Shownotes
Summary
Roger Espasa, the esteemed founder and CEO of Semi Dynamics, engaged in a profound discourse on the intricacies of innovation within the realm of high-performance computing and machine learning. Central to our conversation was the notion that the journey of entrepreneurship is fraught with challenges, yet it is the persistent pursuit of completion and excellence that defines success. Espasa articulated his personal motivations for founding Semi Dynamics, stemming from experiences in larger corporations where pivotal projects were often abandoned, leaving him with a profound sense of frustration. He emphasized the significance of perseverance and the relentless drive to see initiatives through to fruition, akin to reaching the summit of a mountain after a challenging ascent. As we delved deeper, Espasa elucidated how the RISC V revolution has provided a fertile ground for innovation and competition, contrasting the restrictive nature of established architectures. This episode serves as a compelling narrative for aspiring entrepreneurs, highlighting the importance of resilience and the willingness to chart one’s own course in the ever-evolving landscape of technology.
Notes
Roger Espasa, founder and CEO of Semi Dynamics, shares his journey from working in major tech companies to starting his own venture, emphasizing the importance of control over one's professional destiny. He highlights his frustration with project cancellations at large firms, which fueled his decision to create a startup focused on high-performance RISC-V cores. The conversation delves into the challenges of building a company in Europe, particularly regarding venture capital access and recruiting talent in a less mature ecosystem compared to Silicon Valley. Roger also discusses the impact of rising AI model sizes on chip design and the necessity for efficient memory bandwidth to support these developments. Ultimately, he reflects on the rewarding experience of seeing customer interest and the fulfillment that comes from pursuing innovative technology solutions.
Roger Espasa's dialogue with Jothy Rosenberg unveils the intricate tapestry of his journey as the founder and CEO of Semi Dynamics, a Barcelona-based innovator within the RISC-V ecosystem. Espasa’s narrative is one of resilience against adversity, characterized by a series of disheartening project cancellations in his previous roles at large technology firms. This pivotal moment catalyzed his resolve to establish a company where he could steer the direction of his work without the looming specter of arbitrary corporate decisions. His story serves as an inspiring testament to the power of entrepreneurship in reclaiming agency over one's professional destiny.
As the conversation unfolds, Espasa delves into the technical nuances of RISC-V architecture, elucidating its significance in high-performance computing and machine learning realms. He articulates a vision where data throughput to processing units becomes the linchpin for unlocking the full potential of AI applications. This technical discourse is complemented by a broader commentary on the evolving landscape of semiconductor technology, where efficiency and performance must align to meet the burgeoning demands of modern computation.
The discussion further accentuates the distinct challenges faced by European startups, particularly in securing investment and acquiring talent in a region where the silicon innovation ecosystem is still maturing. Espasa’s insights into the venture capital environment in Europe illuminate the disparities that exist compared to more established markets, underscoring the need for a supportive infrastructure to foster innovation in the semiconductor domain.
Takeaways
- The journey of establishing a startup is laden with challenges, particularly regarding funding and cash flow management.
- Roger Espasa's motivation to create Semi Dynamics stemmed from a desire for autonomy and control over his work.
- The evolution of technology, especially in the field of AI, necessitates constant adaptation and innovation from startups.
- The RISC V architecture has revolutionized the industry, providing a cost-effective alternative to proprietary architectures like ARM.
- Effective data handling and memory management are critical in developing processors that meet the demands of modern AI applications.
- Building a successful startup requires resilience, continuous learning, and an unwavering commitment to one's vision.
- Roger's LinkedIn
- Semidynamics - The world’s only fully customisable 64-bit RISC-V processor IP
- Jothy's TEDx talk - On why those with disabilities tend to overachieve
- The Who Says I Can't Foundation - Jothy's non-profit that helps disabled kids get back into a sport they love to restore their self-esteem
- Our YouTube channel
- Designing Successful Startups — Jothy Rosenberg | Author | Speaker | Podcaster - Site with all podcasts
- Tech Startup Toolkit book
- Please leave us a review
Transcripts
Foreign.
Speaker B:I'm Jonathi Rosenberg, the host of Designing Successful Startups, where today's guest is Roger Espassa.
Speaker A:Roger, finishing things, right?
Speaker A:So, so this thing, this company is not finished.
Speaker A:We're not done.
Speaker A:We're not, you know, making billions or millions or we're not even, we're not there yet.
Speaker A:Right?
Speaker A:We're doing great, but we're not there yet.
Speaker A:There's a ton of work still in front of us.
Speaker A:And, and know if you go, if you like hiking, you tend to enjoy getting to the peak of the mountain.
Speaker A:Do you go hiking and midway, you say, oh, it's been a great day, it's enough we go down.
Speaker B:Roger Espassa is the founder and CEO of Semi Dynamics, a Barcelona based IP supplier of two RISC V cores targeted at high performance computing and machine learning.
Speaker B:Prior to Semi Dynamics, Roger was at Broadcom working on an arm V8 wide out of order core.
Speaker B:And before that Roger worked at intel developing a vector extension for the X86.
