Quantum computing is capturing the fascination and the imagination of the technology world.

It’s built upon the new and unfamiliar field of quantum physics — where principles such as superposition and entanglement offer the intrigue of unlocking entirely new opportunities. The world’s formerly unsolvable problems of logistics optimization, drug discovery, cybersecurity, and material design could now have a tool that potentially cracks those codes. Futuristic opportunities like solving climate change or operating in the space economy eagerly await on the horizon.

Yet for all of the promise that quantum computing holds and all of the attention it’s gained from the world’s greatest scientists, it’s still taking a frustratingly-long time to move beyond the R&D stage. Cloud service providers like Amazon (Nasdaq: AMZN) and Microsoft (Nasdaq: MSFT) have functional quantum computers that can selectively used in certain capacities. But a commercially-useful, error-corrected quantum computer still out of reach for the world’s business leaders to harness. We’re still a long way from finding Richard Feynman’s Killer App.

So what should investors do about this intriguing yet commercially-frustrating quantum opportunity? Are recently-public pure plays in the space like IonQ (NYSE: IONQ) about to make a quantum leap with upcoming breakthroughs? Are deep-pocketed tech companies like Alphabet going to completely rewrite the semiconductor supply chain? Or will quantum computing still remain in the “too hard” pile — fraught with technical risks, uncertainties, and no generally-agreed-upon commercial path forward?

To help us answer those questions, we’ve brought in an expert. Tiernan Ray is one of the technology industry’s best reporters. He’s covered the tech landscape for more than two decades – from the early days of the internet and the dotcom boom to the rise of cloud computing and artificial intelligence. Tiernan puts tech progress through a much-needed objective lens, helping investors separate hype from true innovation. He offers daily insights in his Technology Letter publication: www.thetechnologyletter.com.

In an exclusive interview with 7investing founder and CEO Simon Erickson, Tiernan describes where quantum computing stands and where it offers the greatest promise. He explains why controlled, multi-variable problems using AI are where quantum could be the most commercially valuable, and how companies will find multiple ways to support and profit from the bigger-picture movement. The two discuss the different technical approaches to quantum computing and the publicly-traded companies who are harnessing them. Tiernan also explains why we’re in the “the transistor age of quantum computing” and a few recent developments we should be keeping a closer eye on.

In the final section, Tiernan shares a few other technology trends he’s excited about — such as why silicon carbide is such an important material for the semiconductor industry and why high-performance computing still offer a long runway for chipmakers like AMD (Nasdaq: AMD) and NVIDIA (Nasdaq: NVDA).

Publicly-traded companies mentioned in this interview include Air Liquide, Alphabet, Amazon, AMD, ASML, BMW, Daimler, Form Factor, General Electric, Honeywell, Howard Hughes, IBM, Intel, IonQ, Microsoft, NVIDIA, On Semiconductor, Rigetti, Tesla, Texas Instruments, and Wolfspeed. 7investing’s advisors and/or guests may have positions in the companies that are mentioned.

Interview timestamps:

00:00 – Intro: Where do we stand with quantum computing?

03:33 – Where would quantum computing be the mort commercially useful?

06:33 – Which technical approaches to quantum computing are showing the most promise?

13:43 – Publicly-traded opportunities to invest in quantum computing

19:30 – Supply chain and manufacturing challenges related to quantum computing

31:42 – Two trends that Tiernan is very interested in: silicon carbide and high-performance computing

This interview was originally recorded on April 21, 2022.

Transcript

Simon Erickson  00:00

Hello, everyone, and welcome to today’s edition of the 7investing podcast where it’s our mission to empower you to invest in your future. I’m 7investing founder and CEO, Simon Erickson. Today we’re going to be talking about quantum computing. My goodness, there is so much opportunity in this field. But it’s also something that’s got a lot of challenges, it’s taking quite a bit of time to get figured out. I’m really excited to welcome Tiernan Ray to the show today. He’s the editor of The Technology Letter. He’s also, in my opinion, probably one of the greatest reporters on technology that I’ve followed for more than a decade. Yes, seriously, I’ve been a huge fan of his work for a long time. Tiernan, I’m really excited to have you as part of the 7investing podcast today. Welcome to our show.

Tiernan Ray  00:41

Thank you for those kind words. And thanks for having me back. Simon, it’s great to chat with you again.

