Fourier Is Solving Long-Duration Storage with On-Site Hydrogen
Siva Yellamraju is the Co-founder and CEO of Fourier, a company building modular, on-site hydrogen systems for long-duration energy storage and distributed hydrogen production. Fourier’s approach combines lithium-ion batteries for short-duration needs with hydrogen storage for longer durations, packaged into a single system designed to operate like a microgrid.
In this episode of Inevitable, Yellamraju explains why hydrogen has long been considered promising for multi-day, seasonal, and even multi-year energy storage, but has struggled because of the cost and complexity of transporting it. He argues that producing hydrogen where it is needed can remove that transportation challenge and unlock a different cost curve for long-duration storage.
The conversation explores Fourier’s modular “hydrogen blade” architecture, its first storage deployment in Gujarat, India, and the growing demand for alternatives to diesel generators as data centers require increasingly large amounts of backup power. Yellamraju also discusses why levelized cost of storage matters more than round-trip efficiency, India’s potential role in energy manufacturing, and his transition from software and data centers into building physical energy infrastructure.
Note: Fourier is an MCJ portfolio company.
Episode recorded on August 5, 2026 (Published on September 1, 2026).
In this episode, we cover:
(0:00) An overview of Fourier
(1:33) Why hydrogen could work for long-duration energy storage
(4:12) Why centralized hydrogen production creates a transportation problem
(5:31) Making hydrogen storage operate like a battery
(7:08) Why levelized cost matters more than round-trip efficiency
(10:08) How hydrogen stores and returns energy
(12:22) How long hydrogen can remain stored
(13:45) Fourier’s battery-plus-hydrogen microgrid architecture
(14:24) Why Fourier uses a modular “hydrogen blade” design
(15:51) Combining lithium-ion batteries with hydrogen for all storage durations
(16:19) How Fourier’s software optimizes storage, lifetime, and ROI
(17:47) Fourier’s first commercial storage deployment in Gujarat
(19:53) Why Fourier is expanding from hydrogen feedstock into energy storage
(20:32) How AI and data center growth accelerated demand for long-duration storage
(24:24) Replacing diesel generators and solving multiple power needs with one system
(26:55) How hydrogen storage compares with iron-air batteries
(27:50) Why India could become a major energy manufacturing and hydrogen hub
(30:10) The engineering and manufacturing talent Fourier is hiring
(30:39) Transitioning from software to physical energy infrastructure
(34:11) Scaling Fourier to meet growing demand for energy storage
More episodes:
-
[Cody Simms] (0:00 - 0:43)
Hey, it's Cody. Before we get into today's episode, I want to tell you about something happening during Climate Week NYC. Jeff Lawson founded Twilio and turned it into a $4 billion public company. Now, he's running Inertia, building laser fusion, and he just closed a $450 million series A. We're taping a live episode with him on September 22nd in New York at the JP Morgan Chase Tower. The link to join us will be in the show notes or just go to inevitable.fm/NYC2026. Hope to see you there.[Cody Simms] (0:42 - 2:02)
Today on Inevitable, our guest is Siva Yellamraju, co-founder and CEO of Fourier. Fourier is attempting to solve the tricky problem of long-duration energy storage by enabling on-site distributed hydrogen production. The company's bet is that lithium-ion batteries are too expensive for multi-day storage, and that hydrogen has historically not been a viable option due to the expense of transporting it.Fourier builds modular, on-site hydrogen production systems that are arranged like a server rack, with swappable hydrogen blades instead of one large electrolyzer stack. The company has a commercial storage pilot running in Gujarat, India, powering a facility around the clock on solar and hydrogen. MCJ is an investor in Fourier, alongside General Catalyst via our venture capital funds.
From MCJ, I'm Cody Simms, and this is Inevitable. Climate change is inevitable. It's already here.
But so are the solutions shaping our future. Join us every week to learn from experts and entrepreneurs about the transition of energy and industry. Siva, welcome to the show.
[Siva Yellamraju] (2:02 - 2:04)
Thank you, Cody.[Cody Simms] (2:04 - 3:06)
Well, it's been a minute since you were on here, and you all have made tremendous progress. And I think the topic I'm really interested in diving into with you is about hydrogen's potential for long-duration energy storage, and really unpacking why hydrogen could be a solution, and also why it hasn't yet been a solution to that space.Explain a little bit. I mean, I think theoretically, people have been talking about hydrogen as an energy storage solution for some time. NREL has had a report for a long time saying it could be really one of the only solutions for seasonal storage or truly multi-month energy storage.
