Why quantum has been "10 years away" for 30 years | E2316

Alright everybody, welcome back. It's This Week in Startups from Tokyo. Yes, we're at Founder University. I'm back in Tokyo. I go twice a year. I go in the summer, it's 96 degrees and 100% humidity. It's the exact opposite of what happens in January when I come here to do Founder University and ski for a week. Um, but I love coming to Japan if you're a founder or an investor, and you're in the This Week in Startups or the Launch or Syndicate family, you can join us for Angel University. You can apply for that. We do that twice a year here in Tokyo. If you're a founder, you can apply for Founder.University website. You'll see Japan, you'll see Riyadh in Saudi Arabia, and then you'll see the US edition. You can come to any of those editions. You can apply. It's a bit competitive to get in. We like to look for teams of 2 or 3 founders. For Founder University, you don't even have to be incorporated yet, but you should have a project, a product in market, maybe one customer, a pilot. And then for our Launch Accelerator, you need to have a couple of customers, and that's where we invest in companies. And finally, we have thesyndicate.com, where if you're an angel investor, you can join. And twice a month or so, we share one of the deals we're investing in it, uh, and with you. And as an angel investor, you can put as little as $5,000 or $10,000 into a deal. Well below the cap or the minimum that most founders charge. So, uh, you know, join us for any of those programs and make sure you, uh, subscribe to This Week in Startups as well on your podcast player. When I'm in Japan, my team looks for the most successful, the most inspiring founders for me to meet while I'm here, and we're very lucky today. We have Kazu Nakashogi. Nakashogi?

Yeah, absolutely.

Oh, I got it.

This Week in Startups is brought to you by Rippling. Don't settle for AI that's all talk. Head to rippling.ai/twist and get the only AI built to give you full visibility across your startup and take complex actions across your entire organization. That's R-I-P-P-L-I-N-G.ai/twist. Sign up for exclusive access today. DigitalOcean. Want to see what building on a true AI-native platform looks like? Head to do.co/twist to start building on DigitalOcean's AI-native cloud today and cut your AI workload costs by up to 50%. And PayPal. Built for payments, growth, and agentic. PayPal Open, built for all business. Visit paypalopen.com to get started.

Uh, and you are with a company, Yakumo.

Yakumo.

Yakumo, which is spelled Y-A-Q-U-M-A, and you're building quantum computers.

Hardware.

Yes, hardware. And hardware is hard.

Hard, yeah, of course.

Hardware is very hard. Quantum computing has been 5 to 10 years away for the past 30 years. Why is it taking so long?

That's a really good question. Most of the difficulty part is coming from hardware itself. Ah, yes. So we try to generate very special computing units, QPU, quantum processing units.

Quantum processing units.

Yeah. That's why you call it QPU instead of CPU.

Got it.

Right. But it's really hard to create those kind of things as a hardware. And we have to make a bunch of number of QPU to do very useful calculation.

And from my understanding of qubits, this is instead of 1 or 0, you get to have 1, 0, or neither 1 or 0.

At the same time.

At the same time.

Yes, superpositions.

So you could have each of those positions simultaneously.

Technically, yes.

Technically, yes. Yes. So explain to, founders in technology who use GPUs and CPUs all day long, why is this the white whale? Why is this the holy grail of computing? Why is this the next paradigm? Why is it so powerful?

Thank you very much. So first of all, I think there is some big misconception about quantum computer.

Okay.

Once we use quantum computers, Maybe people believe that, oh, maybe we can do more nice, very fast calculation compared to CPU or GPU. But this is, first of all, it's not correct. Yes, we can apply quantum computer for particular questions or problems. For example, combinations or like RSA crypto.

Yes.

Those kind of difficult question, we can only, we can apply those kind of. Problems to quantum computer or quantum calculations.

And we saw Google and Microsoft and some other quantum computer companies went public over the last maybe 2 or 3 years.

Exactly.

Lots of excitement.

Of course.

And we had the same wave of excitement 5 or 10 years ago, but it seems like it's getting closer and closer to having an application in the real world. When will we see a company, a startup that uses a quantum computer to solve a problem in the world, as opposed to all these companies spending billions of dollars trying to get a quantum computer to work?

Yes.