Simon Erickson  00:46

And and let’s let’s chat at the 10,000 foot level about quantum computing. And you know, without getting too too in the weeds on this one just yet. We’ve seen a prediction from Gartner that they believe that 40% of large enterprises will have quantum computing related projects by the year 2025. That’s only three years away, which is shocking, almost, considering that less than 1% have any quantum computing projects at all today. Mostly because there’s really not a commercially viable quantum computer to work with for them.

Where do you think that quantum computing as a whole stands today? It seems like it’s still very early. But maybe more interestingly, where do you think the big commercial opportunity for this lies?

Tiernan Ray  01:29

There’s a number of computers you can get, as you know Simon, from Amazon Bracket. You can go to AWS and you can get, I think there’s five different computers now, from all from startups. D-Wave, Rigetti (Nasdaq: RGTI), Oxford Quantum Circuits, IonQ. And you can run a job, just like you run in any AWS job and they’re priced at. So recall, I think it’s 30 cents per circuit. So in quantum terms, if you run a single, basically the equivalent of a single line of a computer algorithm through one of these machines, it costs you 30 cents.

And so, to Gartner’s point, what that translates into for me is it’s live now on AWS Bracket, it’s live on Azure, it’s live on Google GCP, and you can go put up a credit card and run something on an actual hardware machine. They’re incredibly primitive machines. And more important, most people don’t know what a circuit would be to them or their business. They don’t really know what that means. It might be the equivalent of a basic “if then” statement run in quantum fashion.

So really primitive stuff. So you’ve got easy access via cloud to multiple machines from exciting startups. But primitive machines, and you have a whole cohort of people in business and science who might vaguely know what they would do, but are just beginning to learn how this runs in this quantum circuit, these quantum gates if you will. And have a long, long road to cover to figure out what a program, a full fledged program, would look like to run on that. So that’s where we are, in practical terms in terms of what you can do today.

Simon Erickson  03:31

There’s almost an endless number of potential opportunities for this, right? And when we look bigger picture, quantum kind of keeps running into the same ideas that are out there. The optimization problem, the traveling salesman, the cybersecurity issues, material design keeps coming up, drug discovery keeps coming up. It seems like quantum could potentially fix all problems the world ever has ever had about anything.

But it just seems like that might be hopeful thinking Tiernan. I mean, do you think that there are pockets of the industrial world or the commercial world that this would be most commercially useful? Especially considering it’s not a cheap price tag to get stuff done?

Tiernan Ray  04:11

I think those are problems, Simon, that people who work on incredibly complex multivariate analysis are interested in solving and could see applying their in controlled settings.

For example, PsiQuantum, one word, is a startup, right? They’ve got a paper out within the past month or so that was a research paper all using their computer. They have their, like IonQ and Rigetti and D-Wave and Oxford, they have their own machine. And so they’ve used this to show a new kind of approach to lithium ion battery chemistry. And so these kinds of areas in chemistry and physics where there’s multivariate analysis, there’s complex quadratic equation solvers that have never have been really done. Well, a traditional computing could be approached. And so I think in areas such as physics and chemistry, these kinds of breakthroughs are conceivable. I think we’re seeing this, but it’s very controlled. Because it’s not someone going AWS Bracket, as far as I can tell, is someone going to the vendor like PsiQuantum and saying, “let’s work with you on this problem.” And it’s a laboratory that has been working on this problem for a long time. It’s sort of analogous to the way DeepMind with deep learning AI approach to protein folding. Right in the test, they’ve been running this test for decades, to predict protein folding. And DeepMind got involved with that very controlled problem. Space is a very controlled set of participants. It’s a very controlled technology controlled by Deep Mind in their case, and every entrant had their AI model.

So that AI model of solving a well defined problem that’s been worked on where there’s been data for decades is, to me, what is the kind of thing you’ll see in quantum. Where we have a very well defined problem, it’s under the control both the experimenter and the hardware software provider, you know, the vendor of quantum. And then there’s a whole bunch of data that can be crunched to make it work. So highly controlled, not you and I, kind of just going to AWS Bracket or your GCP and running, you know, we just came up with today would be cool.

