Omnitron: From Self-Driving Cars to AI Data Centers

Startups get the party. Startups get the party started. Startups get the party. Startups get the party started.

You want tech and tacos, Tacos and tech, you know we got it.

Hey.

Startups get the party. Startups get the party started. Startups get the party. Startups get the party started. Startups get the party. Startups get the party started.

You want tech and tacos, tacos in tech, you know we got it.

Hey.

And welcome back to the Tacos and Tech podcast. I'm your host, Neil Bloom, and I'm thrilled to have on Eric Aguilar, co founder and CEO of OmniTron Sensors. Hey, Eric, what's up? Welcome to the pod.

Hey, thanks for having me.

It is a pleasure to have you on. I feel like it was meant to happen at the right time, and this feels like the right time to tell this story. We are at quite an interesting point coming right off a few weeks off of CES when the world suddenly woke up to physical AI and robotics and everything. Except you and I have been living in that world for 10, 15, 20 years. So this is a good crossing of the point, I would say, right. You have this conversation. You want to give us a quick high level what Omnitron is?

Yeah, at a high level, we've built the most powerful micro machine. And that's really something exciting to share more in this podcast. But that's, that's, that's what we do.

Okay.

Tiny machines out of silicon.

Okay. We are going to unpack that in a variety of ways so that everyone can have their own little micro box of knowledge here to understand the huge potential. But we got to take it back because one, why now? Why is this the moment? And why you, to a certain regard, why. Why have you been building to this moment, which I think is a pretty awesome story. Some personal pieces to that that we'll get to and we'll unpack. But where does it all start? Where'd you grow up? How'd you end up in the. The tech path here?

Well, I'm a nerd. I always been curious about how things are built and made. That kind of led me to engineering as a skill and I wanted to just kind of figure out how stuff was working and then innovate there. And I got my degree at Cal Poly and then worked down at San Diego at the Navy Research Labs. And I was really fortunate to be able to work on some really cool sensors. This was computer chip type stuff as 6 and, and different types of sensors to help drones and weapons understand the physical world. Right. Because it, it needs to, in order for it to guide accurately to its depth destination and complete its mission. And in that a group of us were thinking like hey, you know this, this is a really cool sensor that we could maybe commercialize. And that's how a company called Lumadon, a San Diego based company was founded. Was they licensed IP from the Navy and worked to commercialize that ip. It's, it's known as an imu. It's that sensor on your phone that does the tilting, you know, when you tilt your video or whatever. That's a naive. And we, we had wanted to, wanted to figure out if we could scale it on a commercial technology. So that's kind of where the company really started and kind of pivoted. They were working on more. What was it at the time?

Wasn't there some oil and gas?

