Investing Billions - E127: Interview with CEO of AI Company Backed by Founders Fund and Khosla Ventures

Episode Date: January 7, 2025

In this episode of How I Invest, I sit down with Felix Ejeckam, Co-Founder and CEO of Akash Systems. Felix discusses how Akash is leveraging synthetic diamond technology to revolutionize cooling for high-power microchips used in AI servers and satellite communications. He shares insights on their $68 million CHIPS Act funding, the importance of U.S. semiconductor manufacturing, and Akash’s journey through tough times to becoming a leader in cutting-edge technology. A must-listen for anyone interested in semiconductors, AI, or transformative tech innovations.

I think a lot of companies or investors would like to replicate the process that you went about identifying new markets and executing against that. Talk me through the process. How did you gather the intelligence and the data in order to repurpose your technology? Paying attention to what your customers are saying very closely, paying attention to mega trends going on in the world. So folks who may be your customers, maybe don't know about your customers today, but are they going to be potentially your customers down the line?

So listening to these massive mega trends and knowing that you're going to intersect the world and the puck where it will go to, not where it is today, and the puck where it will go to, not where it is today. I think that that's a key element to trying to position your company for where the world is emerging into. Otherwise, you just missed the boat. So what is Akash Systems? Akash Systems is a company that uses synthetic diamond to cool down microchips. These are microchips that are so powerful that we are able to do things like what chat GPT does

or what a satellite does. Very powerful electronics. We're able to use synthetic diamond to cool them down and make them perform much, much better than otherwise. And when I say much better, I'm talking about in terms of raw power or efficiency. We've picked two products to start. One is the AI server, the types of systems that use GPUs, which you may have heard about. And then we have also picked satellite radios.

These are very, very effective satellite communication systems that beam down from space. Data, such as you might see in your GPU, your GPS system, when you pull up Google Maps to navigate your drive. We grow diamonds in the lab. These are synthetic diamond materials. And we use these diamond materials to cool down these chips.

How much of the cost of a server chip has to do with heat and power? Oh my god. Most of it. So in a typical data center, for example, something like 40% of the cost of operating a data center, which is essentially a very large warehouse collection of servers, 40% of the operating costs monthly

is cooling air conditioning, trying to reduce that heat. Uh, similarly in your, something as small as your iPhone. I've seen some pretty crazy numbers thrown out there in terms of market size for data centers, data warehouses, chips. How do you quantify the opportunity set for investors? We estimate our market size to be anywhere from 200 to about $400 billion.

Keep in mind that we sell servers as well as satellite communication systems and a radio that fits into a satellite is about the size of an iPhone, maybe twice the size of an iPhone. These systems sell for hundreds of thousands of dollars per unit. And so we estimate that our market is on the order of $200 to $400

billion just looking at those two markets. There are other markets which we haven't started to address today but are in our roadmap such as you might have in EVs, electronic vehicles, such as in smartphones, communication systems, and not just in the phone, but also in the base stations that route the calls from phone to phone. And then of course, in radar systems, which is used in defense. Radar is a kind of listening device that depends heavily on the kinds of cooling chips that we make. You just received $65 million from the CHIP Act.

Congratulations. Tell me a little bit about that. Thank you. We're excited about it. So we just received $68 million in CHIP's funding from the government, the Department of Commerce's CHIP and Science Act. And the funding is coming to us to help us scale, scale our manufacturing and assembly of our products, namely the ones I just mentioned, the servers, the Diamond Cool servers, as well

as the Satellite radio systems. This funding will help us set up a plant, a semiconductor fab in Assembly House in Oakland, and just help us scale in terms of getting space, equipment, hiring people, and doing so as fast as possible. This is non-trivial to do. There is a race right now to scale. All of us who are manufacturing gear in tech have a challenge in trying to scale. And this funding helps us address that. We just received as well a large customer order for servers. We've seen bipartisan calls for onshore and critical technologies in the US and for on-shoring critical technologies in the US and not to rely on foreign countries or foreign adversaries.