Simon Erickson  06:33

It’s really kind of fascinating. Because, you’ve seen the development of cloud computing, where people have different offerings for cloud. There’s different license agreements and things like that you can do. But it’s still storage, computing, and applications, right? Machine learning, you’ve got different algorithms you can use for machine learning. But it’s still algorithmic. It is software.

Quantum computing is a different beast because it seems like there are so many different scientific approaches to taking this. Like you had just mentioned, PsiQuantum, they wanted to use photonics for doing quantum computing. You also mentioned IonQ. Honeywell is another company that’s also similarly wanting to use ion trapping. And this is very different from like what Google was wanting to do or IBM (NYSE: IBM) was actually trying to do. Supercooling down to almost absolute zero. From what I’ve seen here, and in the conferences that I’ve attended, it seems like we’re almost shotgunning a whole bunch of different ideas out there. And everybody says, “Yep, I’m gonna have a quantum computer that’s ready in five years.” Five years goes by and they say, “You know what? Actually, it’s going to be a five more years from here.”

Are you starting to see that there’s one or a couple of approaches that are really gaining traction, from a science or R&D perspective? Or is it still kind of just, we’re throwing a whole bunch of stuff up against the wall and seeing what’s gonna stick?

Tiernan Ray  07:43

It’s the latter Simon. And in fact, I have a metaphor for that. An analogy for that. There’s one of my favorite books in the entire world is this book that was last published, it’s out of print now, but it’s a masterpiece, 1982. Revolution in Miniature. Oh it’s green so you can’t see that.

Simon Erickson  07:57

[Laughs due to the Zoom background blurring out the book’s cover.] It looks really funny with the background of the Greek islands behind you.

Tiernan Ray  08:01

It’s called Revolution in Miniature. And what this small section down here is showing is if you turn it that way, this is the cover of The New York Times from July 1 1948. July 1 1948, is when Bell Labs announced we’ve invented a thing called the transistor. And the transistor was the beginning of the solid state era. So if you want to look beyond the 1940s vacuum tube computers like ENIAC to the modern age of digital computing, the idea of digital computing was there with vacuum tubes, but there was a thing that was necessary to be mass produced to make it viable. So along came this transistor from Bell Labs, and it led to all these firms such as Shockley semiconductor, which produced Fairchild which produced Intel (Nasdaq: INTC).

But what happened in the 1950s after a decade after the New York Times announced Bell Labs has announced transistor. So all these companies arose to sell individual transistors. There was no there was not yet a thing called the integrated circuit, where there are multiple transistors fabricated on a single silicon die from a wafer. There were individual transistors cut from a crystal of silicon. So all of these companies got into it.

And so on page 61 of this beautiful book by Ernst Braun and Stuart MacDonald, Revolution in Miniature, one of the greatest charts of all time for any industry is table six. The percentage of total US semiconductors market. And it’s a list of companies which is headed by General Electric (NYSE: GE). Okay, General Electric sold the vast majority at some point. Raytheon was at some point in 1955 or so the top transistor maker. Newcomers were interested. Texas Instruments (Nasdaq: TXN) is a company which is still important in chips. Transitron, a company I’d never heard of, had 12% of the market. Howard Hughes (NYSE: HHC) which we know from aerospace was selling 11%.

So all these companies, a dozen companies, most of them today are — aside from Texas Instruments — are not important in chips. And why is that? Because they were in a more market for individual discrete transistors that were being sold for things like hearing aids and radios. Before someone had figured out how to create an integrated circuit, which solved the problem of wiring together transistors.

So to me, what that tells me, is we’re at 1957. We’re in the “transistor age of quantum”. Where Oxford Quantum Circuits and IonQ and PsiQuantum, and D-Wave, and Rigetti all have neat things. And Honeywell and Google all have neat things to create these discrete things, and they can wire eight of them together or 64. But they’re wiring them together. And so this substrate that revolutionized digital electronics, the integrated circuit, beginning in the early 60s, which was invented simultaneously, by folks at Texas Instruments and Fairchild, this breakthrough of large scale integration has not happened. And from 1957, the heyday of individual transistor sales, when all these dozen companies were leading the pack to a decade later, when Intel was being formed and become the one of the most dominant integrated circuit companies in decades. It took another several years for desktop calculators to become the main use of integrated circuits in a volume market. It took another decade for the personal computer from Steve Jobs and Steve Wozniak and Microsoft applications to make a mass market use of microprocessors. Meaning, you know, a CPU that was more than just a microchip, basically. And it took another 20 years or more after the integrated circuit to provide as a substrate for Moore’s Law to become a rule of incredible progress and for there to be a volume market to use it.