Yes. Sensors for oil and gas. Correct. That's right. Oil and gas exploration. And, and then they're like huh, I think this, this technology can go into a phone. So that's when I joined and I was like oh, that seems exciting. I'll leave the Navy and join when, when they pivoted to that direction. And that's also where I met my co founder Trent for, for Omnitron. We, in, in, in two years we were able to go from like a, a new idea to fabricated chips. And that led to an acquisition by Google that was a, a really fun, fun journey. And from that point I joined Google to work on drones again, but this time in a different capacity. I was responsible for the avionics system on a drone project. The avionics is the brain of the drone, right where all the sensors and actuators go to help it navigate to its destination and deliver food or, and it's still in operation. Project Wing and I wanted to kind of better understand the application space of sensors. Like how are they, how are, how are people actually using this, how are the engineers and scientists actually using this to, to make things work in the real world? Learned a lot and then went on to work at Tesla again. Similar capacity is now doing sensors and software for Model 3 and trying to figure out on a large scale capacity and the impact of the automotive industry in terms of robustness and reliability really started giving me some realizations on some of the challenges that are still pretty prevalent in how, how sensors understand the real world. They're the holes, the gaps, what they can't see, what they can't understand, how different it is from how you and I perceive the world. And, and after That I, I started advising various robotics companies and another self driving car company and started working on LiDAR. And this is kind of where the story of Omitron begins. I was responsible for making sure that. Well, let me take a step back. Lidar is a, is a very powerful sensor for robotics. If you're going to put any sensor on a robot, and that's true today, lidar is the one you want to go with because it gives the robot an understanding of the world in a way that makes sense to measures an object. It'll measure where that bookshelf is and figure out can I navigate to it, do I need to navigate around it? And it's all by doing depth measurement and you need high resolution and that's the ideal sensor for it. But this sensor has been plagued with issues of cost and actually much more importantly reliability. And that was a frustration that I was trying to deal with. The sensors that we were using at the time would fail after two or three months of operation in the real world. And I ended up spending a lot of my time on just structuring operations and maintenance routines for keeping our vehicles on the road. Just, you know, how do I replace this LIDAR unit in a cost effective recalibrate, get the cars back on the road. But that's, that's not a scalable solution, right? So I started talking to the suppliers. I go, why do your sensors keep breaking? What keeps breaking? You know, and as a good engineer, this isn't sexy by the way. I'm sorry, this is just practical, pragmatic, right? As a good engineer I'm like okay, what keeps breaking? What's wrong with this drive? To root cause, right. And, and I realized that the, the, the scanning element, the, the, the spinning thing that's on top of these, you know, waymos and self driving cars keeps failing. It goes out of alignment or starts wobbling. And that kind of propagated. A, a, a real fundamental issue with optics. Optics requires really, really tight tolerances, sub micron level tolerances. And then you have to hold that tolerance under all the temperature fluctuations. You know, everything expands and contracts as a result of temperature and that, that moves microns of distance, right? And, and, and then you have vibration, right? The vehicle's constantly shaking on the car and on the road. And, and these lighter units are using, you know, 1960s discotech technology, for lack of a better word. And it's like there's nothing better. And, and, and so I was sharing these frustrations with my co founder Trent, who's got a strong Background in, in, in mems, Microelectrical mechanical systems or simple. Another way to put it is using silicon processes to make actuators to make something move. You actually have a bunch of these on your phone already. But he's like, hey, I, I think, I think I can solve it. I have this idea that I had years ago I just, you know, kind of put in the back of my mind. But I think if we build a much more powerful little motor out of silicon, we can build a scanner that is much more effective than, than what's, what's available today. And that's where we started Omitron. That's kind of where we first met you and your team. We were raising our seed round and I remember you guys going through diligence and you guys brought in some big hitters. You brought this guy Curtis Ray to do diligence on our company. And in the diligence process. I remember him kind of posted when Trent shared the key aspect of rip, he just went like this. It's like, that is so obvious. I wish I would have thought of that. So shortly after we finished the diligence and raised the capital, we hired him.

Amazing.

But yeah, we've been on that journey and in fact I have it here. Let me show it to you. Let me find it. Yes, we actually, we actually built it. We, we have, I mean we started with building it at a prototype university foundries, but, but this, this bad boy right here, this is the largest, most powerful MEMs chip in the market. What you're looking at there is two of them in this package. 10 millimeter diameter mirrors. That's, that's considered ginormous in our industry. This thing articulates 60 degrees, it moves really fast. So it does what the spinning thing does in silicon and that addresses all of those problems on reliability and cost. And our customers love us, right? We have over 800 million in LOI contracts. We're going through automotive qualification right now because we know that this is the catalyst of what is going to open up the lidar market, not just for automotive, but the rest of robotics. So it's, it's a really, really exciting journey. And it's also why you haven't seen LIDAR prevalent today is, is that is the root cause.

I think this is a good time maybe to explain a little bit of like the war on lidar that's happened a little bit. And I think you're in a good position having worked at Waymo and Tesla, the two biggest players to come out in autonomous, who Pretty much, I think are introducing the world to general purpose sensing. Right. For kind of robotics. Why do you think there's these. I don't know what these camps are going two different ways and one's gonna. Do you think one's gonna prove right, one's not. And maybe take us how that's making you think forward for the auto industry and then feature applications.

