoft, Amazon, IBM and the U.S. government, are spending many millions of dollars in a race to develop and build the first practical quantum computer.

Startups focused on quantum technology attracted about $2 billion last year, according to a McKinsey & Co. report, as investors pile into an industry that could have nearly $100 billion in revenue within a decade.

There isn’t much business today, though. In total, quantum computing companies generated under $750 million in revenue in 2024, according to the same report.

But more and more, we’re hearing about a big breakthrough.

In the past year, Microsoft unveiled its first quantum chip, Google executives said the technology may only be five years away, Amazon showcased its error-correcting quantum processor and IBM outlined its plan to build a meaningful quantum computer by 2029.

Joining them are scores of smaller companies and universities working on the underlying mathematics, software or potential business model. Some of the companies are even publicly traded, and can see their stocks soar or collapse based on a kernel of news.

In January, Nvidia CEO Jensen Huang sent quantum computing stocks reeling when he said 15 years was “on the early side” in considering how long it would be before quantum computing would be useful. He said at the time that 20 years was a time frame that “a whole bunch of us would believe.”

Two months later, he walked back the comments, but also expressed surprise that they moved markets, or that there were even markets to be moved.

“How could a quantum computer company be public?” Huang said in March.

Right now, there isn’t anything useful that quantum computers can do. They’re purely for research.

But the promise is clear. If the technology works, it can crunch certain kinds of numbers and do some tasks that are currently impossible on a traditional computer, or that would require so much time that the universe would end before they were completed.

To imagine a quantum computer, you have to fundamentally change how you consider what it means to compute.

A traditional computer works because there are billions of transistors on every chip. Those transistors can be ones or zeros — on or off. In large numbers, transistors can represent nearly every number, refer to parts of the system’s memory, and do arithmetic. That’s how every computer in the world works today.

In a quantum computer, the system uses qubits instead of transistors. It’s far more complicated than ones and zeros. Whether qubits are on or off is determined by quantum mechanics, and all of the qubits are “entangled,” which means a change in one will affect the probability of the others.

Making qubits work can require significant infrastructure. For example, some quantum computers have to be operated at very cold temperatures, near absolute zero.

So far, a lot of the applications for quantum have to do with simulating chemistry and physics.

“Quantum computers will not be the compute of choice for every application, and that’s OK,” said Krysta Svore, Microsoft’s vice president of advanced quantum development. “Even if we just use quantum computers for material science and chemistry, 96% of the world’s manufactured goods rely on chemistry and material science.”

Security researchers worry about what they call Q-Day, or the day when an effective quantum computer is created. They predict chaos when passwords and encryption start to mysteriously fail.

“Alongside its potential benefits, quantum computing also poses significant risks to the economic and national security of the United States,” the Biden White House said in 2022, in a national security memo. A cryptographically relevant quantum computer “could jeopardize civilian and military communications, undermine supervisory and control systems for critical infrastructure, and defeat security protocols for most Internet-based financial transactions,” the memo said.

There’s no practical application or algorithm that can be run on a quantum computer that can’t today be accomplished on a normal silicon-based, digital computer.

However, several groups say they’ve proven “quantum supremacy,” indicating that they’ve run a problem on a quantum computer that would’ve taken far longer with a traditional computer. The actions were all abstract.

Google was the first to declare quantum supremacy in 2019, describing its quantum computer’s accomplishment as a “benchmark.” The task it performed is called random circuit sampling, which is basically only used to test quantum computers.

Google says that researchers gave a computer random instructions to make a problem as complex under quantum mechanics as possible. Its researchers were then able to show that a quantum computer is faster at deciphering the quantum problem. Last year, Google said that its new, faster quantum computer Sycamore had expanded the performance gap.

In terms of future real-world applications, most of the potential for quantum computers is in the realms of medicine, chemistry and materials research.

Google points to drug discovery, or finding molecules that could be useful medicines. It also says that quantum computers will be able to do the science needed to commercialize fusion energy.

When Microsoft announced its first quantum chip in February, the company highlighted chemistry and materials science problems, like why some materials corrode, or how to compost plastic.

There is also some optimism that quantum computers will be well suited for generating training data for AI applications, especially for situations or problems with a huge number of potential solutions.

A Google researcher maintains a webpage that catalogs many of the most prominent quantum algorithms.

The most famous is Shor’s algorithm, which showed that a quantum computer would be able to find prime factors of a large number far faster than is currently possible on a digital computer.