I think we’re kind of like at the start of, if you like the analogy and it were just literal, we would be at the start of 30 years of finding an integrated circuit for Quantum. I’m not saying that’s the case, because history never repeats itself. But we’re at the transistor level. Where people all have a hold of this basic thing, which is now you can build them fairly stable. You can build qubits somewhat stable. But how many do you need for the circuit to be accurate? How many do you need for to do real work aside from just a single if then statement? That’s the integration work. And I don’t know who among these five startups is going to do it. Or if any of them will still be around and significant. You know, when the Intel of integrated quantum is happening.

Simon Erickson  12:34

It’s fascinating. Like you mentioned, you look back at something 70 years ago and where the technology was. And we take for granted seven decades of R&D work and breakthroughs that has gone on to it. We’re still trying to get two qubits. It has got some some neat quantum mechanics properties. You know, of superposition and entanglement and getting them resonating at the same frequencies. But still, you’re building the transistor to get them to actually work. And the next steps.

Tiernan Ray  13:01

I mean, it’s a multi dimensional thing, Simon. Because remember, you didn’t have high level languages when you had transistors. You had people who could program in machine language to make a transistor versus a basic switch. But something had to happen when the integrated circuit was invented. For people to say, “Oh, how would you use a thing where you can get access to many gates simultaneously, efficiently and economically?” You then had to have high level languages like C, which were their own. The software was its own evolutionary path. That happened in conjunction, but also over decades, to enable humans to do things other than having to program you know, a billion transistors, one zero at a time.

Simon Erickson  13:43

It certainly is interesting to see the different approaches of it. I’ve enjoyed watching kind of the development of the quantum computers themselves. Which is so far upstream. And then you’ve got several steps downstream of that, too. You’ve got to build an interface. So you’ve got to find a problem and make sense of these things. There’s picks and shovels along the way. All the cooling equipment, the control electronics. Anything that goes into building the quantum computer.

I mean, do you do you have opinions on that Tiernan? Let’s transition a little bit. From tech and R&D to the commercial side of this now. You know, our audience is mostly individual investors who are watching or listening to this podcast. There’s a bunch of different ways that you can make money off of this. Right? You can build a quantum computer, you can support the development, or you can kind of work on the software and downstream.

You mentioned IonQ. That is a pure play company that raised some money in a SPAC last year. But then you’ve also got the big tech companies. IBM is working on this. Google’s working on this. Microsoft’s working on this. How do you think investors could approach investing in quantum computing? Is it good to go after the early ones that have more potential? Is it just way too risky right now?

Tiernan Ray  14:48

It’s way too risky, but the stocks are in freefall.

I think when I wrote about IonQ in late March, the stock was down 25%. It’s now down 41%. And what happened, Simon, was they did a secondary follow on offering where about a half of the shares outstanding — 105 million — were put up for sale in one form or another, including warrants. So a bunch of people were involved in this SPAC and the PIPE, you know, the private placement, just exited. And so what happened was, I mean, you’re gonna see a lot of this. March 22, I wrote about these companies. March 28, IonQ announced, “Hey, we’re gonna have five times as much revenue this year. Our revenue is not going to be 2 million, it’s gonna be 10 million in 2022.” Great, the stock was like, “Okay.” 5% up the next day. A week and a half later, they issued the prospectus for this sale, it was not totally unknown in advance. But the prospectus comes out for the secondary share sale and the stock was down almost 13% on April 11.

I think so you’re dealing with companies that can be very volatile. The trading can be very volatile. So what I would say is for the ones that are public, like Rigetti, and IonQ, both result of SPACs, you can watch them, but it may be a long time before they show anything meaningful. If suddenly, on March 28, the Street estimates for IonQ went from 6 million revenue this year to 10 million in revenue. Wow. Okay, so you’re talking about like a 75% increase in estimates. But we’re still talking about $10 million for one year. And they’re still gonna lose 10s of millions of dollars on that because they’re burning through money like crazy. As you said, they had a huge public float via the SPAC. And so they have about almost 550 million in current assets on the balance sheet, almost 600 million. And that’s multiples of their current liabilities. So they’ve got huge room for working capital to keep going. But whether everyone’s getting excited about 10 million or 20 million of revenue this year in a market that’s really rough, I don’t know.

