What's up with 'Quantum' lately? The surprising things slowing it down

I'm no quantum computing expert, by any stretch of the imagination. Yet even casual observers have probably heard the breathless promises coming from the field over the decade. Chemistry simulations that would take classical supercomputers millennia will run in hours. New medicines that currently take a decade to develop could be designed in months. The encryption protecting everything from your bank account to state secrets will become obsolete. The future is tantalizingly close.

At least... that's what people were saying. What about now?

What you probably haven't heard is that one of the biggest obstacles standing between us and that future is a colorless, odorless gas that exists primarily as a byproduct of decaying tritium.

Welcome to the strange reality of quantum computing in 2025, where the constraints aren't always where you'd expect them.

The Refrigerator Problem

To understand what's holding quantum computing back, you need to start with a fundamental physical reality: today's quantum computers must operate at temperatures approaching absolute zero. These systems require specialized refrigeration equipment called dilution refrigerators that can reach temperatures in the millikelvin range,[1] typically between 10 and 20 millikelvin for superconducting qubits.[2] At room temperature, the delicate quantum states that make these machines work simply collapse into noise.

These aren't the kind of refrigerators you can order from a catalog. They're precision instruments, and they rely on a careful dance between two isotopes of helium: helium-4 (the common party-balloon variety) and helium-3.

Helium-3 is where things get interesting.

The Isotope Nobody Planned For

Helium-3 is vanishingly rare in nature. The only practical source at scale is as a byproduct of tritium decay, which has a half-life of 12.3 years.[3] Tritium comes primarily from nuclear weapons programs. When the United States maintained a large active nuclear arsenal during the Cold War, tritium production created a steady supply of helium-3 as a side effect. Nobody particularly wanted it. It just accumulated.

After the Cold War ended, tritium production slowed. Helium-3, once accumulating in storage, started to look like a strategic resource. Low-temperature physicists estimate they need between 2,500 and 4,500 liters per year,[4] primarily to fill new dilution refrigerators.

Then quantum computing started to scale.

Every dilution refrigerator can require significant amounts of helium-3. As quantum computing installations expand globally, the supply has become increasingly constrained. The helium-3 supply has fallen short of demand,[5] creating a physics-limited bottleneck.

Tritium decay proceeds at a fixed rate determined by nuclear physics. You can't speed that up. You can't substitute something else. Every quantum computing lab in the world is effectively waiting in line for a gas that arrives at the pace of radioactive decay.

I can't help but feel a hint of irony. Possibly the most advanced computing technology humanity has ever conceived is constrained by decades-old gas from 1950s Cold War nuclear weapons stockpiles, waiting on radioactive decay.

What Looks Like a Bottleneck But Isn't

Here's where the story takes an unexpected turn, because not every constraint in quantum computing is what it appears to be.

If you've tried to experiment with quantum computing through cloud platforms like AWS Braket, IBM Qiskit Runtime or Microsoft Azure Quantum, you've likely encountered limits. AWS Braket allows 35 concurrent quantum tasks by default for its SV1 simulator, with a maximum of 100 in most regions.[6] IBM gives you 10 free minutes per month on its open plan, approximately 400 minutes per year on its flex plan and over 5,000 minutes on its premium tier.[7] Azure Quantum meters access to hardware through various pricing mechanisms.

These feel like bottlenecks. They shape who gets access, how much experimentation can happen and how quickly education programs can scale. Universities and startups often find themselves resource-constrained.

That said, these aren't supply bottlenecks in the traditional sense. They're pricing mechanisms. AWS isn't running out of computing infrastructure when it throttles your simulation. IBM isn't facing a shortage of quantum processing units when your monthly minutes expire. These constraints exist because of commercial allocation policies, not because of any physical scarcity.

You can resolve them in days, sometimes hours, with the right budget or a quota increase request.

Compare that to helium-3, where no amount of money will make tritium decay faster. Or to the specialized fabrication facilities needed to manufacture superconducting qubits, where only a handful of pilot production lines exist worldwide. Those are real bottlenecks.

The distinction matters a lot. When policymakers or investors talk about "accelerating quantum computing," they need to know which obstacles are economic and which are structural. Throwing money at a pricing policy won't do much. Investing in isotope production infrastructure or advanced manufacturing facilities might.

The Hidden Choke Points

Beyond helium-3, the quantum value chain reveals several other surprising constraints that most people never hear about.

Superconducting qubit fabrication is bottlenecked by the lack of production capacity. Building these qubits requires clean-room facilities with specialized processes for depositing thin films and creating nanoscale junctions. Only a handful of sites worldwide offer open access to these capabilities for multi-project wafer runs, creating capacity constraints for the dozens of research programs seeking access.

Cryogenic testing equipment represents another constraint. Once you've fabricated qubits, you need to test them at operating temperatures. That requires probers that can operate at or below 2 Kelvin while precisely measuring wafers. Only a few institutions have these specialized tools, and building and integrating them into existing facilities takes years.[8]

Laser systems for trapped-ion quantum computers face supplier concentration. Companies like IonQ rely on custom-built racks of ultraviolet and visible lasers with narrow linewidths and fast beam steering. Only a small number of vendors can build these systems, and when one supplier fails (as M Squared Lasers did in August 2025),[9] the ripple effects spread across the industry. Custom optical integration causes lead times to stretch to six to twelve months.

Even something as seemingly mundane as real-time error correction decoders turns out to be a severe bottleneck within research and development. Quantum computers will eventually need to correct errors on the fly to achieve useful computation. Right now, companies like Riverlane dominate commercial supply of quantum error correction technology,[10] with limited deployment in labs and commercial facilities worldwide.[11] Integrating these decoders with existing control systems takes months of engineering work.