Well, I don't necessarily subscribe to a winner take all philosophy. I mean even, even Mr. Musk has been open about, you know, there are spaces for using lidar, like outer space. But he's, he's also pragmatic. Right. Like putting it into a car. Unless you address that reliability issue and costs, it doesn't, it doesn't work for a consumer to, to use that solution. It makes sense from that perspective. But when you want to go after three nines or four nines or five nines of reliability or, or operational time, maybe Tesla is at 90% operational time. But I don't know if you've driven a Tesla or you drive one or if you've been one recently. You know, it's about 80, 90 of my commute. I'm like, oh yeah, it can do it. It's that 10% that it won't overcome unless it's got lidar, something simpler. Cameras struggle with dynamic range. In other words, seeing shadow and bright low light and, and highlight in the same instance. Right. Our eyes are really good at that. And lidar doesn't care because it's actually emitting its own beam to, to see the world. So you can see at night, day, whatever cameras can't. So that's, that's the biggest case. Sometimes you'll see it like, I don't know if you'll, you'll notice that I notice these things because I'm in these vehicles all the time that, you know, it'll, it'll, it'll creep into another lane because it doesn't quite see the lane marker or it'll, it'll miss this or miss that because, because it struggles with that sensing limitation that it has with cameras. But you know, it'll get 90% of the commute done, which is really great. Yeah. But that last 10%, even if it's that last 1%, you absolutely need lidar to, to overcome these, these cases to get it fully operational and safe and robust enough.

Which is why, I mean I, I wrote within a 24 hour period in Austin, I was in 10 waymos. Right? And when you're in that, you're like, we're done, we're here, it's set, right? Like this is ready for the mass market, and obviously it is. And, and you can tell that. And I can't wait. And I think now my conversation with people who've been in one versus not is so different, by the way, right? Everyone who hasn't is still focused on the idea and the trust and everyone who's been in it. It's not even a thought. The trust is there within 10, 30 seconds, right?

Yeah, yeah, yeah. You're curious for a little bit. Yeah, yeah. You're like, curious like, oh, look at that thing do its thing. And then once you're like, oh, oh, what's got it? You're like on your phone like, oh, I have so much freedom now, right?

And I think this is a year where the general population gets there, right? They, more people try it, more people do it, and now it just starts to compound on itself, right? And so does that. When general population is not thinking about lidar, first of all, right. They're just thinking about that extra little spinny thing on that car that they took. And then once they're in the car, they're like, well, it got me there. Moving on. Right?

Um, Nobody cares. Yeah, yeah, yeah, Right.

So we're so. Which is kind of back to your point. Like, just solve the problem, right? Why is this not working great? How do we solve it? Solve it. You know, back to kind of your root cause of why lidar wasn't working. So where is this being taken next? Like, where, where are you being pulled? Because you're just solving the problem here.