So when do they have enough power and enough software and maybe the compilers and the tools have to be built as well, I understand. To actually make it accessible to founders at Startup University? Yes, startups at Founder University.

Of course, of course. So that is really, really important question. So there's two ways to understand these circumstances or at the same time. So first of all, if you're looking at question or some kind of problems that quantum computer going to solve, in the past, for example, 5 years ago, people believing that we need 1 million qubits, QPU, to solve some certain problems. But recently that number is drastically decreased.

Oh, okay.

Yeah, close to, for example, 10,000 qubits. Got it. Just 1% of what we expected 10 years ago, for example, since we updated some softwares, algorithms, or some other schemes, for example. Besides, in terms of the hardware, looking back, for example, 10 years ago, we only have maybe 1 to 10 qubits.

Wow.

Well, it's super tiny.

That's it.

Super tiny. But recently—

It's less than like a punch card.

Exactly. From an old computer. Exactly. But recently, for example, one of the United States universities demonstrated 6,000 physical qubits.

Wow.

It's amazing. So what I mean is like the number we need for doing actual calculation is decreasing. The number of the physical qubits is close to that numbers. So we believe that we reach to like two of them meet each other close to 2030, for example, in next 4 to 5 years.

Got it. So we'll be sitting here in 5 years and there'll be companies who have quantum computers in their data center using them?

Of course, we can imagine that. And some of the quantum computers are already installed into data centers already. Yes. For example, in Japan, Toyosu-Obare.

Yeah, there's one there.

Yeah, of course.

Got it. And it's being used for what? Just doing simple math calculations now and just trying to make it work? Because my other understanding is it's not consistent yet. It has— it's maybe fragile or brittle. Yes. So why is it fragile or brittle?

Yes, thank you very much for question. So once we do quantum calculations, they made a lot of error. For example, every 1,000 times, every 100 times they make error. Oh, oh yeah, it's a problem. It's problem, right?

Yeah.

And this is big problem. So we have to correct this error. So-called, we call it quantum error corrections.

Quantum error corrections.

Exactly. So we call it QEC. And by doing so, we have to protect the information that qubit has. Ah, so by doing so, we can create very scalable, very robust quantum computer. So this is recent approach the human being try to take.

Got it.

Did you ever notice how you spend hours shopping online, then pause before you actually hit buy now? That tiny hesitation where you wonder if this is trustworthy can make or break a sale. And with AI changing how we discover and compare things, that split-second trust question matters more than ever. That's exactly where PayPal comes in. For more than 25 years, PayPal has been a leader in online payments, and now they're at the forefront of agentic commerce, making it work for businesses while letting them maintain control of their brand and their customer relationship. So even as the way we shop changes, the moment that matters most feels familiar and deeply dependable. Built for payments, growth, and agentic. PayPal Open, built for all business. Visit paypalopen.com to get started. And we had these in the early days of computers, the hard drive or a floppy disk, the memory tape, or the memory in the computer would go bad, or it would, you know, have to be rebooted and you have to clear all the memory out and start over. Yes. And you maybe sometimes I'll have somebody who's having a problem with their laptop and they say, when's the last time you rebooted? And they're like, I never— I've never rebooted my computer. I've had it for a year.

Yeah.

And I say, oh, I'm from the 1980s. In the 1980s, you would reboot your computer every 2 hours to make sure you didn't have memory leaks or problems and the hard drive was working okay because you wanted the software to— the operating system to be fresh and it would run faster. And then by the end of the day, your computer would be sluggish, memory leaks, and you have to unload one program to run the other. It reminds me of what you're going through right now.

Yeah, that's true. That's definitely right. Probably what we expected or experienced like 60 or 50 years ago for classical computer.

Yes.

We experience in the quantum computer at this moment.

Now you're doing something different at Yakuma.

Yes.

With the chemicals or the material science. Yeah.

So you're talking about application or hardware?

No, the hardware itself is slightly different.

Yes.

Or it's advanced and they're using different materials now, I understand.

Yeah, thank you very much for asking that. So I think there's a couple of methodologies to do quantum calculations. For example, if you're looking back the history of classical computers, let's say if you're using this kind of CPU, this is like most of the CPUs created by Silicon, right? But maybe 60 or 50 years ago, I don't know. We try to use vacuum chamber.