Tell me a little bit about the CHIPS Act. Was that part of that effort or is this just a totally unrelated trend that you're seeing in the market? The CHIPS Act is absolutely connected to the bipartisan effort to try to bring back to the US manufacturing and assembly of semiconductors. We are extremely excited about that effort because the CHIPS Act helps us not only bring back semiconductor manufacturing to the U.S. but also shores up our national defenses. So being able to make your own semiconductors the thing that powers all defense equipment is an essential part of a country's national security and we view

our work to be very, very vital and central to that national security strategy. It helps us in securing our supply chain for all of our nation's major and critical economies. We will never again have to depend on other countries, certainly not our adversaries, to make the things that we need. And making these semiconductor chips and materials in Oakland is central to that. This is also a question of our economic security. We're building jobs.

We're going to hire hundreds of people in the order of four to five hundred new people in Oakland over the next four to five years. These are all things that every American is excited about, certainly on both sides of the aisle. And I should say that the hiring we're doing is going beyond California. We have a site in Texas that will benefit from this as well. And you mentioned this space is really heating up.

Obviously, there's a lot of demand from AI companies. Who are your biggest competitors, and how are you guys differentiating against those companies? In the area of sort of high powered, highly cooled, high efficiency servers, I don't think that we have any competitor at this point. Certainly someone who we have to worry about in terms of performance. In any conversation with a customer, so far we always win because of the performance that we bring to bear. Just to illustrate that example

of how Diamond is so effective, we do this in front of customers, and every time, as soon as we do it, people will just sort of light up, even before we show how the server performs. If you don't mind, I'd like to show you how diamond works by moving heat as rapidly as possible. This is a piece of diamond that we've grown in the lab.

It's about half a millimeter. Absolutely. Very, very thermally conductive, five times more thermally conductive than the next best material in nature, that being copper. I'm gonna show you some ice cubes. And the ice, obviously it is cold. My hand is naturally warm, my body temperature.

I'm gonna hold this ice and you're gonna watch as it slices into this ice cube as if it were butter and my fingers are gonna feel so cold you can't tell but I can I can tell you that that's what I will feel but what you will see is the diamonds moving heat from my fingers into the ice cube rapidly and that'll just be and you'll see that because it goes goes right in like butter can't know if you can see it but yeah it's going to write in just sliced it right there I'll do this other one here as you can see so there's some thermodynamic factors coming into play here so there's a transfer of heat from

your fingers to the to the ice and from your ice to your fingers with the diamond being a conductor? That is correct. And that's exactly the feature that we bring to bear in a server, in a diamond cool server, which we sell. And again, we're making a very big announcement next week, a big order, a massive order from a customer, one of the biggest cloud AI compute companies in the world. And this feature is what is making that possible. Explain exactly how that works within a server or within a chip. So we take this diamond and the diamond, the features of the diamond, the way we grow the diamond, the dimensions of the diamond are all proprietary,

and we use proprietary adhesives, and we apply it to the GPU. Okay, so just like you saw with the diamond touching the ice, the diamond is on top of the GPU, the main heat source. We also do the same thing with other chips in the server, and because of that, because you have the diamond being such a great thermal conductor, it is able to spread the heat and reduce the temperature of the server. You reduce the temperature

of the server, it means that your fans or your liquid cooling do not have to work as hard. The burden on those cooling systems is greatly reduced. In fact, we have found that we're able to reduce the electricity going into the fan, the cooling system by 90%. This is money going straight to the operators of the data centers, the folks who are paying the light bill. If you are doing inference work,

it means you can overclock the GPU a lot faster. You can get more clock cycles. You can train your models faster, all because we're able to, by thermal conduction, extract the heat directly from the GPU to the diamond and then to whatever liquid cooling or other fan cooling systems that you may have in the server. And we do this demonstration for customers all the time.

