Andy Cobin is a microwave engineer and the founder of Quantum Microwave. His company manufactures cryogenic microwave components for superconducting quantum computers. He started the business after identifying a need for reliable, commercially available parts for researchers in the quantum field.

Nate Wheeler is the host of the popular Manufacturing Insiders podcast. He also owns weCreate, a nationally recognized marketing agency that helps manufacturers grow, save money, and become more efficient.

In this episode of Manufacturing Insiders, Andy Cobin discusses the challenges of building hardware for an industry that is still largely experimental. He explains how major research institutions were struggling to find microwave components that could function in the extreme cold required for quantum computing. Cobin saw an opportunity to create a business that designs and manufactures these parts specifically for cryogenic environments.

Cobin shares how his company addresses the engineering problems of making components that operate near absolute zero. He details his approach to testing and validation when the necessary equipment costs hundreds of thousands of dollars. Learn how he built an e-commerce marketplace to streamline the supply chain for this niche.

---

Nate (00:00.834) Welcome to Manufacturing Insiders. Today I have Andy Cobin with me. He’s a fascinating individual in a fascinating field. Quantum computing is something we hear thrown around a lot, and he has insider information about how quantum computers are made. He manufactures some of the coolest components I’ve ever seen – we’re talking gold-plated superconductors.

Andy, welcome to the podcast today. I can’t wait to learn more about what you know.

Andy Cobin (00:34.025) Thanks for having me, I appreciate it.

Nate (00:37.26) Could you give us a little background? You started off getting a degree as a microwave engineer and you’ve been working in this field for almost as long as I’ve been alive. Tell me a little bit about your story.

Andy Cobin (00:54.367) I went to Wentworth back in the 80s and graduated in 1985. Then I worked for a company called Microwave Associates, where we made microwave components. I was a design engineer and a product engineer there. I moved on from there to other companies and even had my own microwave company years ago where we were a microwave assembly company.

Then in 2016, I decided to start Quantum Microwave, building cryogenic microwave components for superconducting quantum computers.

Nate (01:42.402) I would have to assume that you know quite a bit about how quantum computers work, given that you’re making components for them.

Andy Cobin (01:54.228) That’s not really the case. It goes back to 2016 when I visited IBM, Google, and Johns Hopkins APL. They all said the same thing to me: “Andy, we’re PhDs, researchers, and scientists making these superconducting quantum computers. Nobody makes the proper microwave components for quantum computers. So we’re buying everyone’s components, putting them in our dilution refrigerator at 10 millikelvin or four Kelvin, and finding out which ones work.”

That made no sense to me. That’s when the idea came that I should start a company designing components to actually work properly in quantum computers. I’m not a quantum computer expert, but I’m an expert in the cryogenic components that are needed in quantum computers. Does that make sense?

Nate (03:12.234) It does. When we talk about the fundamental unit of quantum computing, it would be the qubit, right? To me, it’s really mind-blowing to try to wrap my head around this. I understand the double-slit experiment where they proved that an electron or a photon could act like a wave or a particle. That’s the fundamental concept here – there’s this duality of the qubit where it can be either one, a zero, or both of them simultaneously.

Andy Cobin (03:21.843) The qubit, yes.

Andy Cobin (03:58.558) Exactly. A qubit can operate in multiple states where conventional computers are binary – it’s a one or a zero. Qubits can be a one and a zero simultaneously. All the researchers are trying to come up with the most powerful quantum computer, and my job on the hardware side is to protect that qubit.

I’m making components that amplify the qubit, components that split the qubit, sending qubits in certain directions, filtering the qubit, and protecting the qubit from interference reflecting back and damaging it. We’re making the components that allow the qubit to get to the outside world and still operate properly. That’s the goal of what we’re doing.

Nate (05:09.248) Are you actually manipulating photons?

Andy Cobin (05:21.119) The qubit is in a microwave frequency. The qubit is generated by Google, IBM, Amazon, or any of these quantum computing companies. They make something called a quantum processor, and the quantum processor can be 10 qubits, 100 qubits, 200 qubits, thousands of qubits – eventually they want to get to millions of qubits.