Yeah.

To do calculation, isn't it? For classical computations. Same. So at this moment, quantum computer market try to test which kind of QPU is the best for quantum calculations.

The quantum processing unit.

QPU, exactly. So for example, Google or IBM, this kind of large, amazing company try to create QPU by using superconducting qubit.

Superconducting cubes.

Yeah, that's why they have to put those kind of chips into huge refrigerator to make it super, super chill.

It has to be super chill.

Absolutely.

Yeah.

And some people try to use ions which have a charge like plus.

Ions, yeah, like plasma.

Yeah, yeah. They have charge and we use atoms, neutral atoms. We don't have any charge. We try to configure it or try to create array of the atoms in the vacuum chamber to use one atom, an atom as a qubit. Got it. Yeah. And some people try to use photons or semiconductors to do quantum verification at the same time. But there's so many very severe, like, you know, competitive landscape happening in this moment.

So let's talk about, this is your first startup, company building.

Yes.

We have some hardware startups here. Hardware is hard.

Congratulations.

You've picked the hardest of the hardware because you have a you know, 5-year window to commercialization maybe, and you came out of a university.

Yes.

And your company is based in Kyoto.

Yes.

So you get to have a beautiful life, a beautiful city. Thanks. And all that talent from Kyoto University, yeah?

Oh, some of them, yes.

Yeah, so you came out of the university. Explain to us how technology in a university how they do the IP transfer into a commercial organization, not just at Kyoto, not just your company, but just generally how that happens. Because some venture capitalists— and Stanford does this— and some venture capitalists specialize, or investors specialize, in going to universities and saying, what are the students working on? What are the professors working on? Is there anything we can spin out? I think they will generally say a spin-out.

Yeah, yeah. Of course.

Or is there a term for taking IP out of the university, like a technical term or a term of art? Do you know one?

Probably I'm not super familiar with the situation in United States. Got it. But compared to overseas, Japan or Japanese university have not so much experiences to use IP as a startup, as a spin-off. Got it.

Yeah, so it's pretty new here.

I would say so. Normally we call it deep tech startup.

Deep tech startup, sure.

Yeah, in Japan. Which is using the IP coming out from university. Right. And at this moment, there's not very solid methodology how to use IP from university to startup. Some people choose stock options, pay for university.

So the university can get some equity.

Exactly.

Or in the United States, sometimes we hear about a royalty for some number of years.— and yeah, this is— or maybe even a cash payment. I guess people could pay to take it out. But the university wants— the university's motivation is to recoup some money that they invested in the research. So you were there for some number of years. They spent some money on you and maybe gave you some resources. And then also maybe they could make a profit, some kind of profit that then helps the university invest in the next set of—

Reinvestment, yes.

Yeah, so it becomes evergreen.

Yeah, I hope so.

Yeah. And this is a negotiation that occurs between the university and the founder yourself?

It's actually, yeah, tough negotiation, to be honest.

Oh, is it tough negotiation?

Sometimes it's not easy. Yeah, yeah, yeah.

So they want to make sure they get a good deal and you have to get a good deal or you can just leave and not take the IP.

Of course.

Yeah. So it's— let's talk a little bit about how you've raised over $10 million, I think, for the company. You got some grants, you have some venture capital, you've done a seed round. How do you think about runway? Most companies here, if they're building software or maybe consumer hardware, they can think about 12 months of runway, 18 months of runway, maybe getting to their first dollar of revenue in 6 months, 3 months, 9 months, depending on the company type. How do you think about runway and when you have to make money and how you keep pushing the team to get to revenue even though it might take you years to commercialize?

Thank you very much. So that's pretty subject to the type of the startup, of course, hardware startup, software startup, but for our case, we expected, for example, 18 months long way after we just completed our seed extension round.

Oh, congratulations.

Last month. We have been invested by All Night Ventures from United States.

Which one?

All Night Ventures. Oh great, yeah, yeah, yeah. And that was, to be honest, that was their first investment to Japan.

Oh wow, how did you meet an American VC and convince them to invest in a company in Japan? Tell us the story.

