The Quantum Computing Bubble Popped, and That Is Exactly Why It Is Now Investable

Late last year, three quantum computing stocks were trading at valuations that would have made dot-com era investors blush. IonQ, Rigetti Computing, and D-Wave were riding a hype wave that began crashing the moment Nvidia CEO Jensen Huang told a CES 2025 press Q&A that useful quantum computers were still 15 to 30 years away. That single comment wiped tens of billions of dollars off the sector in a single trading session. IonQ fell roughly 39 percent. Rigetti dropped 45 percent. D-Wave collapsed by 36 percent.

The speculative bubble that built afterward was almost as ridiculous as the crash that preceded it. But something quietly important happened in the months that followed. The hype cooled, momentum money rotated out, and the actual technology kept moving forward. Three concrete commercial milestones hit in the same week recently, and the sector has now graduated from science fiction speculation into something serious investors can model.

Jensen Huang Quietly Reversed His Quantum Stance

Two months after that CES comment, Huang appeared at Nvidia’s GTC 2025 conference and softened his timeline noticeably. By the time he reached Paris three months later, he was openly saying that quantum was approaching an inflection point and would start solving real problems within a few years. For anyone who tracks Nvidia closely, this kind of public reversal is almost unprecedented. Huang does not change his mind on technology trajectories without a structural reason behind it.

The reason became obvious when Nvidia shipped an open source software stack called Icing two weeks ago. The toolkit handles the messy work of splitting computational problems between traditional GPUs and quantum processors so that each handles the parts they are best suited for. Nvidia placed IonQ on the short list of early adopters and tied the release into their existing CUDA-Q quantum platform, which already underpins research efforts across the major hyperscalers and national labs.

What Nvidia’s NVQLink and Icing Actually Do

NVQLink is a high speed connection that lets a GPU send instructions to a quantum chip and read results back in a few millionths of a second. That round trip speed matters because quantum chips require constant error correction, and only a fast classical co-processor can manage that loop in real time before decoherence sets in. Icing layers on top by accelerating quantum calibration, a process that used to take days and now takes hours.

The strategic message from Nvidia is clear. They are not building their own quantum chip. They are building the connective tissue between everyone else’s quantum hardware and the GPU infrastructure that already runs the global AI economy. Quantum is being positioned as a specialized accelerator inside the existing data center stack, not as a replacement for it. That distinction matters enormously when sizing positions in this sector.

The Math That Makes Qubits So Valuable

A classical bit is binary. It is either zero or one, like a coin showing heads or tails. A qubit behaves more like a weighted coin that holds the probability of both outcomes at once until you measure it. Link multiple qubits together through entanglement and every additional qubit doubles the number of states the system can represent simultaneously.

A 300 qubit entangled system holds more possible states than there are atoms in the observable universe. That is why quantum machines can crack certain problems that would take classical supercomputers longer than the age of the universe. Cryptography, materials simulation, supply chain optimization, and specific classes of machine learning all sit in this special category. The catch is that qubits are extremely fragile. Heat, vibration, or even a stray photon can knock them out of their entangled state, a failure mode called decoherence. Stabilizing qubits long enough to do useful calculations is the central engineering challenge of the entire industry.

IonQ Builds the Most Stable Qubits in the Industry

IonQ uses trapped ion technology, which holds charged atoms in place using electromagnetic fields and manipulates them with lasers. The approach runs slower than competing architectures but produces qubits with significantly lower error rates. Lower errors mean longer coherence times, and longer coherence times mean more useful work per calculation.

The bigger advantage is networkability. IonQ’s photonic interconnect, which the company demonstrated for the first time two weeks ago, links two separate quantum processors so they behave as one larger machine. The industry has been chasing this milestone since the 1990s. Scaling from a hundred qubits on a single chip to the tens of thousands needed for commercial cryptography and chemistry simulation requires either building one impossibly large chip or networking many smaller ones together. IonQ just proved the second path works in practice.

On the same day, DARPA added IonQ to its quantum benchmarking and scaling program, better known as HARK. The program effectively underwrites IonQ’s scaling roadmap with federal money. Financially, the company posted $130 million in revenue for the full year, more than triple the prior year, with $62 million in the most recent quarter alone. The order backlog sits near $370 million, and the balance sheet holds roughly $3.3 billion in cash and short term investments. Be careful with the headline net income figure, since most of it came from a non-cash warrant revaluation gain. The underlying business is still operating at a loss, and management has guided for larger losses in the year ahead as they reinvest in scaling.

