Startup Series: Distributed Hydrogen with Fourier
Siva Yellamraju is the Co-founder and CEO at Fourier.
Fourier's mission is to make hydrogen universally accessible with on-site and on-demand production. Fourier is not Siva's first startup, it's his fourth. He sold his last company to Apple the one before that to Google and the one before that to Polycom, so he knows a thing or two about entrepreneurship.
Fourier is barely two years old and they raised a seed round in mid-April led by General Catalyst. They've largely operated in stealth. In fact, there's nothing on their website at fourier.earth other than their thesis statement for starting the company, so we were excited to learn from Siva about why he transitioned from a very successful entrepreneurial career in software to tackle a business in the energy sector. We were also eager to learn specifically about what Siva’'s building in distributed hydrogen and what he's learned from the process thus far.
Episode recorded on July 25, 2024 (Published on Aug 1, 2024)
In this episode, we cover:
[2:01] Siva’s entrepreneurial background
[4:46] His decision to work on climate and energy
[7:35] Customer pain point understanding
[12:44] Potential use cases for distributed hydrogen, including power backup solutions and data centers
[15:04] Approaching hydrogen production as a data problem
[21:47] Customers working with Fourier as part of a broader movement around clean energy
[24:01] Potential use cases for distributed hydrogen, including power backup solutions and data centers
[27:14] Advantages of hydrogen over batteries for long-term storage and heavy transportation
[33:20] Geographic markets for hydrogen deployment; the US, Germany, India, and Indonesia
[35:33] Fourier's success to date and plans for commercial rollout
[37:12] The goals for Fourier's pilots in the US and India
[40:16] Siva's learning curves in building an energy company
[42:29] Potential for India to become a leader in hydrogen
[44:21] The name behind Fourier
[45:34] Fourier's current capital and hiring needs
-
Cody Simms (00:00):
Today on MCJ's Startup Series, our guest is Siva Yellamraju, co-founder and CEO at Fourier. Fourier's mission is to make hydrogen universally accessible with on-site and on-demand production. Fourier is not Siva's first startup, it's his fourth. He sold his last company to Apple and the one before that to Google and the one before that to Polycom, so he knows a thing or two about entrepreneurship.
(00:34):
Fourier is barely two years old and they raised a seed round in mid-April led by General Catalyst. They've largely operated in stealth. There's nothing on their website at fourier.earth other than their thesis statement for starting the company, so I was excited to learn from Siva about why he transitioned from a very successful entrepreneurial career in software to tackle a business in the energy sector, and I was eager to learn specifically about what he's building in distributed hydrogen and what he's learned. But before we start, I'm Cody Simms.
Yin Lu (01:12):
I'm Yin Lu.
Jason Jacobs (01:13):
And I'm Jason Jacobs, and welcome to My Climate Journey.
Yin Lu (01:20):
This show is a growing body of knowledge focused on climate change and potential solutions.
Cody Simms (01:25):
In this podcast, we traverse disciplines, industries, and opinions to better understand and make sense of the formidable problem of climate change and all the ways people like you and I can help. Siva, welcome to the show.
Siva Yellamraju (01:39):
Thank you. Thanks for having me, Cody.
Cody Simms (01:41):
Siva, I was preparing for this conversation with you and you've got quite a background. You have been a successful entrepreneur over the last little over decade, I guess, 12, 13 years. Maybe let's start there. Chart your own entrepreneurial path and the companies you've built and exited to in recent times.
Siva Yellamraju (02:01):
I've been fortunate. I grew up learning computers quite young. I'm a programmer at heart. I fell in love with computers, did my undergrad in India at one of the I IITs and came here actually for grad school, although I dropped out very quickly and I started on this entrepreneurial journey. The first company we did was actually called VView, which is actually like Zoom, browser-based video conferencing. A couple of years ahead of Zoom, we sold that to a company called Polycom. In fact, that company was cool because we raised money through 2007, '08. Pretty cool story.
Cody Simms (02:33):
Tough times to raise money back then for those who weren't paying attention to the tech market back then. Those were certainly tough days.
Siva Yellamraju (02:40):
It was probably the toughest time ever to raise money, a hundred percent. We sold that to Polycom. I left Polycom fairly soon. Then another company called Baarzo, where we were actually one of the first companies to commercialize, what is now very common, deep CNNs for video and audio and image understanding with a group ImageNet from Stanford. And now as we built this company, it eventually became search at YouTube.
(03:04):
Google bought us. I used to run engineering for YouTube at Google. Video search projects like Google Photos, et cetera were part of our stack. Left Google. Started another company called Akruta, where we were doing a smart speaker with a 360-degree camera on it. You would say, "Hey, Google, call my mom." The camera itself could be moved around, but the view will always follow you. Basically, it's like a photographer you can carry with you, will automatically frame people as they are, however you're sitting.
(03:32):
That eventually got merged with Apple. That technology itself became what is now featured in iPads called Center Stage, which I used to lead. We used to work on all the new camera features at Apple like Portrait, Cinematic, et cetera. All the new features. That's my background, very classic Valley software, deep AI background, thick into the computer science. I left Apple about three years ago, took a hard right turn into what I wanted to do, which we'll talk about, but that got me into working on Fourier. But that's been mostly my background.
Cody Simms (04:02):
Well, I love hearing that story and I can't tell you how many people in the MCJ universe clearly don't have exits to Google and Apple under their belt like you do, maybe, but have built their careers in traditional tech and at some point decided to dive in and focus on climate change, the energy transition and all of that. I love hearing the stories of how folks like you decided to make that switch. So let's go there next. What prompted you to say, " Hey," clearly, Siva, you could go build another company in video AI right now, which is one of the hottest topics on the planet and do quite well for yourself if you wanted to, but you decided to go a different direction, tell me more about why and how.