Thank you very much. So we split the round into two pieces, C round and C extension round, So Seedland, this company is spin-off from Kyoto University. So that's why we have been invested by Kyoto iCAP. It's sort of CVC coming from Kyoto University. And they have a bunch of network in like EU or United States. So they nicely connected me to All Night Ventures. And yeah, and there is—

Oh, so they, there was a group that was able to put you in touch with them and that venture firm had shown interest already?

Yes.

Or they cold called them?

Oh yeah, they have like soft connection. Got it. Through the OnVentures. Ah. And they kindly connected to me to the OnVentures.

Are you a Japanese company or do you file as like an American company or both for the venture capitalists in America to invest? Usually they have to be a Delaware company or something.

Yeah, yeah, yeah. So we based in Japan.

Based in Japan?

Based in Japan. There's no launching overseas.

So you don't have a corporate entity in the United States?

Yes.

Or you do?

No.

No. Okay. So you have to be here. So was that, that wasn't a blocker for the venture firm? They figured out how to do it?

Oh, I didn't think so. It cannot be a super, super big deal. Oh, got it. Yeah, I guess so. And we also have our investment from the VC named Qornation. As you can see from the name of this VC, this is a top VC for quantum computer hardware from all over the world. So they are coming from France. That was their first investment to Japan at the same time. Got it.

And did the venture capitalists come visit you for the board meetings and they get to come to Japan a couple times a year?

No, no, no, no, it's like we talk online basis.

Oh, okay.

Online basis. But I've been—

See, that's the opposite of me. When JETRO and Japan said we want to do Founder University with you, I said, oh, I get to come to Japan twice a year? I'll do it. That's why I'm doing it also, to meet great founders. But I was coming anyway for skiing in January. So every time we see a revolution in how software is built and used, one company ends up owning the infrastructure that everyone else depends on. The next great platform is being built right now. That company is DigitalOcean, and they just launched their AI-native cloud. This is not just another place to rent GPUs or a hyperscaler overwhelming you with features and services but leaving you on your own to patch it all together. No, we're talking about a full-stack platform that comes pre-assembled and that sends you just one bill at the end of the month. Workato runs a trillion automated workloads on DigitalOcean with 67% lower inference cost, 79% lower latency, and it's 2 times faster to production. If you want to understand what building on a true AI-native platform looks like, go to do.co/twist. That's do.co/twist. Start building on the DigitalOcean AI-native cloud today and cut your AI workload cost by up to 50%. That's do.co/twist.

You went to Niseko, for example?

I go to Niseko.

Oh, that's beautiful.

I like to go cat skiing.

Okay.

You know the tractor that goes up the mountain?

Oh yeah.

So there's an abandoned ski resort. I won't say the name anymore 'cause when I say the name on the podcast, they sell out all the seats.

Okay, got it.

And then people get disappointed. There's an abandoned ski resort, maybe 40. There's many. There used to be 1,000 ski resorts. Now there's maybe 300 or 400 active ones. Wow. Because there was many more children and a bigger population, and so it's come down a little bit. So a lot of those small resorts— and people move to the city, so in those local resorts, some of them shut down. Oh, so an American fell in love with this resort outside of Niseko. It's a good story. And he asked them every year, can I have this resort?

Why don't you make businesses there?

I might need to. And he said it took him 4 years. Every year he went to the city council, to the elders, and said, hey, I would like to do this. They said, oh, tell us your plan. They said, okay, yeah, maybe next season. And then he came back, I'd like to do this, I'd like to And when I went to see the ski resort, half of the lifts were— oh yeah, uh, broken down. The cables were down, the ski lifts were down, and he cleaned it up and he left one run perfect. And all the local people get to have a ski ticket for maybe $50 for the season if they live there, and they ski that one run, and then they make very good katsu curry and nice ramen. And then they have two cats with 8 seats in each, so 16 people come from around the world.

Oh.

And they drive you up the other 8 runs. Okay. So one run for the locals, the other runs are broken down. They drive you up, it takes 15 minutes to go up. Okay. And then it takes 10 minutes to go down.

Okay.

And you do 8 runs in one day.

Oh, wow.

Wow. $1,000 a day. That's $100 a run, maybe.

Okay, okay.

But it's worth it because every run is fresh powder. J-Pow. Japanese powder. J-Pow.

Is it so? Oh, no way.

Beautiful.

Oh, right, right.