And I'd love to hear from you some of your perspective on what's held it back and also what the potential is.
[Siva Yellamraju] (3:06 - 4:53)
To start with, from a first principles perspective, like you said, I obviously agree with you that hydrogen has a large role to play in energy storage, anything that's more than a few hours. The main reason for that is hydrogen is one of those high-density fuels, which is very unique in its price structure. Any other or most other storage solutions are priced for storage density.It's dollars per kilowatt hours. The more you want to store, the more you have to pay. It's usually a straight line.
The slope obviously is different. For lithium, it's probably like $120, $130 a kilowatt hour right now. For something else, the slope might be slightly different, but it's a straight line.
Hydrogen is kind of weird in the sense it's usually priced for power density, so it's all dollars per kilowatt. No matter how much you want to store, there's a straight line, very steep. You have to pay upfront, but the more you want to store, it's just another tank.
It's just another tank. So it's not a straight line that way. So at some point, no matter what other storage technology you take, the line will cross over and hydrogen will become one of the cheapest ways to store power at a certain duration.
And depending on what other, like if it's lithium, there is a certain duration. If it's something else, it will be another specific duration where hydrogen will come the more meaningful way. So it makes sense from a first principles perspective.
It's also clean. It's easy. The chemistry is well understood.
The reason it doesn't get realized, and this is why we actually started Fourier to begin with, is two problems. One, how do you make it efficiently at different scales? And how do you transport hydrogen?
Because while it's a great solution for storing power or energy, it's really a molecule that does not want to be transported. It's a molecule that's very light. It's very flammable.
As a result, the transportation of hydrogen becomes a big problem.
[Cody Simms] (4:54 - 5:15)
And I think what you mentioned too about the way it's priced, which is basically priced per unit of density, has resulted in the phenomenon where to get to economies of scale, you end up with giant centralized facilities building large amounts of it, because then you're just like you said, add another tank, add another tank, but you end up with this transportation issue. Is that right?[Siva Yellamraju] (5:16 - 6:13)
Yeah, that's fundamental. I would say what most folks or most incumbents miss is the economies of scale that typically applies for an IC engine or a large industrial hardware equipment doesn't really hold for this. In fact, you really don't want to create these large central hubs for hydrogen.The hydrogen production as an efficiency doesn't necessarily scale with the larger the facility. In fact, the transportation costs add up, the installation costs add up. So our approach to it is you have to do it where you need it at the edge.
So you don't need to go these massive hubs and you kind of remove transportation altogether. And you have to do it in a way where it's modular so it can be done at the edge. Depending on the scale, it could be a small hospital, it could be a medium-sized data center, it could be a gigawatt scale facility.
You can do it at the edge, you remove transportation, and actually the efficiency can scale at any of those scales that unlocks the potential for hydrogen.
[Cody Simms] (6:13 - 6:42)
That assumes those edge use cases, those edge users actually want to be in the business of producing and storing hydrogen themselves, which I guess if you're getting down to using it as a storage mechanism, you naturally are assuming they would want to. But does it require more expertise on their end to set up a hydrogen production and storage facility on site than it does to plug in solar and batteries on site, for example?[Siva Yellamraju] (6:43 - 6:49)
Yeah, it's no different than a battery. That's how we solve it. They shouldn't think of it as a hydrogen expertise.[Cody Simms] (6:49 - 6:52)
And when I say battery, I mean lithium-ion battery, obviously.[Siva Yellamraju] (6:52 - 7:49)
Again, when a company chooses to store power with a lithium-ion battery, they're not experts in lithium chemistry. So same thing will apply for hydrogen also.You need a package where you think of it as a box, you store hydrogen. Chemistry is hydrogen, but it still operates as a battery. All you need is a use case for the need to store energy for beyond four hours, which obviously is an amazing tool for that.
Once you have that use case, it just becomes another chemistry. If you can containerize it, if you can deploy it as a box that you deploy it, then it's just another storage system that you're doing. So we definitely don't think the companies should be thinking about, oh, I'm going to make hydrogen.
I'm going to store hydrogen. I'm going to produce power from hydrogen. There are three different pieces right now.