Siva Yellamraju (04:46):
Whenever I go to investor pitches these days, although I'm working on something, the first question I get asked is, "Are you working on something AI? And, "Why you're not working on something AI?" So that's a whole different story by itself. I left Apple about three-and-a-half years ago, I think in the middle of the COVID waves. When I came out, to be honest, I wanted to work on something in AI and that's what I thought we would work on. But when I came out, I also was very clear in my head that eventually I want to make an impact in the climate energy space. Motivations were obviously there's a problem that we all face, which needs to be fixed, but also having young kids changes your perspective on what you want to do.
(05:22):
But it started as, initially, a 10-15% exercise of, "let me go explore and then see if there's something to be built that I can build," because the only way I know I can contribute to that is actually build something. I'm a builder. I want to build things. Although there are other ways to contribute to climate, this is the only way I think I can contribute. So it started as an early exercise. I thought, "Okay, let me go spend some time." And the more I got into it kind of took over me. There's no other way to explain it. There's no other reason. It just grew on me. Eventually we actually had a term sheet for the other idea we were working on, and it hit me that, I don't want to work on that anymore. I just want to embrace myself into energy.
(06:01):
The reason when I looked at energy, there's a lot of focus on consumer side of problems like EVs or solar panels for the home. But energy is a much broader domain. If you were to transition energy from what it is to where we want it to be, the way I wanted to do it was not necessarily an activist mindset. More or less, how do we create a better, cheaper, faster energy that just doesn't constrain people, that just doesn't ask you depending on how your political directions are or otherwise? We just want to make it better.
(06:32):
One of the things there is for me that's very important is energy abundance. How do you create abundant sources of energy that can work across different platforms, different applications. I got increasingly drawn to it, got increasingly curious about high density fuels because you need that for storing large amounts of power. Without large amounts of storage, you don't have an energy play. Got interested in hydrogen, spent time in talking to folks, which will spend a lot of time talking about hydrogen today, anyway. But I would say the short answer to your question is, organically, I got into it, but started with a clear motivation that the only thing I can do to contribute this whole problem is to build something.
Cody Simms (07:11):
When you built your first video company, maybe you were new to the space, but quickly you learned the industry, you saw where there were customer pain points, you saw where there were business pain points, and for your second and third company, I would guess you were able to attack problems from a technology standpoint where you already had lived experience and knew that there were real needs. You knew that there was demand, that there was a need for the technology you were building. When you shifted industries and sectors, I understand you took this broad theory approach to, "Hey, we need to solve energy abundance. Are there challenges we could go after?" But how did you get the customer pain point understanding to know what to actually go build?
Siva Yellamraju (07:51):
It took a lot of conversations. There's no sharp way to do it, unfortunately, or fortunately, I guess. So it started with this thesis that you would need storage. It became clear to me that hydrogen is one of the candidates. Without that, there is no story. But the more important part for me, that's what, if anything, the last three startups told me, you need to have a commercial path towards getting towards the future. I didn't want to work on something that's 15 years down the line that will work. I wanted to make something that's practical day one and go on.
(08:21):
For us, it was very important to understand the current market for hydrogen. We spent a lot of time on that, where hydrogen is used today, where it could go. And hydrogen actually has a large market today. It's about $200 billion as a commodity molecule that's made. Ironically, most of it is made is using methane. But within that, we spoke with chemical factories, independent companies all over the globe. The problem that we realize is the biggest in realizing hydrogen, not only today, but also for the future, is distribution of hydrogen itself. Hydrogen is a great molecule in terms of mass to energy ratio, but it's actually a pretty sucky molecule in terms of volume to energy ratio or weight.
(08:57):
So as a result, if your truck is carrying hydrogen from one place to another place, you're just moving an empty truck back and forth. There's maybe few hundred kilograms of hydrogen in a five-ton, ten-ton truck. So the companies that bought hydrogen today and delivered farm, it's about 15% of the total market, the more we talk to them, the more we realize that they are paying the highest dollars per kilogram for hydrogen, and then they want an on-site solution, but there's no on-site solution for them because everyone else wants to make large electrolyzers, which are geared towards replacing SMR, like steam methane reformers, just to replace them with electrolysis. And we realized there are multiple technical problems.
(09:32):
Something that I'm also very passionate about is distributed energy, as in not creating these massive central hubs of energy production, but push it to wherever you need it. It's always easy to move current and other sources than moving actual atoms of hydrogen. So our whole thesis started forming after talking to these companies into, "Hey, can we create a truly on-site and the edge production of hydrogen? And that can scale for across applications, starting with feedstock applications." It took a lot of conversations. It took almost a year-and-a-half of meeting all these companies.
Cody Simms (10:04):
All right, I want to back you up, and you mentioned the existing hydrogen projects that are out there that are not steam methane reformation projects, but that are electrolysis projects, green hydrogen, so to speak, that most of them were simply trying to replace large scale steam methane production capabilities. Why is that? Is it because they could plug into the existing distribution mechanisms, however inefficient they may be?
Siva Yellamraju (10:28):
Yes. It's also logical if you really look at it one way, where if you're a company which raised, I don't know, a lot of money or if you're Plug Power or Siemens, you want to go after companies that give you a $10 million contract. So you're going to go after the Shells, ExxonMobil. It's a self-fulfilling industry where you have 85% of the market is SMRs, and then they want to replace either ammonia factories or refineries which have these SMRs. So it's kind of a natural way to start for them.