Every qubit is its own individual frequency. The qubit signal is very low, so it’s not possible to get that signal out to the outside world. The first thing we do is amplify the qubit to make the signal better. Then we isolate the qubit using components called isolators, circulators, amplifiers, and filters.

We make components that start at room temperature and decrease the power. You decrease the power so you’re decreasing the heat, going from room temperature down to millikelvins in multiple stages. We have to keep reducing the power until we get to 10 millikelvin where the quantum processor is operating.

Then we use components that amplify the signal in a superconducting mode. The idea is we want to amplify the signal, but we don’t want to use any DC power which generates heat. That’s called a superconducting amplifier. We sell a traveling wave parametric amplifier and a Josephson junction parametric amplifier.

Nate (07:11.854) Mmm.

Andy Cobin (07:12.713) We get down to the millikelvins and we’re controlling the qubit. Now we have to bring the qubit back up to the room temperature world. So we come down to the qubits and then we have to bring it back up to the room temperature world. We make those types of components.

Here’s the corrected version with all filler words and meaningless interjections removed:

Nate (07:53.42) It’s been a while since I had my last physics class, but with Kelvins, the number 273 stands out in my mind. What is a millikelvin? How would that translate?

Andy Cobin (08:04.349) I think 10 millikelvin is 10 thousandths of a degree above absolute zero. I don’t remember the exact temperature, but it’s minus 400 and something – 470 degrees Fahrenheit or something like that. It’s very, very cold.

Nate (08:26.446) Yeah, so zero Kelvin is negative 273 Celsius, or negative 459 Fahrenheit. We’re talking about some very, very, very cold temperatures. What are they using to achieve that type of cold?

Andy Cobin (08:48.095) I think they do it with helium.

Nate (08:52.596) That makes sense. So this business, Quantum Microwave, that you have now kind of started off with you reselling somebody else’s product, and then you decided that you were going to get into the actual manufacturing process, right?

Andy Cobin (09:09.213) Right. We started off – and we still do – selling for a company called Low Noise Factory, which is out of Gothenburg, Sweden. Low Noise Factory makes an LNA, which is an indium phosphide HEMT. That particular material, as you go colder in temperature, the noise keeps dropping – the noise figure or the noise temperature keeps dropping.

When you amplify the qubit, you have to amplify it with no noise or as little noise as possible. This company makes the best cryogenic low noise amplifier in the industry. Everybody needs that amplifier. They’re in Sweden and they asked if I would open a company here in the US to distribute their products.

Andy Cobin (10:07.123) That’s what we were doing. As I’m distributing the products, I asked the owner of Low Noise Factory if he would mind if we developed complementary products to his product. If he’s making the amplifier, someone needs to make an isolator and someone needs to make a circulator. Someone needs to make filters, couplers, and all of the different components that go around it.

That’s what we started to do. We kept adding them to our marketplace. Then other companies came to us and said they wanted us to sell their products. We’re now probably about five or six companies that we distribute, plus the products that we make.

When you go to our marketplace, you can buy almost every cryogenic microwave component that the Amazons, Googles, IBMs, and research facilities need, and they don’t have to wait six months for a product. They go on our website, place an order, and the product ships that day. We’re an in-stock online marketplace for state-of-the-art cryogenic components.

Nate (11:32.898) Did you bring some samples for us? You said you were going to have some samples, or should I pull up your website?

Andy Cobin (11:39.519) Pull up my website. I got tied up and didn’t get that done. Sorry about that.

Nate (11:45.504) No problem. I wanted to show off some of this stuff because I thought it was so cool – gold-plated components.

Andy Cobin (11:56.136) They’re not just gold-plated components – it’s the material under the gold plating. Go under where it says cryogenic components. That’s probably the best one because we make other products too. Then scroll so you can see the components.

Nate (12:14.286) I’ll click on the whole category.

Andy Cobin (12:25.629) Yeah, keep going. The first one where it says GQE – that’s a company we sell for called Gateway Quantum. Those are very low frequency filters, 25-line filters. The material will be made out of OFHC copper, which has very good thermal conductivity.

Andy Cobin (12:54.077) Then it’ll be plated with gold, but it won’t have any nickel under it. We make these filters so they’re good thermally, they have good RF performance, and they’re non-magnetic. Magnetic fields also hurt the qubit.

