The Quantum Computing Bottleneck Is Not the Qubits

Everyone talks about qubit counts. The race to 1,000, then 10,000, then a million qubits is how the industry tells itself it's making progress. But the machines that actually run hit a wall before they run out of qubits. They run out of room.

Here's the constraint. A superconducting quantum processor needs to sit at roughly 20 millikelvin — colder than outer space. Every qubit needs control signals sent from outside the refrigerator, through coaxial cables that conduct heat inward. IBM and Google have handled this by adding more cables, proportionally to the qubit count. That works for 100 qubits. It breaks for 10,000. Each cable brings in heat. Every cable takes up space. By the time you have tens of thousands of qubits, you need a cryostat the size of a room and cooling power that doesn't exist.

The problem is so well-known that multiple teams are working on it from different directions. The obvious fix: move the control electronics inside the refrigerator, next to the qubit chip. But regular electronics dissipate too much heat to survive at millikelvin temperatures. They would warm the qubits and destroy the fragile quantum states.

So you need transistors that can switch fast and burn almost nothing. Superconducting transistors.

That's what a Finnish startup called S-Transistors is building. On August 31, 2026, it raised 2.6 million euros in a pre-seed round led by Lifeline Ventures, a Nordic early-stage investor that recently raised a 400 million euro fund. The company was spun out of VTT Technical Research Centre of Finland. Its co-founders — Heorhii Bohuslavskyi (CEO) and Andrey Generalov (CTO) — were VTT researchers who published a 2024 paper demonstrating graphene Josephson field-effect transistors fabricated at wafer scale using CMOS-compatible processes, with room-temperature yields exceeding 90%.

The device is unusual. It uses graphene — a single layer of carbon atoms — as the channel between two superconducting electrodes. The resulting Josephson field-effect transistor can be turned on and off by a tiny voltage, and when current flows it dissipates essentially no power. In the research, critical currents reached about 3 microamperes for a 50-micron-wide channel and could be tuned by an order of magnitude. The devices held up through thermal cycling. They were fabricated on 150mm silicon wafers.

The first product is a multiplexer designed to plug into existing cryogenic setups, expected to ship to early customers within the first year. The longer-term goal is what the company calls a "quantum motherboard" — a classical control layer operating at millikelvin temperatures, sitting next to the quantum processor.

This isn't a unique idea. The approach of moving control electronics into the cold is the dominant research direction. SEEQC, an American company that raised over $64 million and is planning a Nasdaq IPO at roughly a $1 billion valuation, uses Single Flux Quantum logic — a different superconducting digital technology — to integrate control circuits at the 20 millikelvin stage. SEEQC published results in Nature Electronics in March 2026 showing five-qubit control with gate fidelities above 99.5%. Google, Intel, and IBM are all developing cryo-CMOS variants of conventional silicon transistors optimized for low-temperature operation.

But S-Transistors is different from SEEQC in one concrete way. SEEQC builds monolithic superconducting chips using proprietary processes in its own foundry. S-Transistors' graphene Josephson transistors are CMOS-compatible. They can, in principle, be co-fabricated with standard silicon circuits on the same chip. If that holds at production scale, the company could ride existing semiconductor manufacturing infrastructure instead of building its own foundry. That matters because foundries are expensive and slow. The ability to use the infrastructure that already exists for every other chip in the world is a real advantage — if the physics holds up under manufacturing stress.

There's a catch that the research itself admits. The graphene-to-superconductor interfaces in the published devices have low junction transparency, with the measured critical-current product at only about 20-40 microelectronvolts, well below the theoretical limit. The authors flagged improving interface quality as the next milestone. In a product context, lower-than-theoretical transparency means slower switching or higher error rates. It may not matter for a multiplexer, but a full motherboard controlling thousands of qubits would be unforgiving.

And there's a deeper uncertainty about whether the control layer is the bottleneck or just the next bottleneck. Some researchers argue that materials quality and qubit coherence are still the harder problems. You can have perfect control electronics and still lose computation because the qubits decohere. The wiring problem only matters if the qubits themselves get good enough first.

None of this is an investment in the traditional sense. S-Transistors is private, at pre-seed, and will likely need multiple rounds of funding before any revenue. Quantum computing companies routinely consume tens of millions per year with uncertain paths to profitability. The 2.6 million euros will fund prototype development, a cryogenic lab, and a pilot manufacturing line. It won't be enough to reach product-market fit.

But there's something worth watching here. If S-Transistors' graphene Josephson transistors prove manufacturable at CMOS-compatible foundries, and if they can switch cleanly enough to control quantum processors, this could become the kind of company that no quantum computer builder can ignore — the way TSMC matters to anyone who designs chips but doesn't make them. A semiconductor play embedded inside a quantum infrastructure play.

The way to evaluate this isn't to wonder whether quantum computing will arrive. It's to watch whether the qubit builders hit the wiring wall. If the qubit counts keep going up and the control cabling stays the same proportion, the bottleneck will become impossible to ignore. If companies start announcing control-layer partnerships or acquiring cryogenic electronics startups, that's the signal that the industry has reached the physical limit of the cable approach.

What to watch for: whether S-Transistors' multiplexer actually ships within the year, and whether any of the major quantum computing companies — Google, IBM, IonQ, Quantinuum — announce it as a supplier or partner. That would be the first evidence that the graphene Josephson transistor works well enough to leave the lab.