Imagine the California Gold Rush of 1849. Most prospectors went broke. The two reliable winners were Levi Strauss (denim pants for miners) and Sam Brannan (the only guy in San Francisco selling shovels and pans). Quantum computing is in its 1849 moment — except no one knows yet whether the gold is even in the river. Trapped-ion qubits, superconducting qubits, photonic qubits, neutral-atom qubits, silicon spin qubits, topological qubits: there are at least six different ways being raced against each other right now, and most will not become commercial in the next 5-10 years.
But every single one of those approaches — except photonic and trapped-ion — requires being cooled to 10 millikelvin (1/100th of a degree above absolute zero, colder than deep space). The only commercial machine that can do that is a dilution refrigerator . Two companies make them: Bluefors (Finland, private) and Oxford Instruments (UK, public). They share over 70% of the market.
Add to that: every quantum computer needs lasers (Coherent, Lumentum, Hamamatsu), specialty vacuum and gas systems (MKS), microwave control electronics (Keysight), and very specific cryogenic cables. You don't need to know which qubit architecture wins. The picks-and-shovels companies get paid regardless. And the deeper choke — helium-3 , of which the world has about 1 kg per year of natural decay supply against demand of 40,000 liters per year — is the resource that limits the entire industry's growth rate.