So I would say you watch these companies that are public. You watch for when they have new competition from privates, like PsiQuantum coming to market, if anything comes to market in this environment. And in the meantime, I think you look at either does it move the needle for Amazon, Microsoft, Google, IBM, or Honeywell? Or you look at the third choice, maybe the best choice in my opinion, in stocks that are enabling technologies. Form Factor, which is a very well established company, an excellent company in chip tooling. So making conventional semiconductors. And making almost a billion dollars a year in revenue, that’s a good, solid company that’s profitable.

There’s another one that looked interesting to me, which is Air Liquide in France. Which is a company that develops basically, isotopes to create those sub zero temperatures you were talking about at the top of this discussion, that are necessary for Google and others to have quantum hardware. So there’s some of these enabling companies, that I think are really interesting.

So you’ve got three plays, basically. You’ve got the giants, the cloud giants, who are currently selling per-use cycles on hardware and are going to benefit in some way no matter who wins in hardware. Just a question, does it move the needle for their stocks?

You’ve got these startups that, who knows if they’ll generate 10 or 20 million in a year. Who knows what that equals? And negative EBITDA? And it’s a horse race.

And then you’ve got the suppliers, who are very well established companies with established businesses who are not going away. That, to me, those are the most interesting edge where you have get an information arbitrage advantages. Does it play out for them?

Simon Erickson  18:47

Yeah, great points here. And you mentioned your piece in late March. The name of that one in The Technology Letter is “Quantum Stocks: Here, There, and Nowhere All at Once” mentioned a lot of the companies that Tiernan just mentioned. I also would like to point out that that was published on March 22, which happens to coincide with my own 40th birthday. And about a topic that I love to talk about. I think it was a sign that it’s time to reach out to you Tiernan and chat about quantum again.

IonQ, you know, you mentioned that’s a pure play. The fact that it came public at a valuation of $2 billion off of $2 million of revenue. So that’s 1,000 times sales. So there’s certainly a lot of expectations that are already baked into.

Tiernan Ray  19:27

This year, it’s going to be 10 million. And so that’s only 100 times.

Simon Erickson  19:30

Thank goodness, thank goodness  [laughs]. We’re down from nosebleed, down to just the stratosphere.

But certainly, like you mentioned, the big tech companies have the resources. But how much of an impact is this going to make even five years out? It hard to tell any of those.

I wanted to chat a little bit about supply chain with quantum. And this is something that, to remind everyone who’s listening might not be as familiar with quantum computers, it is a completely different architecture than we built over the last seven decades. You know, we’re not doing the zeroes and ones. We’re not doing everything binary with the same logic gates and the CPUs and the GPUs that we built upon with semiconductor chemistry. Quantum mechanics is very, very different. A completely different field of chemistry altogether. And you’re kind of building processors on top of that.

The reason I bring this up is there is a lot of geopolitical instability in the world right now. And fabs are not cheap to build. A new semiconductor fab can cost you $20 billion or more, depending on where you’re building it. I chatted with Google for a little bit at South by Southwest this last month — or I guess, a month and a half ago now. And they’re not only wanting to design the quantum computer, but actually they want to eventually do the manufacturing of the components for it as well.

Tiernan, that blows my mind. When you’re thinking about lithography and etching and all the other stuff that goes into a fab. Do you have any thoughts about supply chain and this supply crunch we’re in right now for semiconductors? And would we face those same challenges if quantum gains a little bit more of market adoption?

Tiernan Ray  21:04

Sure. I mean, obviously, Google doing it changes everything because they have resources. But the problem is akin to — it’s very good question regarding the supply chain. The problem is akin to what ASML (Nasdaq: ASML) was up against. ASML is a Netherlands company that sells the most complicated lithography equipment for imaging circuits. At the bleeding edge onto the silicon wafer. And this month, this week actually on Wednesday, they announced results that were good. But results that were still contending with a shortage of parts. So they can’t get as many parts as they need to build a machine to build the parts.