Yeah. So it turns out that this little micro machine can be used for a lot of different applications. And there's been a lot of talk recently about AI data centers, especially the power that they consume today. And the issue with the power consumption isn't necessarily the draw of power. It isn't necessarily the chips that they're using. The majority limitation in, in data center compute and power is the movement of data. Today, data centers, the way data gets moved is, is you take an optical signal coming out of a gpu, for example, it goes into a, a, it gets converted to an electrical signal on a switch. The switch reads the information and goes, okay, I got to go to GPU3212 and then outputs it optically to that other GPU. And that loop burns an enormous amount of power. It's adding latency and driving up the cost of these data center buildouts. And Google, who used to work for, has been quietly working On a solution. They've been at this for 10 years in production now using a. Avoiding that optical, electrical, optical conversion and just sticking to purely optical. And the way they do it is they have an array of mirrors instead of this electrical thing, they have an array of mirrors that's just using them as switches. Right. The GPU 17 goes to GPU 320 and. And they just move these two mirrors or array of mirrors to get the signal from one GPU to another. And they've been able to show that they're able to save 40% of the power in data centers as compared to what Nvidia, for example, does or any of the other data centers. Yeah, and this is such a big news that last quarter Google announced a partnership in using their, what they call a tpu. Google calls it a tpu, winning a, winning a contract with Meta that ended up costing 220 billion in, in market value to, to Nvidia as a result of this announcement. And these little mirrors are what are creating all of that buzz. And then Google has been also very open about the limitations they have with their current switch. And they have this Apollo white paper where they say this is great for this generation, but what we need is more force mirrors that have more force mirrors that have more articulation, better stability and better manufacturability. And it's like, well, we've been working on that for years now. And I'll say that we now have two, I'll say multiple trillion dollar hyperscalers going, hey, Omnitron, do you think you could design a mirror array for us? So we're actually, you know, they've given us an rfi, we've responded to it, we've done a design and fabricating now an array of mirrors for this particular application. And that's been really exciting. That is super exciting.

This is your second Curtis Ray moment to a certain regard of where they said, whoa, why didn't I think of that?

Yeah, yeah.

Here you are with that opportunity. Can you just explain a little bit maybe when you're talking about, you know, pointing two mirrors at each other every, I don't know, 10 milliseconds or something. I mean, these is. This is a lot of cycle time in a sense of moving the mirror, right? I'm guessing, yeah. Which is why down to your, your chip level of needing to take hundreds of thousands, if not millions of cycles on this chip or can you explain that a little bit? Just kind of orders of magnitude for us.

Yeah, yeah. So, yeah, there's a few Things there's. So I'll reference what Google is publicly acknowledging. All of their data centers use these switches. It's a box that has 100 or so channels, so a couple hundred mirrors that are, that are there and some of these mirrors are, you know, switching every second or so. Others are sitting there for months just, you know, locked into this channel. Because some of these large language model computations, they just have to sit there and, and compute. Right? So you have different mirrors doing different things all, all throughout there. So not only do you have to have a mirror that or a design that you know, sits there for a long period of time, but you have to have others that are able to, to dynamically move. And you know, they've been doing this for over a decade and they've got tens of thousands of these boxes inside of their data center. So it's, it's, it's trying to capture like how many times what's the cycle life. You're talking about billions of switches that these things need to be reliable for. And that is something that, that they're really concerned about for their next generation. One in fact one, one of the things that they've acknowledged in their white paper is, is right now their mirrors have issues with stability, especially short term. Like when they, when they move, they have two mirrors for example, that are right next to each other. Using my hands as mirrors. And it takes like 300 volts to move it right. And the one right next to it will feel the field effect of that voltage and also move. And they're like, I lose stability when I have this happen and I need help solving that. Right. And that's something that's unique to our iPad. Uh, but that's, that's some of the challenges that they're seeing today. And, and what we look to address.

Amazing. Wow, so many use cases here is, let's talk a little bit about like I guess maybe alongside this, the history of MEMS a little bit because I feel like MEMS has also been waiting for its moment in a few different industries. Is this also a bit of like the MEMS coming out in its next generation big application here?

You think I, I gotta, I gotta start here. I had, I had not just one investor, but multiple investors. Say MEMS is a four letter word, right. And they don't mean like, you know, they mean that it's a bad word, right. When it comes to the VC ecosystem or investment ecosystem because it's expensive. But when you're building physical things it gets expensive quickly. We're Going through that right now as we're, we're churning through wafers to improve our, our process throughput in our yield. And that just takes money, right? There's no, there's no shortcutting that. There's no like simulation that, that gets you an improvement in yield. You have to actually churn through that. And that's, that's hard. And, and you know, as a result of that, there's been little investment relative to everything else in, in mems. In fact, the, the technology node that Google is using is over 15 years old. And there really hasn't been any great technology node in MEMS in up until us. But what drives that? It's not, it's not the drive for innovation, it is addressing a real problem. And I think that is the catalyst that is catapulting this industry today. It's like, oh, I'm addressing a real pain point. And it justifies this capital infusion to make this happen. But that is really what is getting more excitement in this industry. Because, yeah, I've been at these conferences for years. My co founder has been in these conferences for decades and he's like, I've been hearing the same thing year after year after year. No one's been like promoting something or doing something new. But there also hasn't been like that market pool. Like this is a critical problem that addresses this huge market opportunity. We're at that inflection point now.