(10:56):
But the problem with that is then you are making yourself go with, "Oh, it costs about $1, $1.50, maybe, a kilogram if you're using natural gas. Electrolysis is not there today, so you need to make it cost competitive. Then you are relying on government subsidies. It's just a steep curve. But if you're putting in billions of dollars backing, you want to actually rather have three companies giving you $100 million contracts. So that's why people have gone with that.
(11:21):
And traditionally, the other way from a technical standpoint is if you're making electrolysis stacks, which are basically cells, if you think of them as large metal plates with membrane, the modern electrolysis, at least, the conventional wisdom in the industry is, "Let's make large stacks," because that's how economies are scale wise, "The largest stack we make gives you the cheapest balance of plan." And large stacks are naturally more easier to replace SMRs. It's kind of an IBM mainframe story, if you will, that's the analogy I use the best, where instead of disrupting it from the day one. On the other hand, the way we look at it is we look at it as, well, there is a small market, maybe 15% of the market, which uses distributed hydrogen. We can go with that. And we think, the way hydrogen market will grow is also distributed, where you have fueling stations, where you have power backup solutions. And what is the architecture that will address that, as opposed to replacing SMR?
(12:13):
So we are actually not as interested in replacing SMRs, as to creating an onsite. It's more efficient, cleaner, and there's a lot more fun doing modular stuff.
PART 1 OF 4 ENDS [00:12:04]
Cody Simms (12:23):
SMR, just for listeners, STEAM Methane Reformation, just to clarify the acronym there, for folks who don't know the SMR acronym.
Siva Yellamraju (12:29):
That's right. I've been in this industry for a little bit, that I tend to use my own jargon. The only reason distribution of hydrogen makes sense is if you're an oil and gas company, because you want to distribute something. Otherwise, there's no reason for anybody to actually transport hydrogen at all.
Cody Simms (12:44):
And so if I understand then, if you're trying to solve the use case of companies today who have to buy hydrogen that's transported to them, and thus are not big enough to command large scale onsite generation, that would mean you're probably not working with large chemical companies that are using hydrogen to build ammonia for agriculture in large scale ways. Rather, you're looking at smaller companies who might need hydrogen as a small part of the input process to what they're doing, or other use cases where hydrogen is supporting their business but isn't core to their business. It's more of a mid-market kind of strategy. Is that correct?
Siva Yellamraju (13:25):
It's a mid-market kind of strategy. As we scale, hydrogen could be very important to these companies, to their dedicated plants, which do hydrogenation. For example, active pharma is an example. They're metal sintering businesses, which use hydrogen as either a fuel or hydrogenation.
(13:40):
So anywhere where they cannot get onsite large presence today, that's our sweet spot. It's not that hydrogen is not critical into their mission, it's just that their overall production volume, or consumption volume of hydrogen, is not at a stage where ammonia plant would have, or a steel plant would have.
(13:58):
Having said that, we will expand to the larger and larger sizes. It's just how we choose to go there. In fact, what we believe is doing even large plants, is to do with small modules, just stack them up. The analogy that works best is Tesla Megapack, in a way where you use small batteries but you create an array architecture.
(14:16):
Day one, we are focused on the mid-market, tail end of the market, not tail end, but mid-market side of it. The way to look at it is a hundred kilowatts per megawatt in capacity.
Cody Simms (14:25):
You use the IBM mainframe analogy. Where I would go with that would be, in order for you to be cost competitive, essentially building at the edge as opposed to building at the center, you need to be able to use fairly cheap materials to actually construct these things out at the edges. And essentially, your hardware hopefully becomes somewhat commodity over time. Is that a correct way to think about it?
Siva Yellamraju (14:49):
I would say it's not going to be a commodity. What will create is a better analogy, probably is a battery systems, where we create a closed loop learning system, where there's a significant amount of data intelligence in the system, which will in turn reflect the stack's design, and then that'll deflect the operational algorithm.
(15:04):
But yes, the hardware, as much as PCs have become a commodity, but still there's enough innovation there. The way I look at it, when you look at hydrogen electrolysis, it's fundamentally an electrochemical problem. You have catalysts, you have a membrane, and a large metal plate. Think of it like a large metal plate with membrane in the middle, and then current passes across it, and then water breaks. Our water passes through in the middle.
(15:26):
So far the focus has been, what is the right electrochemistry? What is the right catalyst we use? How much catalyst coding we use on the stacks itself? What we believe is, it actually is an electrochemical problem, but it is also an operational intelligence problem, as in, you can take the best stack that's possible. If you don't run it at its peak efficiency, there are many parameters, like what current you run it at, what temperature, what pressure you run it at.
(15:48):
And then minor degradation curves, as a stack grows. The operational side of things hasn't been focused on, so what we do is we make small stacks to begin with, and that allows us to use more aggressive catalyst techniques, where you can actually quote less. We can actually do more aggressive stack side optimization, but also allows us to do, instead of having a large stack, we have thousands of small stacks all connected independently in array, and then have it almost push the complexity of the overall system from a hardware material science problem to a software control system problem, where we can actually look at hydrogen production as a data problem.
Cody Simms (16:22):
There is some complexity to the actual electrolyzer itself, in terms of you having to used very rare metals, expensive metals, and whatnot to build these systems. Is that accurate? Assuming you're using a PEM electrolyzer footprint?
Siva Yellamraju (16:35):
We're a PEM-based architecture, although we are not married to any [inaudible 00:16:38] chemistry. Tomorrow, if AEMs are more, AEMs are anion exchange membranes, as you may know, we could move. But we are using PEM. You're right that it does use precious metals, internally, like platinum, iridium oxide, etc.