When the algorithm was discovered in 1994, it ignited some concern from militaries around the world. Many of them use an encryption method called RSA, which needs the process of factoring large numbers to be difficult in order to keep data secret.

“Without effective mitigation, the impact of adversarial use of a quantum computer could be devastating to [national security systems] and our nation,” the Pentagon said in 2021.

Microsoft has acknowledged the national security factor, and has even framed quantum security as a race against China.

“While most believe that the United States still holds the lead position, we cannot afford to rule out the possibility of a strategic surprise or that China may already be at parity with the United States,” Microsoft President Brad Smith wrote in a blog post in April.

The government has led an effort to move encryption to so-called post-quantum methods, which can’t be broken by a quantum computer. Companies such as Apple have already started to integrate post-quantum encryption into its services like iMessage.

But past communications can still contain secrets. Intelligence agencies and other hackers often collect encrypted data in the expectation that one day it can be decrypted.

For now, much of the work in quantum is still fairly academic.

Most of the advanced hardware companies today are working on “error correction,” or a variety of methods meant to reduce the number of errors, and make them less harmful when they happen.

In present-day quantum computers, the qubits fail as often as 1 out of 1,000 times they are used, according to Microsoft researchers. Microsoft said last week that it was able to reduce the error rate by 1,000-fold thanks to a new approach.

Several improvements in error correction have been announced over the past year, which is one reason why researchers and engineers are increasingly confident that they’ll be able to build a quantum computer.

The next issue to address is scaling up the computers.

Google’s new Willow chip has 105 qubits. Microsoft’s Majorana chip has eight. IBM’s Starling plans to have 200 qubits. Amazon’s Ocelot chip has 14 qubits. In the coming years, these numbers have to go way up. Google and Microsoft say a truly useful quantum computer will need 1 million qubits.

WATCH: Quantum computing is reaching an inflection point

Nvidia stock rose for a fifth consecutive day Friday as the chipmaker hovered near fresh highs and investors shook off China concerns.

The rise in shares has helped the artificial intelligence chipmaking giant regain its seat as the most valuable company.

The stock is up 66% since hitting its 52-week low in early April. Its market capitalization last stood at about $3.8 trillion, putting it ahead of Microsoft and Apple.

Wedbush Securities analyst Dan Ives estimated that both Nvidia and Microsoft will hit the $4 trillion market cap club this summer, and reach $5 trillion over the next 18 months.

Nvidia had a rough start to 2025 as fears of tariffs and China export controls on semiconductors dampened sentiment. Earlier this year, the Trump administration told Nvidia that it would need an export license to ship its H20 processors to China. The chip was introduced following rules introduced by the Biden administration.

CEO Jensen Huang raised concerns about getting locked out of the massive $50 billion China market and said the changes effectively cut off sales without a “grace period.” He said Nvidia’s recent quarterly results would have been better if the company could sell chips in the world’s second-largest economy.

Nvidia held its annual shareholder meeting on Wednesday, where Huang called robotics the biggest opportunity for the chipmaker after AI. The company’s business unit, which includes automotive and robotics segments, reached $567 million in sales, or about 1% of revenues last quarter.

“We’re working towards a day where there will be billions of robots, hundreds of millions of autonomous vehicles, and hundreds of thousands of robotic factories that can be powered by Nvidia technology,” he said.

WATCH: Nvidia hits all-time high as semi boom broadens

One of Germany’s data protection watchdogs on Friday said DeepSeek’s app illegally sends user data to China and asked Google and Apple to consider blocking the artificial intelligence service.

Berlin’s data protection commissioner Meike Kamp said in a statement that DeepSeek’s transfer of German user data to China is “unlawful.”

There is not a readily available way to get in touch with DeepSeek. CNBC has reached out to DeepSeek’s privacy team.

Chinese firm DeepSeek made waves this year when it launched an AI model that it claimed was created at a fraction of the cost of competitors, using less advanced Nvidia chips.

The company also has its own global chatbot AI app, which has been downloaded millions of times, garnering scrutiny.

If the German case against DeepSeek progresses, it could lead to a European Union-wide ban for the app, some experts say.

“It is certainly possible that this incident could lead to an EU-wide ban because the rules that apply in Germany are the same elsewhere in the EU and also in the UK,” Matt Holman, specialist AI and data lawyer at Cripps, told CNBC by email. There are a few steps before this would become reality, however.

This is not DeepSeek’s first run-in with regulators in Europe. Italian data protection authorities in February ordered DeepSeek to block its app in the country. Meanwhile, Irish authorities in January asked DeepSeek for information on its data processing.