Amazing. Does that mean though, you have to bring the whole ecosystem with you? Do you need a tsmc, like, you know, revolution in terms of manufacturing for MEMS to work at this scale? And materials have to. Any new materials beyond silicon that needs to come with it alongside you or what's that like, you think?

Yeah, yeah. So, so we did develop this process at the universities and we've been able to successfully transfer it to a large, the largest MEMS foundry in the world. It's Silex, it's public information. It's bigger than TSMC when it comes to mems. And we've had to transfer that process there. There's no new tools. It's a sequence. We've changed order on how things are done. That's one unique piece of our ip. And we've actually reduced the number of process steps that for example, Google would be using for, for their mirrors and that, that helps a lot in terms of being able to get this to market quickly. But yes, you're right, you have to have a partner that can help you bring this to volume in our industry is Silex TSMC does it as well and we could work with them, but we're starting with Silex first.

Oh sure, yeah. I meant more as an example of like they're the face of that brought chips, you know.

Yes, yes.

And now we need I guess Silex to, to move at that speed as well. What about in terms of. So there's a lot of things changing all the same time as you mentioned, a little bit of that is also the supply chain itself. Bringing things, maybe reshoring, bringing things back. Do you see that in the areas that you are in, are you being pulled to look at, you know, an American based supply chain versus others or where is that going to lead you in terms of mass production here?

We have combed through the US to try to find a facility that could build our chips at the volume that we needed. There still is today no capacity to achieve the volume that we need. And when it comes to building something for our customers in the automotive industry and now going into this, this other market and data centers, the first thing they ask is like, can you achieve that volume that we need? Right. And I can't bring them a prototype facility on the table to make it happen. Right. And there are prototype or smaller scale fabs here in the US but to reach the volumes of thousands of wafers per month. Can't, can't do it here in the US Yet. Now there are, there is one MEMS foundry that is, is, I can't say who one. I don't have the permission to. But there is one that is, is ramping up. There is one that is working on, on bringing that here to the US and they've actually got some money from the CHIPS act and, and very supportive of, of them bringing that, that here. Our customers don't care, you know, as long as it's not made in China. They honestly they're like, they're like, and, and the semiconductor market is global inherently, you know, for better or for worse. Like you're building your chips here, you're, you're packaging over in Malaysia and then you're shipping them over there for, for assembly and then shipping them over there to the cars. And it's, it's, it's always been like this, this global supply chain. But that also brings up the geopolitical issues that come along with it, as you mentioned.

Yeah, well you're, so we're talking two huge industries already. Auto AI data centers. Are there other applications we should be talking about that you're already looking into? Like where where is this headed? What's next? Or is this. This is plenty. Multi trillion industries.

I have enough. Yeah, I have enough on my plate. But that does bring up like when I started, when we started Omnitron, I had no idea I was going to go into data centers. That is something beautiful about working on, on chip technology is it has so many different applications, right? When, when the transistor was developed, it was to replace vacuum tubes, right? Amplifiers, analog amplifiers. And they never thought like could use a. I could use this as a, as a binary switch and change the world, right? So, so there's a lot, I don't know, a lot of don't know where, where this could go. But it is exciting to, to be here at this, this juncture.