If you want to set up this solution, for example, there are three different pieces. You go to different companies to do all these things. What Fourier is trying to do, for example, is you don't need to do any of that.
It's a black box, kind of a microgrid. We install it. It operates as a battery after that.
[Cody Simms] (7:50 - 8:39)
So historically, hydrogen held back from the long duration energy storage world largely because it's cheap to produce in large quantities in a centralized scale model and then really expensive to move around, which just doesn't make sense if you're wanting to use it all over the place as a onsite storage mechanism. I think we'll get into what you guys have built in terms of your distributed nature. When you compare hydrogen to lithium ion, one of the shortcomings, I think, is around roundtrip efficiency.And maybe you can unpack some of the deltas there, but you gain this much, much longer duration storage mechanism than lithium ion can even physically do. Am I correct in understanding those differences as well?
[Siva Yellamraju] (8:40 - 10:03)
Yes, you're correct. Oftentimes people talk about roundtrip efficiency as a primary metric. It's definitely not the primary metric you should be focused on.You should be really focused on dollars per levelized cost of storage, dollars per kilowatt hours over lifetime. And that's the real number that matters. For example, internal combustion engines are 20% efficient.
We still use them because it's cheaper to do it in many places. So the roundtrip efficiency is important as a construct, but it's not the only metric that you need to look at. And if you go with levelized cost of storage, like you said, a lithium battery system growing beyond a few hours becomes practically impossible because the costs just scale up.
And because that's $120 or $130 a kilowatt hour, no matter how big it is, number one. And number two, the area that you would require to put these batteries, because they're not as dense, it's also immense. And last problem is the batteries are very efficient.
The roundtrip efficiency you're talking about, it's great in a short duration. The longer you store, it's actually not that great. It actually drops off also.
If you ever have an EV, you leave it for a few days. It'll discharge itself. There's a self-discharge problem also.
So it's not 95% efficient across all durations. So eventually the efficiency also will dwindle down with batteries. But the right metric to use would be an LCOS or some sort of a unit cost for eventual lifetime.
[Cody Simms] (10:05 - 10:49)
When I went to pull out my battery electric powered drill yesterday for an at-home project, I was dismayed that I did not have my battery sitting on the charger for sure. I got about three turns of the screwdriver and that was it. Help me understand, we'll get into Fourier's specific solution, but describe at a high level, just hydrogen as a long duration energy storage product.What does that look like? I think of hydrogen, I think of electrolyzers. I think of the inverse process of using a fuel cell to generate power.
Describe both production and consumption of hydrogen in a battery and what that looks like. And then we'll talk about your architecture specifically, but I think first just paint the broad picture.
[Siva Yellamraju] (10:50 - 11:18)
At a high level, the chemistry is very simple. So you have excess power, excess solar. You have energy coming in.You want to store it. You use that excess power to break water into hydrogen and oxygen, of course, but mostly hydrogen that you care about. And then you store that using electrolysis.
Electrolysis is an electrochemical process where you use current or energy that's coming from a solar panel, for example, to break the water molecule into hydrogen and oxygen. And you store that hydrogen.
[Cody Simms] (11:18 - 11:20)
This is all quote unquote "green hydrogen" in this case.[Siva Yellamraju] (11:20 - 11:22)
If it's solar, obviously it's green hydrogen.[Cody Simms] (11:23 - 11:25)
Using solar and using an electrolysis based process.[Siva Yellamraju] (11:25 - 11:56)
If it's solar and wind, it's all green. If it is something else, the colors are...We actually spend very little time thinking about and talking about colors in four years. I apologize ahead of time if I'm not going to be talking in those color spectrum, because all that matters to me is a sidetrack. It's just levelized cost of storage.
Just cost is what matters. Anyway, going back, you take that hydrogen, you store it in a tank, and that tank can stand for however long you want. Days, seasons, years, if you want.
And then whenever you need that power back, you run the same process in reverse and you get power back.
[Cody Simms] (11:57 - 12:04)
So generating the hydrogen is running it through an electrolyzer. Turning the stored hydrogen into electricity is running it through a fuel cell.[Siva Yellamraju] (12:04 - 12:23)
Yeah, there are different options. The fuel cell is the most obvious option. You can also burn that hydrogen in a typical IC engine where you combust it to get power.It could also go into a different kind of a fuel cell, not just necessarily a hydrogen fuel cell. For example, companies make different fuel compatible fuel cells.