We just had the US election. There's been a lot of exuberance in different parts of the tech ecosystem. From a business perspective, how does that practically affect Akash and your business? We're very excited because as far as we are concerned, the things that we extol, the virtues of our business, the fundamentals of our business have benefits that apply to all Americans, regardless of your political lenient, whether you're on the left or on the right, middle, if you're pro-business, pro-jobs, whatever you may feel, we think that there is universal appeal

for our commercial security, our national security, for creating jobs, bringing manufacturing back to the United States so that it is not outsourced to our adversaries in China. There is universal appeal and we have just received nothing but support from everyone who has heard our message. Folks who saw that we won the CHIPS Act

have sent in very positive support to us and we're excited about that. I invested in 2019 in the Series A and then you guys had a tough time between 2020 and 2022. You did another round at that point. I'm very glad I reinvested then. But tell me about those times and how did that affect the organization?

Those were tough times. I think it was those were the COVID years. Very difficult for everyone in the country. I think other than the handful of businesses, this was something that every country, every company had to grapple with. And the thing about these tough times, times of correction is that it forces everyone to step back and think about your fundamentals. Do you have a distinctive technology or product that makes you stand out. These quiet times, as I call them, enable companies to slow down and get back to basics. What is your business plan? And that's

what the COVID years enabled us to do. We then focused on literally making our products for space even stronger. And we're going to see some launches. Several companies are launching our satellites with our radios in January, just a few weeks from now in April. But also, we started to extend our products technologies into new markets, chief among them, AI and data centers. That wouldn't have happened had it not been for the quiet time

afforded by the COVID years. That's what we did. We stepped back in a very disciplined way, developed some very new applications of our technology, went into the new markets that we could see coming down the pike, AI, and that's enabled us to launch the products just a few weeks ago and to now sign this large

contract with a major compute company that we'll be announcing next week. I believe that organisms when dealt with any kind of issues either become anti-fragile or die, you're basically either growing or dying. How did that affect a cost systems as an organization? I think that it affects organizations, um, either positively or negatively, depending on those basic business plan questions that I listed. Okay. So I think that, um, in, in tough times, if you're not able to answer those fundamental business 101 questions of, is there something very distinctive that has a 10x

value proposition over competitors that you have? Is there a significant market size for that distinction that you have? I mean, you know, like hundreds of billions of dollars in market size. Are you the very best team on the planet poised to deliver that distinction? If you're not able to answer these questions favorably,

then I think you will struggle. If you do, then I think you have some very good years ahead. I think we've looked at, I read this recently how a lot of our great businesses today were born out of the tough times we've had when there's a correction, the corrections of, exactly, exactly, that's usually the time when there's just a mental, there's mental space

for people to just think calmly about big opportunities and these fundamental questions I described. Otherwise, when there's just a lot of euphoria and hysteria, processions, and there's a lot of slush money going around, it's easy for that type of hysteria to fund everything. And then these questions aren't really answered in a very disciplined way. What would Akash Systems be if that bull market kept on going?

Would it just be the wrong business? Would it just be one product? Like what would have happened if there was no correction? If there's no correction, we were probably would have gone further with the space business and maybe the AI might have been pushed out a bit more into the future. Perhaps the world as we know it

wouldn't have leaned in. I think AI was gonna happen no matter what, but maybe a little bit more pushed out. I think that it's possible that the inspiration for the CHIPS act wouldn't be there, possibly this desire and belief to step back, hey, let's shore up our supply chain. Let's not be dependent on everyone in the world for our very vital microchips.