If you scroll down a little bit further, you’ll see the copper one – that’s an OFHC copper cavity. People can use that for experiments or research. They could put their filter in there or their superconducting circuit in there.

Andy Cobin (13:22.139) That’s just a test fixture. If you look at a few more of them, the ones that say “cryo BRD” – those are kind of like breadboards. You can put a quantum processor in there, or qubits, or amplifiers – whatever your research is. You can put these superconductor components in there, and they’re completely shielded from the outside world. They’re thermally excellent and stop any magnetic fields from going through.

Nate (14:14.05) How big is this item physically?

Andy Cobin (14:17.373) They range. Some could be this big and some could be this big. If you look at the bottom of a superconducting computer, you’ll see all of these things hanging. That’s where all of the magic is happening – where all of the physicists are developing all of their superconducting circuits. We call it a breadboard.

If an engineer were to make that, they would probably spend a year making the bracket, then they’d have to get a new metal shield made, then a superconducting shield and a copper shield. We just make it in all different sizes so the customer can click buy and be doing their experiments the next day. Can you scroll down some more?

Here’s the corrected version with proper paragraph breaks and all filler removed:

Andy Cobin (15:10.013) These ones here that you’re looking at are new filters that we’re putting in. These are high pass filters. What a high pass filter means is it only passes the frequencies that are above the frequency. A low pass filter passes the frequencies that are below and a band pass filter passes to a certain band.

Different places in the quantum computer need different types of filtering. We’ve created a whole family of filters. If they want a low pass filter to protect anything of higher frequency, a high pass filter to protect anything on the lower side, and a band pass for any frequencies on the high side and the low side.

Andy Cobin (15:39.104) If you keep scrolling down some more, you’ll see something that says a cryocoupler. That’s if a customer wants to send most of the signal straight through the line, but maybe they want just a little bit of a signal coming out for some reason – maybe for sampling or maybe to do something else. So we make a cryogenic coupler.

Andy Cobin (16:37.843) You’ll see a bias T. A bias T connects DC and RF. The bias T adds the DC and the RF together. If you see the thing that says the QIBO, those are actually qubits.

Andy Cobin (17:03.527) Yeah, so those are superconducting qubits. People put that in their system so they can characterize their qubit versus our state-of-the-art qubits. If you scroll down here, the one that says silent waves – those are traveling wave parametric amplifiers.

As we said, the qubit comes out from the quantum processor and brings out all the qubits, but we need to amplify the signal. This is the superconducting amplifier that’s used to amplify the qubits. Right next to it, you see something called the attenuator.

Those attenuators are used to drop the power. The power is very high at room temperature, and to get colder you need to drop the power so they use attenuators in each stage to drop the power. When they get down to 10 millikelvin they want to have almost no power because power equals heat, and if you’re heating up the plate you can’t achieve the 10 millikelvin temperatures.

That gives you an idea – we make pretty much every component from the amplifier to the filter to the attenuator to the coupler.

Andy Cobin (18:49.631) Sorry about that. That was my dog that just came in. We’re kind of – I don’t want to say the Amazon of cryogenic microwave components, but we’re something like that. We’re a marketplace where all of the researchers can come and do their experiments. The goal is the quicker we can get the experiments done, the quicker we can win the race to commercialize the quantum computer.

Nate (19:23.406) So where is the quantum computer as a whole right now? Are there functioning quantum computers that are solving problems, or are we still up against some issues we haven’t solved yet?

Andy Cobin (19:42.812) Yes, that’s not my expertise, but I would say we still have quite a few years to go. If you talk to experts, some might say five years, some may say 10 years. To get a good usable quantum computer, we need lots of qubits.

Right now on the superconducting side, people are in the hundreds of qubits and we need to get to the thousands of qubits and eventually to the millions of qubits. We need higher quality factor qubits and longer coherence time qubits. There’s big talk about all the software people doing error correction, and we need really good error correction to get the quantum computers to that usable space where they’re really going to solve real world problems.

I don’t think we’re quite there yet, but they’re definitely progressing.