Rigetti Bets on the Hybrid Future

Rigetti Computing uses superconducting circuits cooled to near absolute zero, the same broad architecture that Google and IBM use for their quantum efforts. Superconducting qubits run roughly 4,000 times faster than trapped ion systems, but their error rates are higher, so the practical speed advantage shrinks closer to ten times in real workloads. Errors also make superconducting systems much harder to scale beyond a few hundred qubits on a single chip.

Rigetti is the only pure play public company in superconducting quantum, since Google and IBM both bury their quantum efforts inside much larger businesses. The company recently made its 108 qubit Ankaa processor generally available on AWS Braket, meaning any developer with an AWS account can rent time on a state of the art quantum machine the same way they would spin up a regular EC2 instance. That distribution channel matters because it puts Rigetti’s hardware in front of millions of potential customers without Rigetti needing to build a sales team to reach them.

The financial story is more speculative than IonQ’s. Full year 2025 revenue came in at $7.1 million, with the most recent quarter actually declining 18 percent year over year to $1.9 million. Rigetti is still firmly pre-commercial, and revenue is not the right metric to evaluate them on. What matters is the cash position, which sits near $590 million against a burn rate of roughly $20 million per quarter. That gives Rigetti about 7.5 years of runway, which is plenty of time to commercialize. The stock works if you believe Rigetti becomes the default quantum partner inside the Nvidia stack and rides the hyperscaler integration trend over the rest of the decade.

D-Wave Already Has Paying Customers

D-Wave is the unusual one in the group. Their core product is quantum annealing, which is not a general purpose quantum computer at all. It is a specialized machine optimized for one class of problem, namely combinatorial optimization. Scheduling, routing, materials design, and portfolio construction all fit this category, which is why companies like Volkswagen, Lockheed Martin, NASA, and Mastercard have run real workloads on D-Wave systems.

In January, D-Wave closed a $550 million acquisition of Quantum Circuits Inc, which added a gate based superconducting platform to their existing annealing business. The company now has two completely different quantum architectures under one roof. They do not need both to win. They only need one of them to mature into a commercial business, which gives D-Wave a kind of optionality the other two pure play companies lack.

Revenue came in at $24.6 million for the full year, up 179 percent year over year, with $2.8 million in the most recent quarter. Bookings have been even more impressive. The company reported $32.8 million in new bookings in just January and February alone, more than the full prior year of revenue in two months. The balance sheet holds around $630 million after the acquisition, against a burn of roughly $70 million per year. That gives D-Wave nearly nine years of runway, which is comfortably long enough to wait for the broader market to mature.

The Risks Are Not Small

Every investor in this space needs to understand three things before sizing a position. First, the revenue is still tiny relative to market caps. IonQ is the only company over $100 million in annual sales, and the entire sector still depends heavily on government and defense contracts. Any major federal budget cut would hit all three companies at the same time. Second, these stocks routinely move 30, 40, or even 50 percent in a single day in either direction. The volatility is structural, not temporary, because the floats are small and headlines move sentiment violently.

Third, all three companies are betting on fundamentally different physical approaches to building quantum computers. Trapped ions, superconducting circuits, and quantum annealing are not interchangeable technologies. Whichever architecture wins the long term scaling race, the other two will struggle. This is not like investing in cloud companies in 2010, where the rising tide lifted every boat. The next decade will probably produce one dominant winner per problem category and several losers along the way. For more on how to think about asymmetric technology bets like this one, my framework for contrarian position sizing covers the rules I use to keep speculative exposure from blowing up the rest of the portfolio.

Picking a Winner in the Quantum Race

If forced to pick a single name today, IonQ has the cleanest risk to reward ratio. Revenue over $100 million, a $370 million backlog, the strongest balance sheet in the sector, and a working photonic interconnect that proves the scaling path is real. IonQ is the most expensive of the three on a market cap basis, but it has actually earned the premium with execution.

Rigetti is the right bet for investors who believe in Nvidia’s hybrid computing thesis. The company is the smallest of the three, but it is also the most directly tied to the world’s most important technology company through the NVQLink integration and the AWS distribution channel. D-Wave fits investors who want exposure to paying customers today plus optionality on a second architecture. The annealing business is real and growing fast, and the gate based addition turns the company into a genuine two horse race within a single ticker.

Personally, I think the right move is to own all three rather than try to pick a winner. The architectures are too different and the science is moving too fast for anyone outside the labs to confidently call the long term leader. This is a sector to size carefully, accept volatility on, and hold for the decade ahead. The bubble already popped. The technology is now real. The businesses behind these tickers are finally starting to look like investments rather than lottery tickets.