100% and I feel like there's going to be deep dives into both these industries. You're already seeing it from an autonomous everything system of just now needing to go underwater into space and everything in between, right. Or into other environments that you know, we suddenly are realizing humanoid robots could do things for humans that we hadn't. We didn't think about before as well too. So that's just. The market is just gonna, you know, continue to expand from that regardless. What are the things, what are still the tech developments you think, within MEMS and you guys looking at Omnitron that, you know, still need to be solved to continue to meet these needs.

I think we're just at the beginning of what AI is capable of doing. You know, it's really gotten good at reading characters, but reading images, video and understanding the real world, there's just so much growth that is needed there. You need much better sensors to be able to digest what's going on in the world and then process that data effectively and then use it in a meaningful manner. Man, we're just scratching the surface, right? I don't know if you've seen some of these videos where you have people wearing these glasses that have cameras that are recording humans doing activities so that robots can learn and adopt those systems. We've got such a vast new market. It's really exciting. And that's just on like the digestion of information, right? There's still a lot of innovation that's needed processing, right? We've got this three pound brain that's doing a heck of a lot more than any data center, right? And the efficiency gains there that this thing is 10 watts compared to 100 megawatt data center. There's just so much innovation to still be had to bring that to something that's reasonable. That's exciting. And then there's the humanoid robot aspect. Right. Or these robots that are going to be interfacing with us on a daily basis. A lot of innovation is happening there and a lot of innovation is still needed on efficiencies, even like battery energy.

Right.

There's, there's so much room for, for growth here. It's, it's, it's really exciting. So I don't think we're anywhere near a limit yet. We're, we're just getting started and that's, that to me is what's exciting and, and we'll just be part of that journey. Right. I think there's a lot of opportunities in so many places. I just want to make sure that we're, we're moving fast enough to keep up 100%.

Yeah. I just heard a, I think someone told me a stat about the human brain related to what you just said there, that if you were to create, it uses the equivalent energy to power the entire city of Dallas for that kind of compute, I don't know if that's anywhere close on the order of magnitude, but just thinking through that to recreate the brain's capabilities to use the power to power Dallas, that's insane. We got a lot of work to do to compress thing down in that regard. And I actually really love the way you kind of, you know, did the biomimicry kind of analysis to where we're all headed there. How much do you think about that? How much do you think about the, the biomimicry copying in biology that already exists for what needs to come in the future?

I, I was, I was fortunate really. Early in my career, I had a great mentor. I'll share a quick story. We were, we were dealing with an issue with our radar system. We, we're trying to detect these cars on the road that, that, that weren't moving. Radar is really good at seeing things that move. The Doppler effect really amplifies the signal over, over the noise. But when it comes to things that are static, even like a car that's parked versus a car that's moving, it'll see the car that but not the car that's parked. And, and we needed to solve that for our customer. And my mentor had the answer, but he didn't give me the answer. He's like, I want you to read this white paper on how our eyes dither. Like, that's weird. So our eyes do this like little tiny vibration all the time. They're Actually doing that even now as we speak, even if you're just staring at something, it's dithering. And as I was reading it, it's like it's actually used to help extract information from things that aren't moving. Right. Our eyes are really good at detecting motion, like radar, not very good at things that are static. So in order to help see that so you don't bump into things and hurt yourself, it creates the motion. Right. So in reading this I was like, oh, maybe I should turn on dithering. Simple aha moment. But then, then I looked at the images of before turning it on and after. And then I could. This, this car that was parked, I was able to see it and, and, and that was to me the aha moment going wow. The answers are there. The answers are out there in nature. They've already been solved. We just have to figure out how to understand it. And, and, and, and that's been a huge influence honestly in my career on, on trying to find inspiration to, to some of the challenges that we're facing today.

I love that bit of first principles in that regard.

Yeah.