Nate (20:56.814) When you look at your own sales – I think you said to me before that when it comes to that super-cooled portion of the quantum computer, you’re kind of the go-to resource. How have you seen it progress over the last five years?

Andy Cobin (21:20.915) We’ve been growing a lot every year. If IBM – and I’m not IBM, so I can’t speak for what their goals are – but if their goals are to double the qubits every year, or Google double the qubits every year, Microsoft double the qubits every year, that means these systems are getting bigger and they have double the components in them or close to double. Every year the component growth is just huge.

We keep growing every year. But you can’t just keep growing computers to the size of a building that take a power plant to operate. As quantum computers are growing in qubits, we have to start learning how to make components smaller.

We’re kind of at that infancy stage when the first computers were made – they took up the size of a building and now they’re the size of your cell phone. We’re in that phase now where all of the components are a certain size and they’re all connected by cables, and somebody’s going to one day invent a circuit that can do the whole thing.

Nate (22:44.738) A quantum watch, right?

Andy Cobin (22:47.453) Yeah, so there’s a long way to go there. Who knows if it will ever get there. We have to make these components that we’re making really good – we have to make them even better. We have to make them smaller.

And we have to make them less expensive because nobody can afford to buy a quantum computer right now. It would cost millions of dollars. There’s a long ways to go.

Nate (23:17.676) Yeah, when one breadboard costs 6,000 bucks and you need multiple on your computer.

Andy Cobin (23:30.239) Well, that breadboard is just a fixture. It’s $6,000 without anything on the inside. Then they have to put their much more expensive components inside that.

What we’re doing is taking the same components we have now and trying to make them so they’re surface mount or hybrid so that they can be built in these tiny little circuits. It’s all doable and it’s all going to happen.

Nate (24:05.538) I can imagine it would be an interesting set of challenges when you’re essentially manufacturing components for a largely theoretical and experimental space. When you’re in a space that’s established, like we talked about valves earlier, you got to turn off the flow of gas at a certain pressure – you know exactly what you’re trying to do. Whereas with you, it’s almost like you have to anticipate what they might be trying to experiment with and build around that.

Andy Cobin (24:41.087) What you need is the professors of these quantum computers and the researchers to work with the microwave engineers to tell us what they need. If we get a professor that will say, “Andy, this is great, but this is really what I need,” that’s how we go in that direction.

It’s interesting because in Europe – and we sell for a lot of European companies – the professors that develop incredible technology in Europe are allowed to take that technology and leave the university and start up their own company. That’s why we have a lot of these state-of-the-art components from Europe.

In the United States, maybe an MIT or Caltech or someone develops something incredible that can change the world, but they can’t take it out of the university. It belongs to the university.

Nate (25:55.099) That’s problematic.

Andy Cobin (25:57.200) Yeah, at least that’s the way I understand it to be. We’ll get someone from Sweden or Switzerland and they’ll say we’ve developed this – will you market it for us? We say yes. We work a deal and we stock them here in the US and they’re available to all of these US companies.

Nate (26:23.118) What’s the most surprising thing about quantum computing that you think most people don’t understand?

Andy Cobin (26:31.679) I really don’t know those answers to be honest with you. We’re down in the nitty gritty, just making very simple components in a way that they work cryogenically. According to the experts, it’s going to change the world, but I don’t know.

Nate (27:01.646) How do you test the efficacy of the components that you’re building? When you say it’s splitting off the high frequency from the low, I would imagine there’s got to be some pretty complex testing procedures.

Andy Cobin (27:19.987) You need to buy a dilution refrigerator and those can range anywhere from $10,000 to a half a million dollars – it’s not something that we can afford. What we do is we’re located in Massachusetts and Massachusetts has something called the Core Technology Program out of UMass Massachusetts, and they offer Massachusetts residents the opportunity to use their systems.

We take our components to UMass, and they have very sophisticated components, and they test them for us at 10 millikelvin and at 4 Kelvin, and they give us the data. We have the same equipment to test it at room temperature. Then we put the cryogenic test information on our data sheets.

We just do a room temperature test kind of in volume. Otherwise, a dilution refrigerator to get to 10 millikelvin takes two days to reach the temperature. Then you spend a day testing and it takes another two days to get back to room temperature. So every test you do takes a minimum of five days.