I think that those questions would have been answered in a different way and Akash, we would have evolved. I mean, we have some very smart people in our management team and throughout the company. I think we would have answered the questions well, given the circumstances that the world presents to us. I think a lot of companies or investors would like to replicate the process that you went about

identifying new markets and executing guns that. Talk me replicate the process that you went about identifying new markets and executing guns that talk me through the process. How did you gather the intelligence and the data in order to repurpose your technology paying attention to what your customers are saying very closely paying attention to mega trends going on in the world. So the folks who may be your customers, maybe not quite your customers today, but are they gonna be potentially your customers

down the line? So listening to these massive megatrends and knowing that you're going to intersect the world and the puck where it will go to, not where it is today. I think that that's a key element to trying to position your company for where the world is emerging into, otherwise you just miss the boat. I think having an amazingly talented people around you,

people that are even smarter than you, in the things that you want to do is essential. I have an amazing team around me. Ty and Dan and Pimmy, these are brilliant folks who are extremely good at what they do. And we're constantly challenging ourselves, asking ourselves the right questions. My VCs, KOSLA and Kearney Jackson, Founders Fund,

these are brilliant folks who just have been very helpful to us in making us, making sure that we answer the questions in the right way and that we steer clear of the potholes that can oftentimes sink a company that's trying to evolve out of some very difficult times. I mean, we essentially, the world had a forced major events, you know, an event that sank most companies, most series A companies in the world. And to navigate that, you need some amazingly talented people.

And luckily we had that around the table. There's this repeated concept in venture capital that you have to be non-consensus, right? I think a lot of the times people don't really think about how painful mentally, economically could be to be non-consensus for several years. How are you guys able to kind of go against the grain and walk me through those difficult years where the market had not yet come to you? As the founder have to have conviction and belief in fundamentals and in your core of your value proposition.

If you don't have that, then nothing else matters. You know, running with the herd is can be great when the herd is right and good times. But when times are tough, then I think having that contrarian view is what can give lasting value. Because when everyone's moving left and you see opportunity by turning right, that's when you can make some serious bets. It's interesting because there are different types of personalities I find, you know, the types of, you know, engineers, smart people, PhDs that we interview.

They're the folks who always are thinking of solutions, you know. It doesn't matter what problem they're faced with, their instinct is to have ideas for moving forward, for solving the problem. And then there are folks I find who have an instinct for identifying the pitfalls, the problems lurking around the corner. Yes, there's a big opportunity here, but look at the 25 things that prevent others from achieving that.

So let's not lose sight of these. And I always want both at the table. I especially want the people who see those those risky hurdles lurking around the things that are uncomfortable to talk about, we may not want to fess up to them. And it's just a party pooper to bring up those issues when there's so much euphoria about an opportunity. I always want everyone at the table. And and that nothing because you know what? We are the most successful

when we address every problem in hardware engineering, electrical engineering and in physics. problems don't just vanish. Unless you fess up to them and deal with them head on, they will be there forever. So yeah, we just have to, you know, if these problems are not being talked about, I get worried. There's some kind of healthy mix of optimism and pessimism you need on every team.

You need the entrepreneurs that see the ideal how things could be, and then you need, as a second step, ideally you need somebody could say, yeah, but here are the different things that we need to hear the landmines to avoid so that we could actualize that future. Yes.

Yes. That's I totally agree with that. You need both. What's next for Akash? We've got a lot on our plate coming up in the weeks ahead. We're announcing this big order next week. We're going to, there's a bunch of satellites going up with our radios in January. We will start shipping the servers in Q1.

There's another bunch of satellites going up with our radios in April. I think that next year we are on pace to have probably one of the fastest growing company curves in a very long time. And I can't even begin to imagine what 2026 might look like. But it's just all about execution right now. It's, you know, go, go, go, go, go, build up as many servers as possible for this AI exponentially growing market that we see, and then starting to get some space hardware assets in space. That's the future for us. I think that we see a place, a world where Akash could become the world's preeminent high-end server company in AI, period. I think that that's well within our grasp. I say future, I'm talking about the next 12 months.

OK, so those are exciting things to look forward to. This has been great, and it's great to catch up, and always inspiring to talk to you, Felix. Thank you. Look forward to sitting down soon. Cheers.