As well too. Right. Related the kind of both the energy and the first principles piece. One last question. We actually, actually pulled some of the LPs who originally invested with us way back. Proceed round for some questions. We got one question around the ambient energy piece. Could this. Could. Could mems and specifically the Omnitron system ever have a perpetual motion kind of piece to actually absorb and reuse energy? Because we know right. About energy loss and moving, moving parts, moving pieces compared to static parts. Where does that play into this? Where does the kind of the energy creation loss and, and harvesting play into what you guys are building?

Yeah. So the short answer is yes for energy harvesting. In fact a simple. They are used. There are applications where, where this could be used as a energy harvester. It's kind of like turning an accelerator accelerometer in reverse. So what it does, what you could do is it's literally a mass on a spring. And as, as the mass moves from acceleration or even vibration even talking like it can pick up that energy and harvest it. And it's all a matter of the mass that you make, the amount of articulation that you get and the spring that you get to, to then harvest that energy to use it to charge a battery, charge a phone, whatever those, those are. Or, or you know, if you have a fan running it can capture some of those vibrations and use it. That's, that's a known use case. And yes, our technology could, could, could use, could be used for that. What and this, this may be a good lead in this is it has, there has to be enough of a pain in the market for us to go work on that project or on, on that, on that concept.

Makes sense. Okay, let's zoom out. Real big 10, 20 years, which I know is so hard to imagine right now. But, but where do you think success looks like if we get this right, right. If we suddenly. Not suddenly, if all this compounding the work you've been doing for 10, 15 years, the kind of compute, the kind of energy we're talking about, what does success look like in this regard when we look back here?

I like working on a company that focuses on first principles to bring innovation and also not just on the technology, but also addressing a pain point in the market. As long as we keep focusing on that, we are set up for success decades into the future. The challenge is making sure that the company stays on that ethos, that stays focused on that and doesn't lose that identity as we grow and scale. And that's going to be hard. I was talking to a co worker recently finished the Nvidia Way. How has Jensen been able to keep that energy and trajectory through, through these generations, decades? And it is like don't lose what, what your core thesis is. Even as the market adapts and change, because it will. Right. There's going to be more demand. What is, what is really important now won't matter here. Right. He's not really doing graphics processing units. Right. So that ability to adapt I think is what's going to make us really successful into the future. And all based on these, these, these first principles, like we built a pretty badass little machine. So, so it's got a lot of, lot of different applications. One that's one that is, I know nothing about, I'll be candid but I find fascinating is on, on biomedical applications. These could be used for little pumps, little machines that swim through your body to, to open up valves or, or address alien something that's ailing you. But what can you do now that you have so much more articulation? What can, what can be realized that, that we couldn't realize before? I have no idea but that, that that's curiosity that I want to explore more later down.

Okay. And for listeners, so many, we have so many who are, you know, tech workers at startups and at big tech. You started in maybe bigger tech, then went, joined startup, then joined big tech through acquisitions. There's so Many different paths towards this. How would you give someone advice to like, when to know to take the leap and what to be taking the leap on when we're in the middle of this. AI gold rush, if you will.

That's a, that's a very personal question and I can only speak from, from my experience. And you know, it, it reminds me a little bit of the story of tsmc, the founder, not screaming, but I was telling my, the companies that I was advising, working for, we need to focus on that. We need to work on this problem. This will change this industry. This will make this thing reliable. I wasn't getting any response. I wasn't getting any, like, support. So that frustration was the catalyst to go, you know what? I'm just going to do it myself. And there's, I, I do have an entrepreneurial spirit. Right? There's, there's, you have to have this, like, yeah, I, I can, I can do this myself. It is a hard journey. It will be challenging. You're going to get a lot of no's. You're gonna let a lot of people think that you're crazy. And that's gonna hurt. It hurt me. It was, it was not easy. It still isn't easy. But I, I, if you're thinking about this, I, I encourage it, do it. And like, I don't feel like I'm working today. I'm like, I'll go on, I'll try to take a vacation and like, why I'm having a blast. So highly, highly encourage it to anyone that's, that's thinking about doing this. There's never a good time. There's never a right moment. You just gotta jump.