[Cody Simms] (12:24 - 12:29)
We just had Mainspring Energy on the show, and they're using an electromagnetic process to convert gases.[Siva Yellamraju] (12:29 - 12:45)
There are different ways to do the same thing. Mainspring's a good example. And all of them have different scales of efficiencies, but let's call it fuel cells.And there's also an even more novel approach where if you design the stacks properly, you can do the same stacks, can do both forward and backward passes too.
[Cody Simms] (12:45 - 12:56)
Got it. So you may combust it, you may convert it electrochemically, electromagnetically, however, but it goes from a tank of hydrogen back into onsite electricity.[Siva Yellamraju] (12:57 - 13:04)
Yeah, as I said, the fuel cells and electrolyzers don't necessarily have to be two discrete things. They could be the same cell doing both back and forth.[Cody Simms] (13:04 - 13:09)
Are there theoretical limits to the duration of the storage of hydrogen in a tank?[Siva Yellamraju] (13:10 - 13:59)
Theoretical limits will be very, very large. Any tank, no matter what you do, will have some leakage. Like if you store it in a liquid tank, there is boil-off.But if you store it in a compressed gas, it's like a tank. It just sits there forever. But there will be some minor gas leaking through, but it'll be in years.
So the theoretical limit is significantly, almost you can argue that it's not there anymore. And you're not losing anything. It's just a tank of gas sitting there.
The theoretical limit comes in, obviously, how long you want to operate and maintain those tanks because these are not just, yeah, it's a tank sitting somewhere, but you need to have a perimeter. You need to have some regulatory safeguards on it. So there's operational costs managing it to some extent.
But then there is no theoretical short-term limit for how long you can store. You can go for seasons and years.
[Cody Simms] (14:00 - 14:10)
So the competition becomes for the duration you want to store, at what point does this solution become more cost-effective than any other storage solution you're using?[Siva Yellamraju] (14:10 - 14:26)
And that's all hydrogen. In fact, I know you said we'd come back to Fourier, but what we are building is it's a battery plus hydrogen together in one kind of a microgrid. So you have a standard LFP, and then you have hydrogen all working together.So you can actually, with that system, you can go all durations.
[Cody Simms] (14:27 - 14:33)
Let's do it. Walk through the architecture that you all have deployed now on the ground in India.[Siva Yellamraju] (14:34 - 15:02)
Yeah, so maybe I'll have to give a little bit more overview of what our architecture looks like. Obviously, we covered it in the past, but I want to maybe run through briefly. Our main hypothesis is that you don't need the gigantic stacks or electrolyzers to solve this problem.You need, in fact, the smaller cells and smaller stacks are more efficient, easy to manufacture and scale very well to any size. Similar to a battery system, modern battery system, which has a lot of tiny cells, if you look.
[Cody Simms] (15:02 - 15:05)
This is the commodity Google server rack playbook, yeah?[Siva Yellamraju] (15:05 - 16:32)
Yeah, and then you have a compute system that controls all that. We are kind of similar to that. Our system is literally like a server rack with blades.We actually call them blades because I come from that world. Instead of server blades, you have hydrogen blades, and then they just rack them up and you can stack them as many as you want. The system is equipped to go from power to hydrogen or hydrogen to power.
So for example, what we did for the first few years of the company was mostly focused on generation market. But we've started with this pilot in India, which we did actually continuing to do, but the deployment happened two, three months ago, where there is a solar plant, which is one of the largest solar developers in India. And they're powering this particular facility and it's obviously still not 24 hours.
So we install our system, which is four towers in their case, and there's a storage component to it. And we connect the solar power. We do the same thing, store it in hydrogen, run it, and then we enable truly around the clock, 24-7 operation, just powered by this solar plant.
And that's basically the first deployment we ever did in that storage market. The storage itself is capable of going beyond 24 hours, but current system actually holds power for storage. For us, if you want to go from there to two days worth of storage to cover incidental power losses or seasonal, even winters, we just add another equivalent tank and go from 24 hours to 48 hours with almost no additional cost.