(16:50):
All the coding that we use is so small now, that that is a significant cost, but it's not something that's worrisome. It's not like you use a lot of platinum, and the cost advantages of using one versus the other is marginal. The actual stack, the cell architecture itself is fairly commodity. We believe there's not a whole lot of fundamental innovation that you can bring in, other than manufacturing efficiencies, like how do you make mass produce yourself? Like roll to roll manufacturing, what thickness of catalysts you would use?
(17:19):
Other than that, the PEM electrochemistry is fairly understood. We didn't want to innovate in that day one, because it's a significant lead time if you want to change something. There's a step function, potentially, but that requires significant innovation.
(17:31):
What we do use is we use artificial chemistry, but we do have our own stack design, which is like I said, because we use small stacks, we have a lot more flexibility in what we can do, in terms of mechanical sealing and how we manufacture these things. So our focus is actually on more on the manufacturing efficiencies. Like you said, it almost making the stacks themselves commodity, with some proprietary design parameters, which link us with the operational side of things. You can take our stack, run it. We won't get the same efficiency as we do, because it's closed loop system.
Cody Simms (18:02):
Will this mean that to deploy a system out in distribution, presumably because these are not large factories that you're building, this isn't a project finance heavy CapEx related play. You're actually selling hardware, essentially, to these companies that are making an economics decision on investing on on-site generation relative to the cost of buying transported in hydrogen. Am I understanding correctly?
Siva Yellamraju (18:30):
Yes, we are also only a year, year and a half into this, so this model is evolving as we speak. But we have customers preferring both. There are customers, for example, have off-tick agreements, that buy hydrogen in just per kilogram basis. And they would like to have a similar arrangement with us, where we deploy the devices, finance them, and then we charge them as a service.
(18:52):
There are other customers who also want to buy the device upfront. Both models are possible, but our deployment model is simple. Like you said, they're literally like cabinets. We try to use a data center, power racks, server racks analogy quite heavily in the company. In fact, we call our modules blades. It's almost like a data center server rack, where you have blades of hydrogen production in the middle, and you can populate as many of them, and you can put as many racks as you want.
(19:18):
Our deployment is we just ship you the devices in racks. It's a skid-mounted, you can just mount them, connect water, connect power. It's a city-grade water. And then out comes hydrogen, you just use it. If you don't like it at some point, we can even take it away. It's not something like a construction project that someone needs to come and construct you for two years.
(19:36):
It's more like deploying, I don't know, a diesel power generator or a car charging station. So we are significantly easy to deploy. But on the business model side though, we do want to hear customers, and figure out how they want to do it. We do know that even with the CapEx that we have, even if you do a hydrogen as a service, at a price below what they're paying for transported hydrogen, anywhere between $7 to $20 a kilogram is the average transport hydrogen price, we actually can get the device priced back in about a year and a half, even if you were to just finance it and deploy it.
(20:06):
So there is significant upside for both the customers and for you. Even if you go down the service path. We are hearing both, so we might end up doing 40, 50% upfront payments, about the rest of the customers doing service. Either way, there seems to be a significant price leverage that we would have, if you look at transported hydrogen.
Yin Lu (20:25):
Hey everyone, I'm Yin, a partner at MCJ Collective, here to take a quick minute to tell you about our MCJ membership community, which was born out of a collective thirst for peer-peer learning and doing that goes beyond just listening to the podcast.
(20:37):
We started in 2019 and have grown to thousands of members globally each week. We're inspired by people who join with different backgrounds and points of view. What we all share is a deep curiosity to learn and a bias to action around ways to accelerate solutions to climate change.
(20:52):
Some awesome initiatives have come out of the community. A number of founding teams have met, several nonprofits have been established, and a bunch of hiring has been done. Many early-stage investments have been made, as well as ongoing events and programming, like monthly Women In Climate meetups, idea jam sessions for early-stage founders, climate book club, art workshops, and more.
(21:11):
Whether you've been in the climate space for a while or just embarking on your journey, having a community to support you is important. If you want to learn more, head over to MCJCollective.com and click on the members tab at the top. Thanks, and enjoy the rest of the show.
Cody Simms (21:26):
Are you seeing customers who are contemplating this purchase in line with some other broader movement around clean energy? Meaning maybe they're doing on-site solar, they're investing heavily in this on-site solar project, and now they know they have excess capacity they have to manage, and so they're going to use the excess capacity to generate hydrogen, et cetera?
Siva Yellamraju (21:47):
You kind of nailed it. We go to talk to a customer, they're already doing solar, and then they're like, oh, we have excess capacity, now we need to store that energy, but also have this hydrogen problem, so buying batteries. Why not just solve the hydrogen problem? That's a big push.
(22:01):
In fact, I haven't spoken to a customer until now, who haven't already started looking at on-site hydrogen solutions. There's a strong push around green hydrogen. Even if it's not green hydrogen, even if it's grid-powered hydrogen, it's still 30-40% lower emissions than transported hydrogen.
(22:17):
In general, the green incentives are significantly pushing these companies to start looking. Although these are independent factories, they may not really care about it as much. But the problem is, they sell to other companies which actually care. So for example, we talk to pharma companies in India, where they sell their intermediaries to European companies or American companies.
(22:39):
Now their customers are also pushing them to lower their emissions. So they're looking at all solutions, and that's a big push for us, in terms of customer traction. Also, the government policies, for example, in India, every company that uses hydrogen is now asked to reduce their emissions, go to green hydrogen by 2030 or something like that.
