I am working with one of these dilution refrigerators of Oxford Instruments in my PhD thesis (not directly quantum computing, but a related topic in the direction of quantum computing). They are an incredible piece of machinery, with the push of a button they cool down your sample from room temperature to about 20 mK (admittedly, it takes ~30 Hours). That's 0.002 degrees above absolute zero.
I wonder how much of the 500k$ is really the material cost. As it was said in the article, the Helium-3 gas is already 40k$. And there is a lot of high-purity metals used: gold-plated oxygen-free copper for the main parts, silver wire for thermal anchoring and heat exchangers, indium wire for vacuum seals etc.
And thats just for the cryostat. To really use the cryostat for quantum computing experiments, you need to buy all of the cryogenic high-frequency electronic stuff. The cryogenic cables mentioned in the article by Coax, Ltd. are indeed really really expensive. Our last batch of NbTi-NbTi cables (i.e. superconducting inner + outer conductor) were about 1500$ per meter. Cryogenic microwave amplifiers (they operate at ~3.5 K to reduce microwave noise) cost about 5000$ per piece. Microwave circulators run at about 1500$ per piece. (According to the google paper, they used 9 cryogenic amps and 45 cryogenic circulators).
It's really very fascinating. But on the other side, I guess there is just not enough demand to justify founding another company building these fridges or superconducting cables. Unless of course at some point you manage to build these systems in large scale. But right now, that is just not feasible because it's no plug-and-play system. You have to constantly adjust and recalibrate the electronics to make the chip work.
The article talks about superconducting cables, but these cryogenic cables are not superconducting. Seem like regular Kapton flex cables, either strip line or broadside coupled.
Misleading title; also the article is mostly fluff. TLDR; quantum computers are expensive because superconductors are expensive to maintain, and companies like IonQ and Delft Circuits are looking into some solutions.
Quote: "The problem is that these huge cylinders, which can cost between $500,000 and $1 million each, are custom-made, and researchers say that only a few companies, like BlueFors in Finland and Oxford Instruments in the UK, are producing high-quality ones."
There, the next unicorn - why only UK & Finland and not a Silicon Valley start-up as well? In the end it's just tech that requires investment, and God help us, there are plenty of venture capitalists looking for next unicorn. I mean the quantum computing field is so hot these days that everyday you find a new article about it.
This is a poor choice for a unicorn. There are incumbents making dilution refrigerators already, and it's unclear as to how your products are any better.
Yes, the lead times are long right now - but why? My assumption is that there is so little demand that these units are built-to-order, and additionally there's probably a bit of customization on each unit for end-customer requirements. I don't think there's a long lead time because they are busy. So, the volume probably isn't there yet for a company to swoop in with mass-produced cheaper dilution refrigerators. And even once there is - the incumbents are probably better positioned to capitalize on that than you are.
I'm a little unclear on the benefit here. Say I have a hard problem and want to spend say $15 million for one machine, and further say that my problem might be suitable for a quantum algorithm, and further say I'm okay restocking all the exotic cryogenics etc.
At that price, I could buy at least 15,000 boring old machines using a boring distributed algorithm.
> "The problem is that these huge cylinders, which can cost between $500,000 and $1 million each, are custom-made, and researchers say that only a few companies, like BlueFors in Finland and Oxford Instruments in the UK, are producing high-quality ones."
It's a wonder that manufacturers of audiophile equipment have not yet jumped into this space.
Audiophile equipment is nothing like a multistage cryogenic refrigerator. It's made of pumps, seals, heat exchangers... and when you're using helium 4 as a coolant you really need zero leakage because it's an extremely expensive.
And yeah, microwave systems need amplifiers going up, and filters going down. But they need to be extremely efficient -- and also entirely non-magnetic. But unlike audio, these components need to be linear in the gigahertz, not kilohertz. There's some similarity, but huge differences.
It's weird that the title focuses on cables. There's a ton of expensive, labor-intensive components that can be bottlenecks.
> Audiophile equipment is nothing like a multistage cryogenic refrigerator.
I was thinking more in big picture terms. Extremely expensive cables that you sell to people who believe their effectiveness depends on observing them...
But here, the effectiveness really depends on using superconducting cables, i.e. there really is a difference if you use superconducting cables or e.g. silver-coated stainless steel.
I wonder how much of the 500k$ is really the material cost. As it was said in the article, the Helium-3 gas is already 40k$. And there is a lot of high-purity metals used: gold-plated oxygen-free copper for the main parts, silver wire for thermal anchoring and heat exchangers, indium wire for vacuum seals etc.
And thats just for the cryostat. To really use the cryostat for quantum computing experiments, you need to buy all of the cryogenic high-frequency electronic stuff. The cryogenic cables mentioned in the article by Coax, Ltd. are indeed really really expensive. Our last batch of NbTi-NbTi cables (i.e. superconducting inner + outer conductor) were about 1500$ per meter. Cryogenic microwave amplifiers (they operate at ~3.5 K to reduce microwave noise) cost about 5000$ per piece. Microwave circulators run at about 1500$ per piece. (According to the google paper, they used 9 cryogenic amps and 45 cryogenic circulators).
It's really very fascinating. But on the other side, I guess there is just not enough demand to justify founding another company building these fridges or superconducting cables. Unless of course at some point you manage to build these systems in large scale. But right now, that is just not feasible because it's no plug-and-play system. You have to constantly adjust and recalibrate the electronics to make the chip work.