[Cody Simms] (16:33 - 16:39)
So you're bringing the lithium ion battery storage on site for the truly short duration use case.[Siva Yellamraju] (16:39 - 16:39)
That's correct.[Cody Simms] (16:39 - 17:00)
You also provide the hydrogen production and then storage and then conversion back to electricity.I don't know if you want to describe, we talked about the various mechanisms of converting back to electricity, sort of what you're building there. And then I assume you have a software layer that sits on top of it and manages where and how power is routed.
[Siva Yellamraju] (17:01 - 17:51)
That's correct. As you said, that's the discrete pieces, but it's like one box or one set of boxes that we install on them. The battery piece is, it's a separate use case.It's power coming in. There is obviously a dynamic management system or we call it cell management system that is deciding whether the power should go to the battery, whether power should go to the electrolyzers and storage. And when you're drawing power, how much power is being drawn from what elements simultaneously looking in real time, optimizing for roundtrip efficiency, but also more importantly, it's actually optimizing for ROI or lifetime.
Because depending on how you operate them, you can actually run them at higher efficiency, but then you lose out on the lifetime. So the system's looking at a future problem of, okay, if I run this, how long it lasts and what is the ROI, really the levelized cost of storage is what's being optimized.
[Cody Simms] (17:52 - 17:58)
So I'd be looking at a shipping container with some pipes coming in and out of it and maybe some storage tanks next to it. Is that sort of what it looks like?[Siva Yellamraju] (17:59 - 18:29)
Yeah, a better way to describe it is it'll look like these towers, like car charging stations have these cabinets that are sitting and then there's one big conduit of power going in, like the solar power going into it and then water goes in. Water consumption is very little because we kind of use all the water pretty much. And then the same conduit will send the power back.But it could be put in a shipping container, but they're discrete cabinets that just get deployed depending on the size you want. And they all are interconnected and they're all daisy chained in that way. There's only one connection.
[Cody Simms] (18:29 - 18:32)
And what's the initial deployment? Who's it with and how's it going?[Siva Yellamraju] (18:33 - 20:01)
I'm not at liberty to say the name of the customer yet, but this is one of the largest solar developers in India. They develop solar plants all over the country. They have gigawatts of solar in the pipeline.And this is in a site in a state called Gujarat in India, which is a very industrial area. And it's going great. As I said, we're powering this office facility for around the clock, 24 hours, fully on.
In fact, they're super excited about it. And it is the price that they get for power is end-to-end with solar, extra solar capacity to call for the storage, to call for the battery component and our system CapEx all together. Cost of energy we are predicting is between eight to 11 cents a unit.
And this is end price for the customer, including all of it together. It's actually cheaper than grid price for them. And they're super happy.
In fact, they have a larger, close to a megawatt size plant that they wanna completely power with the system. They're gonna do a new solar installation next to that and then put in our system. We're actually negotiating a full offtake and a 20 year offtake with them.
So it's going as well as it can. I think I'm excited. There's another pilot that's coming up at UCSF pretty soon.
So we're now graduating it into more commercial pilots in the U.S. And that is also for the large Japanese corp. I guess I can say that name out. It's a company called Obayashi Corp.
And they're building a facility in UCSF and they're piloting our same system there.
[Cody Simms] (20:01 - 20:01)
That's great.[Siva Yellamraju] (20:01 - 20:34)
It's going as well as it can.The whole team is pumped. In fact, I can share a before and after pictures with you later. It's just, it was nothing.
It was just not even level ground. We went in and maybe three weeks is all it took for us to deploy the whole system. And for a company like ours, we actually had to hire contractors to actually level it, build the concrete all the way up.
And it was a pretty cool experience. We have a small team in India, but it's just doing all that seems pretty surreal for our company at this stage.
[Cody Simms] (20:35 - 21:14)
How did you land on the storage market? When we had our first conversation with you a few years ago, it was all about, hydrogen doesn't just have to be used for mobility. It can also be an industrial feedstock serving as an actual chemistry input into heavy industrial process.And it felt like there was a little bit of, oh, everyone thought about hydrogen as a car fuel cell technology. And clearly lithium ion was winning that war. And there had to be some kind of mental redirect away from that toward, oh no, it's actually a chemistry input.
When did the storage problem present itself? And how did you decide to go seize on that?
[Siva Yellamraju] (21:14 - 24:06)
To be honest, the reason I started Fourier because I spent a lot of time in data centers, my past life, I've worked on deep AI problems, myself deploying these models on data centers at Google and other places. The energy storage problem is very real. Eventually this is all becoming an energy problem.And it's obviously tokens. Then you have compute, then you have energy. In that order, all of them have to be relying on energy.