None of these constraints are the kinds of things that make headlines. They don't involve flashy demos or science fiction-like omniscience (like in the 2020 miniseries 'Devs').[12] They're about industrial capacity, supply chain depth and the unglamorous work of scaling manufacturing.

This is the reality of emerging technology. The distance between a lab proof-of-concept and industrial deployment isn't just about science. It's about the entire ecosystem of suppliers, fabricators, integrators and logistics providers that need to mature in parallel. It isn't about a point solution, it's about a value chain.

The Moon Mining Solution (Maybe?)

The helium-3 shortage has sparked some genuinely science-fiction proposals. Lunar regolith contains helium-3 that has been deposited by solar wind over billions of years. Mining the moon for helium-3 and shipping it back to Earth is no longer purely speculative. In 2025, Bluefors (a leading dilution refrigerator manufacturer) signed an agreement with Interlune, a lunar resources company, to purchase up to 10,000 liters per year of helium-3 for delivery from 2028 to 2037.[13]

That timeline feels both audacious and telling. We might establish moon mining operations before we solve terrestrial isotope production. Unless some new huge breakthrough happens that nobody is expecting, the exotic solution may arrive before the mundane one.

What This Means for the Quantum Future

Quantum computing is real. The technology works. Researchers can demonstrate quantum effects, run meaningful algorithms and solve specific problems faster than classical computers.

What's missing is the industrial maturity to scale from dozens of machines to thousands, from research labs to commercial deployments, from proof-of-concept to routine operation.

The path forward isn't entirely mysterious. We need expanded isotope production capacity, whether from accelerated tritium programs, new nuclear processes or yes, maybe even moon mining. We need more fabrication facilities that can produce superconducting qubits at scale, with open access for researchers and startups. We need deeper supply chains for cryogenic equipment, quantum-grade lasers and specialized components.

None of this is theoretically impossible. It's just infrastructure. It's the kind of thing that advanced economies know how to build when the incentives align. The question is whether the investment will arrive in time to meet the ambitions.

The next time someone tells you quantum computing is five or ten years away from transforming the world, you'll know to ask a different question. Not whether the qubits will work, but whether we'll have enough helium-3 to keep them cold. The answer to that question might just depend on whether we can mine the moon before the tritium supply falls further behind demand.

Sometimes the future doesn't hinge on genius or breakthrough moments. Sometimes it hinges on plumbing.

Endnotes

[1] SpinQ, "The Complete Guide to Dilution Refrigerators" (2025), https://www.spinquanta.com/news-detail/the-complete-guide-to-dilution-refrigerators

[2] International Institute of Refrigeration, "Successful test of a dilution refrigerator for quantum computers" (October 27, 2022), https://iifiir.org/en/news/successful-test-of-a-dilution-refrigerator-for-quantum-computers

[3] U.S. Environmental Protection Agency, "Radionuclide Basics: Tritium" (February 6, 2025), https://www.epa.gov/radiation/radionuclide-basics-tritium; Wikipedia, "Tritium" (accessed October 2025), https://en.wikipedia.org/wiki/Tritium

[4] Adrian Cho, "Helium-3 Shortage Could Put Freeze On Low-Temperature Research," Science (October 2009), https://www.science.org/doi/10.1126/science.326_778

[5] Ibid.

[6] Amazon Web Services, "Amazon Braket Quotas" (2025), https://docs.aws.amazon.com/braket/latest/developerguide/braket-quotas.html; Amazon Web Services, "Submitting quantum tasks to simulators" (2025), https://docs.aws.amazon.com/braket/latest/developerguide/braket-submit-tasks-simulators.html

[7] IBM Quantum, "Pricing and access plans" (2025), https://www.ibm.com/quantum/products; Paul Smith-Goodson, "RESEARCH NOTE: IBM's New Flex Plan Fills a Big Gap in Quantum Access Pricing," Moor Insights & Strategy (June 19, 2025), https://moorinsightsstrategy.com/research-notes/ibms-new-flex-plan-fills-a-big-gap-in-quantum-access-pricing/

[8] International Institute of Refrigeration, "Successful test of a dilution refrigerator for quantum computers" (October 27, 2022), https://iifiir.org/en/news/successful-test-of-a-dilution-refrigerator-for-quantum-computers

[9] The Quantum Insider, "M Squared Lasers Placed Into Administration, Seeks Buyer For Assets" (August 28, 2025), https://thequantuminsider.com/2025/08/28/m-squared-lasers-placed-into-administration-seeks-buyer-for-assets/

[10] Riverlane, "Deltaflow: The Quantum Error Correction Stack" (2025), https://www.riverlane.com/quantum-error-correction-stack; Riverlane, "Deltaflow.Decode, the world's most powerful quantum error decoder" (2025), https://www.riverlane.com/products/deltaflow-decode

[11] Riverlane, "Riverlane and OQC Move Toward Fault-Tolerant Quantum Computing with QEC Integration" (July 22, 2025), https://www.riverlane.com/press-release/riverlane-and-oqc-move-toward-ftqc-with-qec-integration

[12] Devs, created by Alex Garland, FX on Hulu, 2020, https://en.wikipedia.org/wiki/Devs_(TV_series)

[13] Bluefors and Interlune, "Bluefors to source helium-3 from the Moon with Interlune to power next phase of quantum industry growth" (September 16, 2025), https://bluefors.com/press-releases/bluefors-to-source-helium-3-from-the-moon-with-interlune-to-power-next-phase-of-quantum-industry-growth/; The Quantum Insider, "Bluefors Enters Deal to Secure Lunar Helium-3 Supply From Interlune" (September 2025), https://thequantuminsider.com/2025/09/17/bluefors-enters-deal-to-secure-lunar-helium-3-supply-from-interlune/