Do you ski or snowboard or neither?

I know I was born in Bangkok. I was raised up in Malaysia, so maybe I'm not super, super familiar with those kind of—

Yeah. Anyway, it's a dream. And then unfortunately, years ago I talked about it on the podcast. And then other people talked about it on social media. And just like Tokyo, all the great restaurants have too many Americans coming to take pictures. And if you really like the ramen at this place, somebody puts it on TikTok and then no locals get to go anymore. So now some of the locals are not putting them— they don't put an English name on the restaurant.

Just for—

and they hide the address more. So only locals.

Okay.

Okay.

And some of them have signs now, only local. But in Japan, in Niseko, there's now at Niseko, the actual resort is too many.

I see.

Too crowded.

Oh yeah, got it.

So I go to the other ones.

Oh, you know Japan more than me, I guess. Yeah.

So talk about this hiring of talent here and the work expectation and then building a company and building talent and culture in Kyoto where it's very beautiful. Thanks. But maybe do people work hard and do they want to be in the office for 10, 12 hours a day grinding? How do you build a culture here in Japan and even in, you know, Kyoto?

Yeah, so that is a really important question as a startup. At this moment, the size of the company is close to 70.

70.

Oh, uh, we just launched this company, um, 15 months ago. Whoa. So 2025, April. And most of them are tech side, I mean engineers, scientists, or we also hire mathematicians who's like, you know, very capable.

So 1 to 2 hires per week?

Oh yeah, sort of. Yeah, yeah, yeah. That's the biggest job for me, to be honest, hiring the people.

Got it. And how many of the 70 are Japanese or recruited to come to Japan?

Yeah, so I think there's close to 50, 5-0, tech people, and I guess 20 to 30% of them is non-Japanese.

Non-Japanese.

Non-Japanese.

Oh.

An interesting point is that there's 3 divisions under the CTO, first management, of course tech side, 2 of them, 2 heads of 3 divisions are non-Japanese. Got it. Yeah, so one of them is coming from Sydney.

From Sydney?

Yeah.

Oh, beautiful.

Yeah, he's very talented. I really want to work with him. I've been to Sydney. I need good excuses to go to that beautiful city.

That's great. Yeah, yeah, Sydney is one of my favorite cities.

Yeah, so beautiful. And the other, another guy is coming from Turkey.

In Turkey?

Turkey, yeah.

So it's a very international organization.

I would say.

What's the selling pitch to say leave Turkey, leave Sydney, beautiful city, and come live in Kyoto, another beautiful city, but this is a culture change for them.

Yeah, yeah, we do scissors, paper, rock, paper, scissors. Yes, no, no, anyways, yeah, no, it's just joking. But, um, I think, um, Kyoto have a very great advantages, um, in terms of the beauty of the city. Yeah, very chill. And we had a spin-off from Kyoto University and Kyoto University have a very profound research background, the history. We, we use neutral atoms.

Yeah.

But one of the professors coming from Kyoto University, his name is Yoshiro Takashi. He studied neutral atoms, for example, yttrium Yb. That's one of the species in neutral atoms, by the way. He studies that species more than 30 years, and he discovered, um, I 60 to 70% of the fundamental scientific study of the turbine.

Incredible.

So yeah, so that professor is very famous. And at the same time, an interesting point, a selling point about this company is that we try to integrate two entities' research output. One is Kyoto University, one is coming from IMS. IMS stands for Institutional Nuclear Science in the Aichi Prefecture. Whose name is Kenji Omori, he's the head of that institute and two professors, two top, try to create independent startup, similar startup. But if you're thinking about the competitive landscape over the world, oh, why don't we make one startup?

These two extraordinary professors, researchers, scientists.

Yes, exactly.

So now you've incorporated them into the startup, they're advisors, board members?

Advisors, advisors, yeah. So This is one of the great history behind this company. And many people know that these two professors co-work together in the past.

So they're rock stars and it draws talent.

Yeah, yeah, yeah. Very talented two professors. So we closely work with two entities, Kyoto University and IMS. And I would say this company is spin-off from Kyoto University and IMS. Yeah, so this is another good selling point since from day zero we have a large scale of research entities as a startup. That is not a good point.

And so is your intent to build the entire computer or be part of working with other companies building computers and empower them as a provider to them?