So for me, I was motivated to solve that energy resilience. That is why Fourier exists. We always knew that storage for these solutions, we have great short-term storage solutions, but we don't have great long-term solution.
And that's where hydrogen, as I said, will play a large role. The motivation for us was always that. That's why we exist.
That's how the system is designed to be dynamically adaptive to that. Having said that, what I did not expect when I started Fourier was how fast the storage market will catch up. I thought it was maybe five years away when we started three years ago.
We really focused on feedstock because that's the market, as you said, industrial feedstock market exists. People use hydrogen. They get transported hydrogen today, which makes it ridiculously expensive.
So what we are doing is do it onsite, no transportation, cheaper than what they're paying for hydrogen, half the price. And that is a real market. We've done multiple commercial pilots with that.
We actually signed off the agreement that it's a revenue generating opportunity for us already. When we are commercially deploying those systems all over, there's one going on in Anaheim right now. There's one in Dallas, Chicago area, Michigan area, Pennsylvania.
So all across the country, that is an actual deployment that we're working on. It's a real market. It's about $7 or $8 billion in the U.S. in that niche transported merchant hydrogen market. While we were doing that, obviously, ChatGPT happened and the data centers exploded in terms of their demand. And if any data center, for example, needs a 48 hour at Google for all of our data centers had a 48 hour minimum power backup. And they all do it with DG gensets without any exception.
And now there are two problems. One, they're completely backlogged. You can't even get a hold of any DG gensets anymore.
Number two, yeah, these data centers used to be 5 megawatts. Now they're talking about 100 megawatts, which is a norm and gigawatt as like what's going to happen. There is no way you can actually do that with a diesel.
Even if you only think about a backup solution, not even forget about the storage, other parts of storage. So they are looking at other solutions to store. And we've gotten increasingly reached out by all these folks, both hyperscalers, but otherwise solar developers.
So the demand has been real, very real. And there's tailwinds around that. And it always followed our vision.
The unit economics makes sense. And this opportunity, for example, in India came on and there's other opportunities. So it's more market driven than anything else, but it was always part of the vision for the company.
[Cody Simms] (24:06 - 24:23)
I mean, we've seen Bloom Energy explode, obviously with market demand for fuel cell generation of power. They're mostly based on natural gas today and that's a generation use case. It's less of a storage use case, as I understand it.[Siva Yellamraju] (24:23 - 25:05)
Yeah, it is. And it's not a store. Even then they need to have natural gas availability to begin with.It's still not clean. Also, this is not going to be one solution for all problems kind of a market. This is a gigantic trillion dollar market where different solutions make sense at different places.
But the reality is the cheapest way to get new power is solar. It's not even a policy decision anymore. The cheapest way to get new power is solar, but solar fundamentally has intermittency issues.
That's why you get gas turbines and other issues to just compensate for it. If you actually have a true solution, which can enable that cheap power into a firm, reliable power that can only be solved with storage, then the new deployment, the time to market for these things will be off the charts.
[Cody Simms] (25:06 - 25:34)
But the opportunity for you, just to be clear, isn't to be the power system powering the entire data center. It's really to be a diesel generator replacement. Instead of going and buying diesel gensets and having to continually manage the OpEx of buying diesel fuel and filling it up, you can plug a Fourier system onsite and be long-term backup power for that facility.But you have to be a pretty sizable system to do that, I think. Am I following the thread correctly here?
[Siva Yellamraju] (25:35 - 27:31)
That is correct. And I'll use some characterization of the system itself. That's a primary use case that people are talking about, which is power backup and diesel gensets.But once you deploy that system, it actually addresses four things. That's why we also have a battery and the whole microgrid. You're talking about four problems that you have.
One is flex, basically in terms of their peak demand. And then there's peak shaving, which is basically shaving off whenever they have spikes of usage. And then there's obviously flex as in the power daytime versus nighttime flex usage.
Then you have UPS, which is basically your real-time, if you lose power. And then you have power backup, which is basically a much longer-term problem. Although the power backup is the main use case that they're interested in, once you deploy it, it can be one solution that addresses all of them.