Yeah, you're right. You, there is no good time. Just start. Because the compound, learn the learning compounds.

So it does 100.

The sooner, the more the learning happens. And back to kind of what you said. Solve the problem. I mean, you gotta be solving problems that you want and your customer is dying for, right? And you will be forced into a deluge of learning, that's for sure.

Yeah, Peter Thiel put it well, and others have, have, have echoed this sentiment. And it took me a while to kind of really digest it. And so I'll say it here for, for that context, make a painkiller, not a vitamin. Like, what the heck does that mean? Like when I, when I'm, when I'm solving a problem that my customer is frustrated by, they'll want to work with the startup. Nobody. So, I mean, I'll be candid. Like when I was working at big tech and, and, and other, you know, the Navy, you don't work with a startup like your startup has to have something so compelling, so frustrated that they're working on that they're willing to, you know, put in capital into this and, and, and, and make an effort into that because they're just so busy with so many other things. Like you really have to focus on what is the painkiller that I'm, I'm providing vitamins. They're nice, but they're not, not essential.

Right. Well, speaking of essential foods that we digest, this being the tacos and tech podcast, tacos are essential. They are not a nice to have. They're part of our daily life. We're here in Southern California and we need our tacos. So we, the audience has to know, they need to know where is Eric's favorite taco spot?

I'm going to go in a slightly different direction because I was in San Diego, I lived there for seven years and there's a place I'm very fond of. Not a taco place, although they, there's. I couldn't, I can recommend a taco place, but I'm not going to Rocky Point Pub. Best burgers I've ever had.

Oh yeah.

Still. And I've looked everywhere. I've been, I've been trying to find them simple burger, patty, lettuce, tomato, mayonnaise, mustard, ketchup in a bun. And it's just like, it's just phenomenal. You get a beer. Well, at the time you got like a beer, fries and a cheeseburger for like 12 bucks. Now it's probably like 25, who knows? But still like the, the price point is phenomenal compared to the the market and great, just great burger. So yeah, I'm going. My, my taco is a burger.

How non consensus of you as a, as a Deep Tech founder. Thank you. Rockies is legendary. What? We've never had it mentioned here because everyone's thinking inside this box and you're thinking out here, which we really appreciate. Well, thanks Eric. This has been fantastic. First of all, love the conversations you're intersecting. Deep engineering, real world pain, massive future impact. And that's exactly what you and the team at Omitron are working on. And thanks for spending the time with us. But even more thanks for spending the time on changing the world in the way that you guys are doing it. So I can't wait to continue following along journey and we'd love to have you back at the next next great story inflection point here. And keep it up. Thanks, Eric.

Yeah, thank you. Thank you for having me.

Of course.

By.

Tacos, tacos in tech, you know we got Tacos in tech is on smash for the green guacamole Keep accountants counting beans Contador Quentan frijoles found us trying to raise some queso la vida San Diego school of rappy rapping Paella, tortillas round jalapenos in these modern times a craft beer and coffee my coating yoga posse fold the body origami we be surfing coronado eating tacos El gordo, the taco stand in puesto lucha libre and sombrero Got a shout out to cost angels by the ocean watch the sunset la Hoya shores Entrepreneurs trying to get funded when San Diego Startup week is popping in the summer Oceanside, the Tijuana St. Be full of hustlers from couch surfing to penthouses right on the beach. Tech leaders who believe stars are within reach we interview and founders who be shooting for the Moon. Hashtag SD TechOS with your host, Neil Bloom.

Come on.

Startups get the party. Startups get the party started. Startups get the party. Startups get the party started. Startups get the party. Startups get the party started.

You want tech in tacos, tacos in tech, you know we got it.

Hey.

Startups get the party. Startups get the party started. Startups get the party. Startups get the party started. Startups get the party. Startups get the party started.

You want checking tacos, tacos in Tex. You know we got it.

Hey, Sam.