Very, very good question. Very, very good question. So I believe that our neutral atom quantum computer is something like car, automotive car. And we try to become the quantum computer company, something like Toyota. So if you're thinking about Toyota, they don't create tire, no, they don't create engines, they design everything, integrate everything to one and sell that car.

Got it.

Into the market. See, so we are the quantum computer company try to assemble the lasers, vacuum chambers, meters, whatever.

So there are many people building these components now?

Exactly, co-work together.

Where are they all based and are they also centered around universities?

You're always asking a very, very good question. So we use neutral atoms and most important device for this quantum computer is laser. Very hyper eraser, very less noise lasers.

So high fidelity.

Exactly. I like that word, Lee, much high fidelity lasers. And this is coming from Hamamatsu Photonics in Hamamatsu, Japanese company. And they also have an amazing subsidiary company in Copenhagen in Denmark.

In Denmark?

In Denmark.

Yeah.

Of all places. Yeah.

But it's like, no wonder. Since Niels Bohr, one of the like an amazing scientists from Denmark, and Denmark is very, very strong capability in terms of the physics. And that company name is Innovative Photonics. They creating amazing high-power lasers. And recently Yokumo, our company, and Hamamatsu Photonics and Innovative Photonics did nice MOU in between Denmark in Japan try to create special devices for new quantum basis quantum computer. I mean, this kind of supply chain is very crucial for industrialization of quantum.

Most AI tools can give you business insights, but they can't actually do anything about it. That's the difference between AI that talks and AI that acts. But Rippling AI is built differently. It's the only AI built on your live global workforce data. So HR, IT, and finance are connected from day one. That means it's not just surfacing insights, but you can actually take actions across your entire organization. Let's say you want to focus on talent retention. Just ask Rippling AI, who are my top performers this year? You'll instantly get a workforce report, comp ratios, performance reviews, engagement metrics, all the data you need. Then Rippling AI will recommend a retention strategy, including a 10% spot bonus for your top performers. So don't settle for AI that's all talk. Head to rippling.ai/twist and get the only AI built to give you full visibility across your startup and take complex actions across your entire organization. That's r-i-p-p-l-i-n-g.ai/twist. Sign up for exclusive access today. Which is fascinating that it's— thank you very much— so many different people around the world are working together, together to try to, to, to, uh, hit this vision.

Yeah.

Is something happened in the time you were starting this company, uh, and quantum computing has been working, uh, towards this very ambitious goal, which is large language models, artificial intelligence, GPUs have become very powerful. Yes. So how has the advances in AI impacted what you're building and how you're building it.

Yeah, thank you very much. So, so-called, we call it AI for quantum and vice versa, quantum for AI.

Explain both.

Yeah, sure. So AI for quantum, this is very, very powerful. So for example, this is very, very interesting story behind quantum calculations. But if you use quantum computer, as I mentioned, error happens and we have to estimate what kind of error happened by using classical computer. Ah, GPU, for example, right? CPU. So they help us, I mean, ask me, quantum computer, to identify or estimate what kind of error, to what extent it happened, and try to correct through classical computation. So, and through this kind of activities, we definitely use, we already use a lot of AI, LLM, transformer, for example, to do quantum error collection. So this is something very tangible example of the AI for quantum.

Got it.

Right.

And just doing that error correction.

Yeah, that's like the most like, you know, and it's so powerful now. Very, very powerful because of the AI. And opposite side. So quantum for AI. So at this moment, there's no practical or tangible example about quantum for AI since the power of quantum computer is not enough.

Not enough yet.

Not enough yet. However, recently the researchers coming from MIT wrote beautiful papers about how quantum for AI happens. We believe the future is not super, super far away.

And so would it be used for inference or model training, both? And then if it does work, let's say we go 5 years into the future, maybe even 10 years into the future, what impact could that have on either of those two practices? And obviously there'll be other jobs to be done and new modalities are being created. But using the current paradigm of the language model and the building of the language model or inference, the customer, you know, getting their query answered, What impact could quantum have on that?

Yeah.

So for example, a very, very interesting example is that after we do calculation of, by using quantum computer result coming from them. For example, if you do fluid dynamics calculation by using quantum computer, that data or result is not normally coming from the classical computer, isn't it? This is only coming from quantum computers. Why don't we use this data to make AI even smarter? We can differentiate the model by doing so.