You don't need to have a separate UPS system, a separate flex system. And then you can actually have, data centers often have this other problem where sometimes they want to have 150% of their input power. They just want to run high end.
They might have interconnect today for let's say 100 megawatts, but sometimes they do want to have 140 megawatts. And this also enables that you can actually have that extra power that you want. There's multiple use cases that it will address, but primarily the DG genset is a valid replacement.
Yes, you're talking about hundreds of megawatt hours of storage. Diesel obviously is cheaper CapEx day one, significantly more OpEx, or so the levelized cost of storage is comparable. And also it's dirty, obviously.
A hundred megawatt hour system with Fourier would be roughly three shipping container size. So you're looking at about 700 megawatt hours to a gigawatt hours per acre density storage wise, in terms of land requirement. It's gigantic.
Gigawatt hours is a lot of power, but still about an acre of land. It is no other system in that market which can address, like for example, if you go with iron air, that's easily four to seven acres.
[Cody Simms] (27:31 - 27:36)
And we're seeing those deals get done now with, you know, Form Energy is landing these massive deals, right?[Siva Yellamraju] (27:37 - 28:31)
It's great. It's very encouraging for us because you brought up iron air and again, there's different ways, there's different systems. More than the land usage, roughly same round trip efficiency, like about 40, 45%.We're about $18 a kilowatt hour landed costs versus $20 that's projected by iron air battery systems. But more importantly, that system can only go 100 hours up and 100 hours down. I would actually argue that they're probably 200 hours up and 100 hours down.
You cannot change the rate. With a hydrogen system, it's much more flexible. You're not stuck with just the backup solution.
You can draw all of it if you really want it. You can architect the system that way much faster. You can store it much faster or you can only run it for 15 minutes and then use it as a UPS system.
So it's a very different system for slightly different use cases, but there is an overlap. And that they're landing these contracts, it's super thrilling. That clearly shows the market's there.
[Cody Simms] (28:32 - 28:53)
One area I'm curious about, and I'm going back a little bit to the pilot that you're running, is one of the hypotheses you had when we recorded an episode a few years ago was that India was going to become a key player in the supply chain for hydrogen, much like China has for lithium ion and solar. And I'm curious how that has been playing out.[Siva Yellamraju] (28:54 - 30:26)
Yeah, I think that's playing out well. It's a bet that we are making also. We have a team now in India.We're going to scale manufacturing there. It's not just hydrogen. I think India will play a key role both in manufacturing and supply chain for energy equipment in general.
India has a large homegrown auto industry. I mean, India is probably one of the biggest auto markets in the world. And a lot of it is domestic.
And as a result, there is a lot of industrial hardware manufacturing expertise that's in the country. Where China excelled, obviously, in semiconductors, silicon, obviously, lithium batteries, etc. But we see this untapped potential and almost, I would argue, that the right geopolitical veins are blowing where you need another alternative for China.
And again, given where we are, it will play a huge role in this whole space. Even more importantly than the technical expertise that is available and the raw material, it's also a gigantic market. For example, India's bets on hydrogen are significant right now because India is one of those rare, large countries which does not have a lot of domestic fuel production.
They are reliant on a bunch of other countries for fuel. And the only way they can be self-reliant is they have a ton of solar, they have a ton of wind, but they need firm storage. And they are betting a lot on hydrogen in that sense.
So it's a huge market, plus a lot of talent that's there in chemical, industrial, manufacturing processes. It's a valid bet, and it's also a full-on English-speaking democracy, which is obviously very great to have.
[Cody Simms] (30:27 - 30:31)
Have you been building the team both in California and India?[Siva Yellamraju] (30:32 - 30:47)
Yes, we have an office now. We just moved to this office in Mountain View, but we also have an office in Hyderabad in India. Hyderabad is one of those industrial hubs in India.And right now the team's largely engineering, but now we're exploring scaling manufacturing also as we scale.
[Cody Simms] (30:47 - 30:51)
What kind of folks are you looking for? What's the hiring ramp up right now?[Siva Yellamraju] (30:52 - 31:20)
Hiring's always the bottleneck, no matter what stage you are in. And we are obviously looking for mechanical engineering expertise. That's the main expertise that we look at.How do you design these systems? Electrochemical engineering, but also electrical engineering. Bunch of what we do is as much electrochemical, but it's also a lot of power electronics.