It’s this irony in the chip world beat. The interesting thing is that this is a company that sells machines in the 10s of machines per year. They’re multimillion dollar machines, they’re incredibly complex. So anyone, be it IonQ — which incidentally has similar numbers, like talking about single units of machines — anyone like IonQ or Google (Nasdaq: GOOGL) faces the challenge that you can be very ambitious. But to build a machine, that is the kind of complexity that we’re talking about, is not easy to rush. And it’s even harder to rush when you cannot get the parts that you need from the supply chain to build anything.

I think that this is a boutique business for a long time to come.

And that’s interesting for a startup. It’s an open question, whether a company of enormous resources such as Google with, let’s say, total vertical integration, making semiconductors. And then packaging it all into a finished system. Along with, you know, the coolers, the freezing equipment, all the attendant things that go into making a machine. It’s a question whether all of those resources can actually accelerate a boutique business. Would Google be better than ASML at making the world’s most complicated lithography equipment? It is a kind of analogous question. It’s not clear it would be. Because you could have the most resources in the world and you’re still up against not being able to get things shipped from one country to another. You are still up against a machine like this needing a heavy qualification cycle. You’re going to ship it to a lab like a Department of Energy Lab and then it’s going to be tested for months at a time. This is why companies like ASML are incredibly efficient and productive and they still ship 10s of systems, not millions, every year. It’s very, very complex.

So I question whether anyone having any resources can accelerate a boutique business. That is extraordinarily complex. It’s one of the most complicated things that human beings build in the world are the machines of this type. And it’s the same for quantum as it is for this ASML lithography.

Simon Erickson  23:54

So the future for quantum’s supply chain seems to look like one of two paths.

One is maybe the Taiwan Semiconductor (Nasdaq: TSM) path. It’s an independent foundry, where they work with designers but then they go out and they source the equipment from ASML and the silicon and everything else they need to produce these chips independently. Or it’s something that looks more like an Intel. Where you’re doing the design and you’re using it internally for your own production. But you might farm out a little bit of production for others too.

Either way, it’s going to be very complex because the components are still so early.

Tiernan Ray  24:21

Right? And the question, people are now asking you if that, in fact, is okay. ASML, you build a really complex product. Maybe you’d have an easier time if you owned more of the parts that go into it. Meaning you’re grabbing off the shelf parts to build these complex machines. That’s a liability now in a supply chain situation. So yeah, you could say maybe the model was a vertically integrated model where Google is like Intel in producing everything for the system. But even if you did that, it’s not entirely clear to me that you could turn up the crank on incredibly complex machines. That take me multiple stages to assemble and to test and to verify.

Simon Erickson  25:05

Two more questions for you Tiernan. My first one is, I suppose we are both the financial media. But you know, I’ve seen a lot of articles in the financial media kind of focused on quantum advantage, quantum supremacy. You know, what’s the number of qubits that’s going into a quantum computer. It seems like we’re kind of obsessively focusing right now on these interim milestones that make the news. But maybe they’re not the thing that really matters.

What is it that you are interested in about quantum computing? What are the metrics or the things we should be paying a little bit more attention to?

Tiernan Ray  25:40

I think that, again, it is the transistor age for quantum. And these incremental steps, as you point out, are a way of wishing we were at the integrated circuit stage. And it’s not happening. And to my mind, it won’t happen by experiments that show additional qubits is going to happen as some kind of manufacturing breakthrough that makes it feasible. To have an integrated circuit analog to for quantum, something that’s highly integrated, and something that makes it a no brainer to just have as many qubits as you need. And we’re not there. And people keep trying to feel like they’re there by grabbing on to the next breakthrough qubits. And so that’s on the hardware side, something that no one has a clear route to, I think. Including Google.

That is we’d be like the science people did about integrated circuits before they made them where they said. Is there a way we could stop connecting these things individually? There must be some way to mass produce them in one or two steps rather than assembling them by hand. So we’re at that stage in hardware.

And I think as long as we’re at that stage in hardware, I think what you might be alluding to Simon, is people get impatient about the applications. And they want to do stuff like optimization, like better lithium ion battery technology, via approaches that are quantum approaches. And it’s impossible to do quantum computing style approaches without a quantum hardware machine. And the way you would do it are various approaches that involve analog semiconductor chips that we already know how to make well. Via approaches that use simulations. They borrow notions that Google has developed with its Sycamore computer, but they do it on a regular computer. And it doesn’t have the speed up that you get with quantum advantage. But it may open a new path to investigating some of these problems.