Got it. So the models become more powerful.

I guess so.

Yeah. This is going to be very interesting.

Very interesting. I believe quantum for AI, once this kind of happens, quantum computer company makes tons of money.

Yes. And it could be like the answer to many problems. Do these quantum computers, are they as power hungry or more power hungry than their equivalent GPU or CPU type computer? What's the energy consumption like?

Very less, let's say. Very, very less. Really?

Yes.

Interesting.

Yes.

Recently, one paper is talking about, we find it out, one paper is talking about the comparison in between classical computers and quantum computers as a function of the consumption, the energy as a function of the number of the qubits. And they mentioned that the number of the qubits is close to 40 or 50 or something. Then we do a particular problem solving. The energy that quantum computer use, even less than classical computers. So in the context of less CO2, less energy, we definitely believe that quantum computer contributing to the AI.

Are they always going to be very large data center computers or will we ever see the desktop revolution equivalent of quantum computing? I know it's a silly question, maybe, but I'm just curious.

Yeah, exactly. I would say again, I cannot say that quantum computers take over all the position of the classical computers. Just one part of it. But I believe quantum computer could be the very, very great accelerator to make classical computer even smarter.

So there would be a hybrid computer with both?

Exactly.

Like a hybrid Toyota Prius?

Yes, yes.

So some jobs go to the quantum processing unit and maybe some other parts of the job go to the GPU and CPU?

They share the loads.

Wow.

And they could share the memory.

Yes.

So there's kind of primary research work happening already in Japan. For example, in Tsukuba. And there's many ways to do demonstration of hybrid system in between quantum and classical.

Fascinating. Okay. Final question. Everybody's a bit nervous about cryptography and that someday quantum starts up and all of our phones are hacked and everybody sees our bank account and you take all my Bitcoin. When will you be taking all of my Bitcoin?

Yeah, that is really, really big question. So recently, one survey review that how quantum calculation researcher think about DORSCON, D-CRICK, for a few happen. And they mentioned that in next 5 to 10 years, we have to be very, very careful about it. But if you think—

And it's true?

Yeah, I think it's true. I think it's true.

Sometimes technologists are a little bit hyperbolic. They get a little bit excited like we saw with AI. So, but it's true.

Yeah, it's true. So we, some research works already proven that if you reach to certain amount of the size of the qubit, QPU, we can solve RSA cryptography, but it won't be happening next 2 years, for example. Don't worry, it's time to sell your Bitcoin. No, plenty of time.

Yeah.

But the point is that maybe not only for Bitcoin, but for example, some like top secret coming from Japan, United States.

Yes. Military applications.

Absolutely. There's kind of dual use cases are very crucial. And for example, some country already collected top secret, which is already protected by RSA cryptography, but they already has that, isn't it?

So they have the data, they just haven't cracked it. Yeah.

And once they create quantum computer, so take now, solve later.

Oh wow. So get the hard drive now, crack it later.

Yes, yes, yes.

Put it on a shelf.

Exactly.

It was very dangerous. And so is there an international body or has the government gotten involved? And created a regulatory body that's in any way monitoring your progress? And do you get a knock on the door that says, hey, you know, we need to have a conversation?

Yeah, that is really, really important, like topics that this kind of dual-use or like, you know, quantum computer-based deep tech startup, that is really important topic. We talk with governmental people very carefully and For example, United States recently announced that they tried to launch or they want to have a very useful quantum computer by 2028. It's aggressive, pretty aggressive. And so in order to nicely communicate this comment to people, we are very happy to discuss our, or we want to like talk transparently with government people, how to use it, how to protect the technologies. Yeah, that is really important.

It's very important. The world is very dynamic.

Exactly.

It's a very dynamic world. It's constantly changing. Yeah. And as a technologist, we can't take it for granted that everybody has good intent. There are some people in the world who have bad intent. And so we have to be thoughtful. We have to get there first.

Yeah.

Yeah. Is China making progress on quantum? Do you see big progress from China?

They, they doing very, very well. So I think one of the top countries is China, United States, of course EU, Japan. I believe these two like area is pretty, pretty aggressive in terms of quantum.