(22:57):
Governments are also pushing them. Even in the US, with IRA, and fortified-e, and all that. So there's a lot of momentum in the market for selecting hydrogen. But on top of all that, the reason why they're most excited about us is, it's actually cheaper than transported hydrogen. This period, everything else is kind of a cherry on top. Without any government subsidies, without any other things, for 15% of this market, it's actually cheaper than what they're paying for hydrogen today. It's cheaper, it's safer, because we essentially we place a large stored tanks of hydrogen with an on-site device that just produces it, mostly without any storage, or maybe a minimal buffer storage. And in many cases, it's actually safer. It is cheaper, and then it's cleaner, and other things come as a value add on top of it.
Cody Simms (23:40):
You talked about some of the customers you're talking to who are just looking to replace existing hydrogen use cases, and looking to swap you in relative to transported hydrogen that's cheaper. You mentioned pharma, you've mentioned chemicals manufacturing. What use cases do you see being unlocked that economically can't be solved today?
Siva Yellamraju (24:01):
There are two use cases that we would grow into. One of the things is actually power backup. Obviously you have diesel power backup everywhere, including data centers. Now we have data centers for example, exploding in terms of power density. There is no way you can actually do all that with grid. And they have problems with onsite like renewables and all that. And these companies are actually increasingly looking at how do we do no emissions or lower emissions if you want to get out of diesel. That's where I think hydrogen will play a lot and that's how our technology is actually tailor made for that kind of applications. Because it's modular, you can install some of it, you can keep growing, it's all fully software controlled. We actually are the most efficient production of hydrogen end end-to-end irrespective of a load. Because we have so many modules, we can distribute the power problem very efficiently.
(24:44):
Whether a data center is at 10% capacity or 100% capacity, we can actually be the most efficient power backup solution. So cost-wise, obviously it's not cost-wise competitive with diesel on paper, but the whole point for us to go through this mass producible small modules, go after the feedstock market is we can actually bring the cost to a point where it's actually cost compared with diesel. And then the moment it's also obviously greener and cleaner. So you can actually look at, the power backup is one of the big areas where we think it's a big place where we'll push in.
PART 2 OF 4 ENDS [00:24:04]
Cody Simms (25:16):
I assume that also when you look at the full picture of economics, back to the earlier conversation about onsite solar generation or onsite renewables generation, if you take the full picture of we might've had to have not used the full amount of generation at any given moment because we have too much capacity. You're essentially getting free power to generate hydrogen in theory, I guess, in those cases.
Siva Yellamraju (25:37):
Exactly. If you have onsite solar, it's very cheap and almost all these applications, especially the data centers that I like to talk about normally, they're okay doing solar and wind and all that locally. But the problem is that cannot address the problem of data centers fully, because you could have over provision your solar, but it's not going to be addressing what if you lose in winter? What if you don't have enough solar? You have a lot of extra power, extra provision power in these places, and at that point the power is cheap. You just already have, it's just extra power.
Cody Simms (26:07):
In addition to onsite generation of hydrogen, they would need some kind of fuel cell or something to then convert the hydrogen into electricity.
Siva Yellamraju (26:14):
We do both stacks. It's the same architecture, it's just reverse. But yes, we would deploy a full system. Right now we are only, at the feedstock application, we are only looking at the electrolysis. But it's the same system, it can have both storage and fuel cell. And again, the reason is because it's software control. A lot of people sometimes say, oh, software is easy, but that's not really the case. If you look at BMS systems and how more modern battery systems are, the stacks and the fuel cells are largely commodities. Now you have the best possible state of the art stack. How you run them is actually difference and that's what we come in. And then going back to the power backup, or these extra power use cases, the cost economics of power come down radically. If you use 2 cents per kilowatt hour as a power price, solar can give you 2 cents a kilowatt hour. Electrolysis becomes as cost competitive as even SMRs at certain Cap Ex. Right now you don't get 2 cents a kilowatt hour.
Cody Simms (27:09):
But ultimately you'd be competing in that use case against lithium ion or LFP batteries, wouldn't you?
Siva Yellamraju (27:14):
We don't. It's actually orthogonal and in fact this is fundamental problem with hydrogen. If you look at Cleantech 1.0, if I can use the phrase, there's been a lot of compassion between oh hydrogen and then batteries one has to win and then there's a lot of stigma around hydrogen on, oh batteries are great, batteries are safe. The problem is, they're completely mutually complimentary technologies. We would never push for hydrogen in places where lithium ion makes sense, where for example an EV or shotgun power storage. Hydrogen, where it comes in is a longer term power storage. Two reasons, one is the price economics of batteries or any battery technology, lithium ion for instance or others like sodium ion or flow battery, any kind of battery technology, the price is linear with the amount of storage you need, so it's price a kilowatt hours, price per kilowatt hour. The more storage you want, the more you pay for it.
(28:04):
As a result, at some duration it becomes price-based prohibitive to keep doing batteries. You cannot have a data center that needs like say two days of power backup with lithium and it's just not possible. There's just so many batteries that you need to put, the cost is going to be significantly higher. Hydrogen on the other hand, is actually favorable power price economics then because it's priced at kilowatts, so you're only paying for what you use and you just add another tank, another tank. The storage is relatively... Incremental cost of storage is significantly marginal compared to the storage [inaudible 00:28:33]. So at some point those curves cross over.
Cody Simms (28:34):
The production cost is higher but the storage is cheaper.
Siva Yellamraju (28:37):
That's correct. So as an example, if you have a data center, let's say you have a megawatt in capacity, it needs a 48 hour pack up. For batteries, you need a 48 megawatt hour battery pack. For hydrogen, I can just need a one kilowatt hydrogen and then just add storage to cover for 48 kilo megawatt hours. It becomes cost efficient. The second reason is batteries are very efficient obviously 90%, 95% in some case round trip, that's instantaneous. The longer you store power in batteries, they self discharge and they lose power. If you have an EV, you'll know as you leave it.