And so what I wonder is do the hardware guys take long enough to reach an integrated circuit stage that the software people and the scientists say, “We’re just going to do something else. We’re not going to wait for you. We’re gonna go to AWS Bracket and use a conventional computer to run simulated quantum jobs, so we get the benefit of thinking about the problem and optimization problem and multivariate problem in that way, maybe with deep learning AI. But we’re not going to wait for a [quantum] computer to be ready.”

Simon Erickson  28:03

It’s going to be really interesting to see how this plays out. You know, just seeing how the computing industry of today has shaken out. From CPUs. And then we went to GPUs, because we figured that rendering for videos was much more efficient doing in parallel. Now you’ve got ASICs that are highly optimized and application specific. And you’re talking about things like FPGAs. Like Xilinx’s AMD acquisition recently. Just super optimized building application specific. We’re at the earliest stage of quantum and we’re just trying to design the thing that can do problems that can never be possible by computing today. And that’s going to optimize over time too, right?

Tiernan Ray  28:36

It is in HPC, the field of high performance computing, in which Nvidia is now the most dominant chip supplier. Arguably along with Intel and AMD, but really NVIDIA is in the forefront. Nvidia is pushing on HPC as a technology for 20 years. Jensen Hwang was pushing on HPC in oil exploration, oil and gas. It deep learning AI came out of nowhere starting in 2006 and accelerated in 2012, to tell the world what chips were really good for. And now it’s become a runaway freight train. For Nvidia, it’s the biggest part of their business is deep learning. But it was this movement of HPC, and that movement still has legs. High performance computing continues to increase in volume of computer systems. It continues to drive Nvidia’s results and AMD’s results and Intel’s results. Nvidia is driving the semiconductor processing stages. So that whole revolution in HPC becoming accessible to researchers. The democratization of HPC is a huge growth stage at the moment, with many investing implications. Chief among them being Nvidia; that’s kind of more real than quantum in a sense.

So you could if you want to talk about big problems on big hardware at a moment in time, HPC is it right now. You know, an interest in quantum. But it’s it’s kind of like you don’t have to go too far into the future. Just look at HPC. I mean, it’s amazing what’s happening.

Simon Erickson  30:05

It’d be really interesting to when you’re talking about what problems would have with millions of inputs that would need a quantum computer. I mean, two that come to my mind, climate and weather is one big one. And space economy. You know, those are multiple billion if not trillion dollar industries that were just in kind of that first pitch of the first inning for.

Tiernan Ray  30:22

Right, yeah. But also portfolio theory. If you go back to Bob Merton in the early 70s, sort of crystallizing, how should people think about annuities? How should they think about utility functions, right? And cash in and cash out? Designing a portfolio for wealth management was a problem with a lot of state spaces. And you know, Merton said, you can’t consider all the state spaces. So you have whole areas that you just label this uncertainty. And then it’s about what is your client’s risk tolerance. So all of our discussion of quantum is marked with uncertainty. But areas of it, with a large and ability to handle large state spaces such as quantum, could conceivably say, “No, it’s not a blob, that’s a labeled uncertainty. We can be more specific and we can isolate the risk here. And we can we can dig into it.” The same way that, you know, deep learning digs into Texas Hold’em poker and says, we can be more precise about what is the likelihood of such a hand in a given round.

That’s what these large computer models can do with larger state space. Is stopping fuzzy humans and saying, “Oh, we’re going to be more precise about what do we think is uncertainty there.” That’s a great one is portfolio theory. Wealth Management is an area where the way humans have done it with conventional computing is, there’s a lot of uncertainty.

Simon Erickson  31:42

It’s not good enough, right? When when you want to unlock the the disruptive potential of something like a quantum computer. It’s interesting to see this, Tiernan, as it plays out.

I do want to shift gears just for one last question while I’ve got you here. Knowing what a fan I’ve been of your work for such a long time, and knowing that you’ve kind of been at the forefront of technology for a long, long time now. Can you tell me maybe two other things outside of quantum computing that you’re really excited about right now? And bonus points if you could make silicon carbide one of those — because I’ve read some interesting articles about it from you recently.