Yeah.

And do they share their progress and papers? No, they don't. They just read ours.

Yeah. So they, of course, they post their scientific activity by papers, some of them, but I don't think they share everything.

Yeah, very close to the vest.

Exactly.

Yeah, it's a very interesting moment in time. Kazu, thank you so much for sharing with the audience. Big round of applause for Kazu.

Thank you.

Excellent. I don't know if there's any crypto and any quantum heads in the audience who have a question. Does anybody have a question?

Anything would be okay.

You do? Okay, come and ask a question. We have one question. You don't have to have a question about quantum computers. I'm fascinated by it. And this is like such a treat for us to see around the corner at what you're building. This is like, this conversation is literally like talking to OpenAI DeepMind 10 years ago, right? Okay, question.

Hey, that's a great presentation and a great interview. I'm Karthik. We're building Glyph Zero Labs. What we're doing is we're building the cryptographic authorization layer for AI agents. And my question is probably two parts. One is right now, Because I've done a lot of work in quantum for the last year and a half before starting my company, and current cost of quantum is really high. Like if you look at SQBM from Toshiba, it costs like $90 an hour. IBM, I think if you want to rent it, you have to pay like $3,000 an hour. So whatever you're building at Yakumo, how is it going to move the needle? Like, is it going to bring the cost of that down, or how are you thinking about it from a business model perspective? Like, do you want to sell quantum clouds? Like through a partnership with AWS or Microsoft to bring it down? Good question.

Yeah, of course.

So recently, one of my colleagues in Japan, he's not in Yakumo. Sorry, he's not in Yakumo. He's creating one of the type of quantum computer, superconducting qubit. So he mentioned that one physical qubit costs close to $1 million USD. It's really expensive. But we use the neutral atoms. And if you're thinking about the price of Qubit, maybe it's maybe 100 times cheaper.

Yeah.

Or 1,000 times affordable. So we believe this methodology is pretty scalable in terms of the financial aspects.

Yeah.

And if you're thinking about our business model as a startup, there's a couple of ways of thinking. However, we're thinking about selling our one full-stack quantum computer next 5 years. But this is not for commercial use case, maybe. It's only mainly for academia who try to use quantum computer as a research box. And after 2030, we believe that we can create very useful quantum computer, so-called FTQC for quantum computers. And we try to sell this quantum computer to data center, for example. In that case, we don't have to sell quantum computer one by one. Maybe 1 byte.

You just sell racks.

Yeah, exactly. That's bulk. And next 5 years after 2035, for example, it's time to think about how to use this quantum computer as data.

For example, you can do quantum entropy service.

You can do just like scale AI.

Quantum as a service.

Yeah, for example. Or for example, there is one great company have been like M&A by Meta, like, you know, Scale AI.

Yeah. Yeah.

They'll sell data for AI.

Yeah, that's quantum data and quantum entropy as a service.

Exactly. So we might expect those kind of things.

Got it. That's actually a part of stuff that I'm working on as well.

Okay, actually, why don't we catch up today?

The second part of the question, do I have time or? Yeah, quick. Okay. So you mentioned that RSA is going to be broken, right? But most of the world, including Visa, MasterCard, all these guys use AES-256. Of course, like hack now, decrypt later is going to happen. But wouldn't it be useful if you just blow up the key bit from 256, from if you make it AES-256 to like AES-2048 or 4096, you can buy time to let the hackers linger on the network and then use the same quantum computer to like trace them out and figure out what keys they're using to enter your system, right? Like, what are your thoughts on that?

Yeah, so that's the reason why we believe that cryptography is just one of the applications. For example, I'm very, very— Yakumo is very keen on the other applications. Application, for example, like quantum chemistry, which is going to create new market. Yeah. So that is something we are thinking about. That's interesting. Like cryptography is just one, like, you know, very, very dangerous, but that's quantum computer company. We're thinking about some other more.

Yeah. You could also end up building like virtual cells. Yeah. And cell models and revolutionize biology and everything.

Absolutely.

Makes sense. All right.

Very exciting.

Thank you.

All right. Let's thank Kazu for sharing. With us for an hour. Thank you very much. Oh my God, so great. Thank you so much for coming. I appreciate you taking the time.