(29:08):
At some point the battery efficiencies exponentially drops off with time. Hydrogen, on paper is not as efficient round trip maybe 50%, 55% round trip, it's static. At some point again, those curves cross down and then hydrogen becomes actually more efficient for a long-term power, so any and each way we don't compete with batteries. In fact, a power backup solution will actually include batteries. You use batteries for two, three hours, one, two hours of storage, and then you use hydrogen for longer term storage. And then you combine a solution which manages everything in an intelligent way. That's how an ideal solution would work. They actually work together.
Cody Simms (29:41):
This is very far from the use cases we've been talking about for Fourier, particularly focused on mid-market distributed use cases. But given everything you just explained, it would seem to me that hydrogen would be a good candidate for seasonal, long duration energy storage. Is that not the case?
Siva Yellamraju (30:00):
I'll even push it further. Hydrogen is probably the only candidate for seasonal storage that I can think of, which is practical. The other ways are there's flow batteries, there's other... Pumped hydro is probably one where geographically if you have that hill, that's one. Other than that, hydrogen, there's no scientific discoveries to be made. Hydrogen is actually a leading candidate. In fact, if you go to NREL, they're only looking at hydrogen as a grid scale seasonal storage. Again, that comes much later in our path. Again, we started with that backtrack to saying, what is the practical way we can get there and this is the line we are drawing.
Cody Simms (30:29):
We haven't talked about transport at all. That seems like the first place most people who don't spend all day thinking about this problem think about as the future of hydrogen, the power trucking or potentially aviation or things like that. The only actual real world use cases I hear about today for hydrogen are things like forklifts at fulfillment centers. I'm curious where you see the evolution of transport and the role you all might play therein.
Siva Yellamraju (30:54):
Hydrogen is probably again the leading candidate in jet aviation. The reason I say jet aviation is large planes, or longer planes and then maritime. It could be either directly hydrogen or it could be a carrier of hydrogen either in the form of efuels which are hydrogen based or ammonia in terms of shipping, or it could be directly hydrogen. One way or other, hydrogen will be the fuel source for all these use cases. There's no doubt, and even heavy trucking, heavy haul trucking is another use case which I think hydrogen will play a large role. The reason we are not super actively looking at is I think it's a little farther away, maybe 10 years down the line, maybe five years down the line. Especially with ammonia. With maritime, it's already kind of realized a little bit. I think there are a couple of companies working on ammonia based transportation.
Cody Simms (31:37):
We had Amogy on the pod recently, they walked through what they're working on. It's a good conversation.
Siva Yellamraju (31:41):
There you go, yeah. Amogy is doing that. I don't know if ammonia is the carrier or hydrogen itself is a direct, but either way it doesn't matter to us. You need to produce hydrogen. The way we look at it is there will be an equivalent version of supercharging stations where you would have onsite hydrogen production as a fuel, especially for trucking. In fact, we do talk to some trucking companies, including the bigger, Amazons, Walmarts, we talk to where not only for forklifts but they have long distance distribution trucks that could potentially be hydrogen, and where they fuel them locally. They charge the last mile carriers with EVs like batteries and then they have the long range ones.
(32:17):
The same setup will have power coming in. Part of it is used to produce hydrogen, part of it is used to charge batteries. That is a vision that I can fully see happening. It won't be a retail, I don't see hydrogen cars ever being a reality, that just doesn't make any sense to me. But anything heavy transportation, construction, mining equipment, for example, dump trucks. Main reason is very simple, hydrogen is energy dense, batteries are really bad in terms of mass and duration. You'll never have a plane or a ship move if you put batteries to power it, large ones, not the EV tools.
Cody Simms (32:50):
I'm somewhat hearing you say, all the areas where people have maybe said hydrogen is a good solution to this, but they're skeptical. Your response might be, well suspend the disbelief that you don't have to have hydrogen pipelines or you don't have to truck it somewhere. But if you could generate it on site, could you solve these use cases? I guess that's the way you're probably approaching the problem.
Siva Yellamraju (33:11):
That's right. So when we looked at it, the fundamental skepticism... Not only skepticism, the fundamental bottleneck is transportation of hydrogen.
Cody Simms (33:20):
Let's talk about geographic footprint a little bit. The US certainly has some initiatives in hydrogen. There were benefits that came out of the inflation reduction act around hydrogen. I feel like EU and in particular France has been leaning in really hard to hydrogen. I'm curious how you see different geographic markets playing out for you.
Siva Yellamraju (33:39):
Germany.
Cody Simms (33:40):
Germany too? Okay.
Siva Yellamraju (33:42):
Yeah, EU has been big into hydrogen for a while now. In fact, US, we are just catching up to EU in many of these industries. Although I think with IRA now US is very fast catching up and now we also have the technology edge here. Germany is big because Germany has stopped producing nuclear they need, then with the war, Germany is big on hydrogen. The way we look at it though is we want to stay away from EU in the beginning at least, because there's significant market in the US and there's significant market in Asia. Those are the big pockets that we are focused on. For example, hydrogen is huge now in India. India is now again trying to be energy independent and hydrogen is a big role to play in all of it. It's one of the largest consumers and they have huge pockets of solar installation, they have a lot of extra power and they need to store it. So those are geographies we look at, India, Indonesia,
Cody Simms (34:31):
And not a domestic gas industry too.