So if we did that on every component, we would never be able to make money or the product would be so expensive. So we give them a lot of 10 pieces and we let them run it at 10 millikelvin. They give us the data. We post it on our website. And then the researchers are able to say, OK, I can see the components from the quantum microwave. They’re going to work at 10 millikelvin. So yeah, that’s how we do it.

 

Nate (29:23.37) Wow. Yeah, that’s fascinating. I was just trying to think of the process of scaling. If they work out all the bugs and there’s just quantum computers going up all over the place, the manufacturing operation would need to have a testing facility and a lot of bars of gold to melt down.

Andy Cobin (29:53.044) Right. Some components that are not superconducting, you can test at not as cold temperatures like 4 Kelvin. Those can get down to temperature in an hour or two. But the ones that go into the superconducting, those are the ones that need to go down to the millikelvin temperatures. It’s a very expensive testing procedure for sure.

Nate (30:21.026) What have you learned about the e-commerce process? You started a website and started selling these. What’s some advice that you might have for somebody that has their own e-commerce store where they might be selling some custom manufactured components?

Andy Cobin (30:41.503) We came up with this idea of every component is its own part number with its own price. This is what you get. You’re not going to get anything else. If you want a different spec, you’re not going to get it. If you want a different spec, that’s a different part number. Everything is its own part number.

By doing that, we can let the customers just go and review all of the technical data and it’s just click buy, click buy, click buy. Every day, it’s great. They click, they buy, we ship it that day or the next day. The money comes into our bank the day afterwards. It’s just phenomenal.

It saves all of the time of preparing quotes. Then the customer has to take the quote to purchase and all these people have to sign it off and you eventually get the PO. Just click buy. It works great.

Nate (31:50.403) You found a hell of a niche. I mean, well, you didn’t find it – there’s very few people that would have the expertise to do what you’re doing. But that’s such a great place to be where you really don’t have a lot of competition in the space. If somebody needs something to operate in this area, the quantum computer, this is where you’re going to get it.

Andy Cobin (32:14.461) Right. We’re very nichey just in that small little cryogenic section of the quantum computer. Very nichey.

Nate (32:21.964) Well, pretty fascinating what you’re doing. Looks like you built a pretty awesome business. I wish all success to you in the future. Certainly, if you need any help on the marketing side, you know where to find it.

Andy Cobin (32:39.177) Thank you, appreciate that. Well, it’s good to talk to you.

Nate (32:42.314) Absolutely, Andy. Anything we missed you wanted to share?

Andy Cobin (32:47.263) No, I could just say we are Quantum Microwave, but we also focus on millimeter waves. If somebody’s making a satellite, we would make the links up in the 30 gigahertz, 40 gigahertz, 60 gigahertz, 100 gigahertz. That’s our other little niche – we focus on millimeter waves.

Nate (33:15.636) Gotcha. Communications.

Andy Cobin (33:19.453) Yeah, so we get a lot of commercial LEO space jobs and super high power amplifiers and radio astronomy actually works cryogenically as well. That’s how they look at what’s going on up in space.

Nate (33:39.608) So are some of these components things that would be involved like if you’ve got a rover up on Mars or something, and you need to obviously send a pretty high-frequency transmission to get it to Earth and to be able to read it, is that the sort of thing your components may be involved in that communication process?

Andy Cobin (34:00.723) I don’t know about a rover on Mars, but like LEO satellites, we could communicate with this. There’s no attenuation up in space. You could send data at 60 gigahertz with no attenuation from satellite to satellite, and then you convert it to a lower frequency and send it to earth. We would be in some of those. Not really sure about Mars or anything like that.

We’re more on the commercial side, but yeah, sure.

Nate (34:34.83) Very cool. Man, you’re in some complex stuff. I’m going to have to do some more reading after this and learn more about all of the language and terminology because I know a little bit but not enough to really dig into it. Thank you, Andy. Like I said, have a great rest of your day and wish you all the success.

Andy Cobin (34:51.66) Well, thank you very much.

Andy Cobin (35:00.347) Okay, take care.