Tiernan Ray  32:13

Thank you for reading the silicon carbide piece. A long piece Simon. Silicon carbide is an amazing technology that is certainly an exotic area of semiconductors. But it is conventional semiconductors. And it’s getting a lot of traction now, specifically in traction inverters.

The traction inverter is the machine that converts the battery in a Tesla (Nasdaq: TSLA) vehicle into the power to drive the motor that spins the axles of a Tesla Model 3, S, X, or Y. So Tesla was the leader here. But there’s tons of companies, every car company in the world is now moving to silicon carbide as this wonder chip that’s powering these more efficient traction inverters. And basically, they can get more power out of the same battery, same voltage, same kilowatt hours. And so this has become a wonder material for cars. Next week, in the town of Marcy in upstate New York, Wolfspeed (NYSE: WOLF), which is the leader in producing the technology of silicon carbide (it used to be called CREE, they sold off their light emitting diode business to a company called Smart Global so now all they’re going to do is make wafers and they’re making larger wafers that are going to build from 6 inch to 12 inch diameter silicon carbide wafers, this is really hard to do incredibly hard to do. The company’s spending 110% of revenue on R&D) — so next week in Marcy, they’ve unveiled their new eight inch wafer factory. And they have a ribbon cutting and they’re going to have the CEO there, Greg Lowe. And so this is a big, big event for Wolfspeed. It’s possibly one of the most interesting chip companies to invest in for years to come. Because they’ll be the only company with this scale of ability to make these wafers. And that means that they’re going to be the place to go to for most of the chip makers who supply the inverter makers who supply Daimler and BMW and Tesla and all these other companies. Elon’s going to come to them for this.

So silicon carbide is this wonder material. Wolfspeed is the pureplay. There’s other companies in there as well. On Semiconductor’s (Nasdaq: ON) a good one. Just totally fascinating area, and it’s going to ripple through all things that use energy, including renewables. Solar and wind need these style inverters. Mass transportation, the form of electrified rail needs these inverters. Factory automation, electric motors in factories need this kind of revolution in this wonder material. And a track system’s the most widely deployed kinds of industrial machinery in the world need. So potential is huge for decades, probably with silicon carbide. It’s really fascinating.

Simon Erickson  34:53

To simplify that down and for anyone, it’s the building block of a more efficient way for them than the silicon that we’ve been using for decades?

Tiernan Ray  35:01

Because our everything in the world is becoming electrified, right? The car, the EV, is the first example going from fossil fuels to electrification. Everything’s going to be like the EV. It’s going to be revolutionized by electric power. And you’re going to need something to make it run more efficiently. And silicon carbide is the chip that makes the electrification more efficient than it will be. There’s a divergence. There’s two paths. Elon Musk and Tesla showed EVs can either be hard to make competitive with gas or they can be super competitive via silicon carbide. That’s it, and everything else in the world is electrified will be the same divergence. You either build it electrified, you know, not so great. Are you building silicon carbide, it’s good.

Simon Erickson  35:44

Do you have one more trend? What’s one more thing that’s on your radar today?

Tiernan Ray  35:48

The super computing stuff that was talking about. HPC. I mean, it just got so many legs. And I think we haven’t yet seen the benefit of the kind of problems. You’re talking about multivariate optimization, these high state space models before they even get to quantum being one in these these systems that are becoming sort of democratized where we have more access to them. That’s a whole level of sophisticated computing problem, that’s going to pay many dividends.

Simon Erickson  36:16

Fantastic. Well, once again, Tiernan Ray, one of the most forefront, innovative thinkers in reporting on the technology industry. You can follow him, he’s the editor of The Technology Letter. It’s a fantastic publication and he’s written a lot of really forward thinking articles.

Tiernan, it’s always a lot of fun to have you. Thanks for being a part of the 7investing podcast.

Tiernan Ray  36:34

My pleasure. Thank you so much.

Simon Erickson  36:36

And we’ll link to the former the last conversation we had with Tiernan in this article. You can always follow all of our 7investing podcasts at 7investing.com/podcast. My name is Simon Erickson. I appreciate you tuning in and we are here to empower you to invest in your future. We are 7investing.