Siva Yellamraju (34:33):
Yeah. They don't have natural gas, a hundred percent. And then US, there's specific geographies in the US like for example, Ohio, Kentucky, Washington, Oregon. They don't have natural gas infrastructure and the price of power is cheaper. And even within that, if you just focus on North America on transported hydrogen. In the regions that I laid out, California would be a bad place because price of power is so high. It's still $5 to $7 billion of market in North America, in US and Canada together. So that's what we obviously focus on day one. Because we're a small company, we're located in Palo Alto, so we want to focus on domestic market first. On that note, we actually did our first pilot with a pharma company in India. It was whole experience shipping our device all the way across the oceans. But India is a big market, we just need to figure out strategically when to go and how to express that market.
Cody Simms (35:20):
All right, well let's talk a little bit about where you are from a deployment perspective and traction perspective. So you've mentioned now this pilot in India, I think a couple of times. Where else are you in the process of getting your tech out in the real world?
Siva Yellamraju (35:33):
As a company, we officially formally existed in about a year and a half. That's when we did our last round general catalyst, led our last round in March. I think that's when we closed the round. And then since then, our goal for this phase of the company was to build a team, build a pilot prototype that actually we proved to ourselves that the algorithms work and then do a pilot, which we did all of them. Plus pilot was kind of a lab pilot for a pharma company in India. We just signed a couple of LOIs in the US for commercial grade pilots. We actually have a prototype of a commercial device just now in our lab that was just finished last couple of weeks ago and now we are scaling it up. So the first pilot that we're going to do will be a plant in Ohio. A chemical factory doing petrochemical products is another one, a fueling application that we're also looking at. So for good or bad, we have actually a lot of inbound and outbound customer interest. We need to pick and choose who we do pilot. We are a small company. We have limited resources on how many pilots we can do, but our goal is, because I come from, for good or bad, again, a software background, I'm far more excited about doing even minimum product and doing pilots even with 80% hardware as opposed to waiting for the full baked system to come in.
(36:47):
Our focus and our style and culture of the company is always go to the customer. So we're going to do a couple of pilots later this year, hopefully by November, December timeframe. We're also working on a design for a manufacturability outfit. We're talking to some manufacturing folks, and start production hopefully in early next year and then commercially rolling out these devices by maybe late next year. So that's the plan I have in mind.
PART 3 OF 4 ENDS [00:36:04]
Cody Simms (37:12):
What are your goals in terms of what you're trying to learn in these pilots?
Siva Yellamraju (37:15):
There are many. I think the safety itself, while it's a big improvement, just working with safety folks in these companies is a whole experience ourselves, what they care about, what they don't care about, what do we automate, what do we not automate? At least my tendency, if left alone, would be automate everything and put everything into software, but there are real things that the customer wouldn't want, wouldn't have [inaudible 00:37:35]. So there are things like that we learned.
(37:37):
The second thing is obviously installation and deployment itself. We have some assumptions on how we think these things are deployed, but unless we actually go do a pilot, customer install it, we'll know the different conditions that are there, things like permitting, things like water quality, things like variability and hydrogen requirements. So there are a lot of things that we would learn from these pilots, which we won't be able to learn if we don't do these pilots, like how do they operate, what kind of interfaces they need? Do they need alerts on their phones? What other system do we need to integrate the software into? I have a whole list of things that we want to learn from these pilots.
Cody Simms (38:12):
How are you determining which of these learnings are generalized and which of them are sector or customer specific type of learnings?
Siva Yellamraju (38:19):
Because we produce hydrogen and it's a commodity at that point, they just use hydrogen in whichever way they want to, the high level parameters that we would learn, like what pressure, what purity, what flow rate, in general the industry, that's largely generalized anywhere. And then the safety issues are also largely generalized. It's just the same issues. The installation could be interesting because most of our current customers who are chemical factories, et cetera, would want the device to be outdoors because that's how their tanks of hydrogen sit today. I don't know if that fully generalizes with some other [inaudible 00:38:50], like a data center where they might want everything to be indoors, or outdoors for that matter. Most of the things we learn, because actually eventually a commodity we produce, they're generalized. All the knowledge that we learn are generalized. Some of the software automation features may or may not generalize, but those are things that we can be very nimble on.
(39:08):
The other thing that we would learn is data. Our whole premise is that we will deploy, we will get more data because we have lots of small stacks. The data that we get is going to be our IP because we have small modules. The hardware, like you said, is largely I wouldn't say commodity, but largely repeatable. What we learn from these deployments, what we learn from these pilots is in different conditions how these stacks perform in different operating parameters. We actually run experiments on those things. Eventually, that becomes our core way of looking at hydrogen production. The more data we have, the more better stacks we will build, and also the better algorithms we build. So, eventually, it's going to be a data play, just like in batteries.
Cody Simms (39:47):
Data feeding your own internal IP though, not data as a sellable asset, I suppose?
Siva Yellamraju (39:53):
It's not something we intend to sell. It's for our own purposes. It's just, the more data we have, the better the stacks will be, the better our algorithms will be. We'd never intend to sell the data. It'll be so precious for us.
Cody Simms (40:05):
What are any assumptions you had going into the company that you felt very strongly about that you have had proven incorrect that surprised you?
Siva Yellamraju (40:16):
When I started looking at this problem, I actually thought electrolyzer stacks are commodity. We'll just get off the shelf stacks and we'll build the whole system and the software and the operation layer all ourselves. That assumption got refined a little bit. I won't say it got completely corrected. What we now realize is the cells or the membranes are fairly commodity. We can get them from anybody. 100 suppliers, almost likely the same thing. There's a lot of innovation in the stacks, owning the stacks itself. How do you construct it? How do you mechanically lay out and all that? So we moved away from buying commodity stack systems to potentially making our own stacks but using commodity cells. We did some cost analysis on that, which actually is very practical. It's a no-brainer. And the more we ran these stacks, the more data we got, the more we realized that there's so much we can do on the stack design itself, given the data that we already have, that we continue that process.