We design and build our own power architecture. There's a lot of power electronics. So main roles we're looking at is in power electronics engineering, mechanical engineering, and possibly manufacturing.
[Cody Simms] (31:21 - 32:07)
I'm struck that you are a founder who comes from, for lack of a better term, traditional tech. I'm laughing at myself because as we were starting the recording, my webcam was blurry and I couldn't get it quite focused on me. And that's technology that you absolutely nailed in one of your prior companies and sold to Apple and is the crux of FaceTime camera technology that tracks your face around.And you also built some incredible technology that became big parts of YouTube in the past, the acquisition. Now you're a few years into a business building molecules and very much hard technology. What has surprised you in making that transition?
What's been easier than you thought it was going to be? And what's been harder than you thought it was going to be?
[Siva Yellamraju] (32:07 - 32:10)
Quick pushback. The technologies in the past were also very hard.[Cody Simms] (32:11 - 32:12)
Yes, of course they were.[Siva Yellamraju] (32:12 - 34:51)
But yes, I hear you. It is a very intentional pivot for me. I mean, I'm a computer and software programmer myself and being there, it's actually started with the realization, like I said, everything actually became an energy problem.And if you don't solve energy, you really have a compromise, either quality of life or creativity. And you cannot see future generations needing to make the compromise. They shouldn't.
Just energy resilience has to exist. And when we talk about data centers that need to be the size of a New York, we need to solve this. So the motivation was there.
What's not different, to be honest, Cody, is the engineering side of things is the same for me. It's an engineering problem. Yes, in the past, I used to work with mostly software engineers, but maybe some product designers, some UX designers, and then products come together.
Now it's a more interdimensional team. You have mechanical engineers, you have electrochemical engineers, you have electrical engineers, all have to work together. But when we go to brainstorm and we talk about engineering, system architecture additions, for me, that is almost same.
Just still a system I look at it. Fundamental first principles of engineering are still applied there. What is very new and what is also very exciting as a result is the markets that we are going into.
One end we sell into these very hard industrial markets and one end we sell into these other side of the data centers or other side of the construction sites. That is a pretty interesting opportunity for me because I've never even, to be honest, I've never imagined that I would go into a random city in Gujarat to do this deployment. That is the most new, most probably interesting, but also most challenging from a personal standpoint for me to adapt and get to.
And that also makes it interesting. You said what's easier. It's actually much easier to hire a top class team and the missions alignment is significantly.
This company, for example, the mission alignment is so obvious in the beginning itself that there's a self filter in what kind of people we attract. So it's actually in some ways easier to attract top talent and then work with them and then keep engaged on a day-to-day basis because it's the why we start and why we do every day is so strong that everything's actually becomes easier to some extent. And there's a lot of other differences, the capital intensity of the business, how much money you have to raise, which is obviously the least enjoyable part of the role.
All that is different. You need a different scale and you burn through cash much, much faster. But the bottom line is seeing something physical in a physical location, dealing with electrons directly.
I think that feeling's just significantly different than pushing off software on the internet and then measuring it using metrics.
[Cody Simms] (34:53 - 35:11)
Congrats on what you all have built. And you've been incredibly capital efficient at doing it and excited by the amount of progress you've made in discovering now multiple markets for your technology and seeing, I would say, a real tidal wave of one open up right in front of you that hopefully you are in the midst of pouncing on aggressively.[Siva Yellamraju] (35:12 - 35:32)
I think we are. We are pouncing on very aggressively. Obviously, it's time for us to scale up under this market.I mean, obviously, our series B is coming along pretty soon. We're hiring. As you said, we are capital efficient, but it is a big bet, right?
I'm quite pumped. It's just the power storage is the true unlock for the scaling up these data centers.
[Cody Simms] (35:33 - 35:44)
Thanks for joining and sharing an update. And congrats on the progress and really excited to continue to follow along. And thanks for including MCJ along for the journey in what you're building.We're proud to be investors.
[Siva Yellamraju] (35:45 - 35:48)
Thanks, Cody. Likewise, you've always been a champion for us.[Cody Simms] (35:49 - 36:15)
Inevitable is an MCJ podcast. At MCJ, we back founders driving the transition of energy and industry and solving the inevitable impacts of climate change. If you'd like to learn more about MCJ, visit us at mcj.vc and subscribe to our weekly newsletter at newsletter.mcj.vc. Thanks and see you next episode.