(41:07):
So that's something that's changed from a technology perspective. When I started the company, I looked at cost economics, which were great. I looked at how we would deploy these things. We thought, oh, we'll make it cheaper. It makes sense, people will buy it, which is largely true. But whenever we go talk to customers these days, and as we did these pilots, the biggest value that they care about is actually safety, which is coming to us much more strongly than even the price. They would even buy it, even if it is higher price than they're paying right now, as long as safety is significantly better. And that value prop is significantly higher, that's what we are learning.
(41:41):
In fact, our device thankfully is a lot safer than stored hydrogen in the factory, but we are now focused increasingly more on as a safety play and safety measurements that we would announce. Those are the two big learning curves for me. One of them is directly out of pilots. Other than that, building a energy company working with various different kinds of engineers is very different from working with just software engineers in your team. So that's a personal journey for me by itself.
Cody Simms (42:06):
On the hardware side, I'm curious if you're starting to see any manufacturing centers of excellence emerge geographically. With semiconductors, we've seen Taiwan emerge. With solar panels, photovoltaics, obviously, we've seen China emerge. When it comes to either the electrolyzer membranes themselves or the stacks, are you seeing any geos become excellent at this?
Siva Yellamraju (42:29):
India. They have the scale of China, they don't have the constraints of China, and then they have huge push into energy these days. Almost guaranteed we'll have a manufacturing center in India. First manufacturing center will probably be in the U.S. There's a lot of talent there in terms of mechanical engineering talent, in terms of manufacturing engineering talent, because India is also a reasonably large auto industry, which is gone very legacy in a way. The same groups of people and the skill set, they're looking for newer industries. Leveraging existing auto industry and then using that talent to produce energy-based manufacturing, I think India will be a huge place in terms of manufacturing these kind of stuff.
(43:07):
It's also helpful that it's closer to Taiwan and other places where you can actually make your electronics somewhere and then just assemble everything in India. So that is something that I'm very excited about. Also, I'm from India, but beyond that, that's an area where I've always had a presence in India for all my startups before, but it was only because I just came from there. But here makes sense to have a location in India.
Cody Simms (43:30):
I may be saying this very naively, but I feel like India has never really had a core IP in the energy space, and fascinating to hear that hydrogen could be a pathway for them there.
Siva Yellamraju (43:42):
Yeah. It is changing because like you said, they don't have natural gas infrastructure, they're largely dependent on external factors for energy, and they've realized with solar and hydrogen they can just be independent. When I go to India and I talk about the company, all the way from senior government officials to senior industrial folks like Tata and Reliance, they're very, very keen about doing anything in hydrogen, anything in energy independence, anything in energy abundance. It is also a very hard market to do business with, to be fair. It's not like America where it's easier to do business, but that's changing too. If you can navigate that process, it could be an unfair advantage on its own if you actually navigate it properly.
Cody Simms (44:21):
Last question. Fourier, explain the name.
Siva Yellamraju (44:25):
A lot of people are familiar with the Fourier transforms in math, in engineering. What a lot of people don't know is Fourier actually invented global warming or invented the greenhouse effect, so he was a first person to propose correctly that CO₂ emissions and other emissions are warming up the planet, more than other planets, for example. At that point, he didn't realize it as a problem. He thought that's what makes us live. But he is known to be the father of the greenhouse effect in many ways, and because it's a name that's well known and I've used Fourier transforms all along, and what we are doing is transforms, that's the reason. We wanted something memorable and I always like playing homage to people who affected us in many ways. So that's the name. It's a simple name. People know it, but more than anything else, it has a very deep history around greenhouse effect. But nothing more complex than that.
Cody Simms (45:14):
Siva, thank you so much for sharing so much about while you're building at Fourier. Your website is still very stealthy, I guess I should say. Really appreciate you coming on here and actually sharing more with all of us about while you're building. Is there anything I should have asked or that we should have covered that we didn't talk about or any help that you need right now that you want to make sure to share with people?
Siva Yellamraju (45:34):
I think at this point we're finished whatever goals we had for the C-state. We're in the process of raising more money. Again, just early stage talking to some investors. Raising money is top of mind, as you know. You interview a lot of folks in this domain. They are capital intensive, in some ways, and the biggest differentiation is raising money at the right time. At least that's been on top of my mind. Otherwise, yeah, we do look for team, hiring, mechanical engineers, and a bunch of stuff, so always we're growing. We are building this. All of the C-state's problems that other companies have, we also have them and opportunities.
Cody Simms (46:14):
Well, Siva, thank you so much. It's been great learning from you, and thanks for sharing with us.
Siva Yellamraju (46:18):
Awesome. Thanks, Cody. I really enjoyed it.
Jason Jacobs (46:21):
Thanks again for joining us on the My Climate Journey podcast.
Cody Simms (46:25):
At MCJ Collective, we're all about powering collective innovation for climate solutions by breaking down silos and unleashing problem solving capacity.
Jason Jacobs (46:34):
If you'd like to learn more about MCJ Collective, visit us at mcjcollective.com. And if you have a guest suggestion, let us know that via Twitter at MCJPod.
Yin Lu (46:47):
For weekly climate op-eds, jobs, community events, and investment announcements from our MCJ Venture Funds, be sure to subscribe to our newsletter on our website.
Cody Simms (46:57):
Thanks and see you next episode.
PART 4 OF 4 ENDS [00:47:09]