Speaker B: Between: Speaker B:Roger got his PhD from the Polytechnic University of Catalonia in 97, has published over 40 peer reviewed papers on high performance process processor architectures and is the inventor on nine path.
Speaker B:Welcome to the show Roger.
Speaker B:I'm really glad to have you here.
Speaker B:We've been looking forward to this.
Speaker A:Pleasure being here Jothi and thank you for inviting me.
Speaker A:Ads.
Speaker A:I'm looking forward to this conversation.
Speaker B:Well, sure.
Speaker B:Let's start with a little bit of context for people.
Speaker B:Where are you originally from and where do you live now?
Speaker A:Well, I'm originally from Barcelona.
Speaker A:I did my studies there and PhD in Barcelona and then I did spend some time in the U.S.
Speaker A:milpitas, California and Shrewsbury, Massachusetts.
Speaker A:But then decided I really missed the good weather and went back home.
Speaker B:You know, it's funny because the two major places that I've lived are on the, on the other side of the, of the Santa Cruz Mountains from that you probably looked at sometimes from Milpitas.
Speaker B:We lived over in Santa Cruz and then many years later we moved out to Boston, where I've been for 27 years, a suburb of Boston.
Speaker B:And, and, and you weren't far from here then, right?
Speaker B:Yeah.
Speaker A:That's funny.
Speaker A:Yeah.
Speaker A:I remember going up that road that goes up the Santa Cruz Mountains, Highway 17.
Speaker B:Very famous.
Speaker B:And, and when I was out there I, I bought a BMW convertible because I wanted a, I wanted a really good car that could handle that road and, and when there wasn't a lot of traffic, I wanted to do it fast.
Speaker A:I had to live on a 400 Ford Mustang, so you can imagine.
Speaker A:And the eight.
Speaker A:But we did go over Highway 17, so that was, that was good.
Speaker A:And three in the car, so that was fun.
Speaker B:Yeah.
Speaker B:Well, good.
Speaker B:So you are CEO, founder and CEO of Semi Dynamics.
Speaker B:And it is a RISC V producer of IP that people.
Speaker B:So someone can basically license your IP and put it into their system on chip.
Speaker B:That's the exact, the idea.
Speaker B:And so tell us a little bit about what made you start Semi Dynamics, what you felt the need was, you know, the sort of the journey that got you to, to that creation.
Speaker A:Well, the need was more personal than anything else.
Speaker A:Right.
Speaker A:I started working in Shrewsbury for the digital people and they became Compaq and then they became HP and then HP sold the division to intel.
Speaker A: years from: Speaker A:But you know, things happen.
Speaker A:And our team got canceled and I was not happy.
Speaker A:So I moved to Broadcom where I designed an ARM out of order CPU for Broadcom also for a couple years.
Speaker A:But then Broadcom got acquired by Avago and again some VP in the west coast decided that my project was canceled, wasn't my project, but the project I was working on was canceled exactly at 3am when I was sleeping.
Speaker A:So I wasn't even in the meeting.
Speaker A:As you know, next morning you get an a bad phone call or email and I said enough.
Speaker A:Working for big multinationals where VPs in the west coast make decisions where I can't affect those.
Speaker A:So this time I'm going to do try something different.
Speaker A:I'm going to set up a company and that's really the motivation reason like I wanted to do, you know, because I had offers to join yet again a big US multinational where I'm sure I would have had lots of fun working on a project.
Speaker A:But it was frustrating to me.
Speaker A:You work this silicon project for so long and you invest three years of your time and you're half a year away from tape out and boom, all out of the window, right?
Speaker A:And nothing is left, right?
Speaker A:The scripts, the work, the, the processes, everything you've created to make that product a success, it just gone, right?
Speaker A:And that, that was very frustrating for me.
Speaker A:And that's, that's the real reason I could have joined again one of these big companies and do for the third time, the third thing.
Speaker A:And I think I would have guessed correctly that, you know, the project might have been canceled again.
Speaker A:So I said no more.
Speaker A:Now I'm trying the startup thing and that's what I did.
Speaker B:It seems to me anyway, and this could be just the over optimistic view that instead of the work you did, of the out of order work and also the Vector extensions for x86 and for tarantula on the alpha 3, that, that what you're doing now is you've got what I think when, and we've looked at a lot of RISC V producers, what I think is probably the highest performance and probably highest quality RISC V cores available in the, on the market.
Speaker A:Well, thank you for those words.
Speaker A:I, I agree to it, I do agree with that.
Speaker B:But you, is it true that you took the work that you were doing and seemed wasted because they canceled it and didn't that inform you and end up giving you the wherewithal to do what you've done with your RISC V core?
Speaker A:Absolutely.
Speaker A:And happy to go technical here.
Speaker A:I mean one, both at intel when we were doing Larrabee, which became the Knights Landing, sorry, Knights Ferry, Knights Corner, Knights Landing line of processors and finally Knights Hill, which I think never shipped, I can't remember.
Speaker A:The thing I've been learning over all these years is growing a passion or obsession figure word is like how do you keep these big vector units and now Tensor units fed with data, right?
Speaker A:So how do you make sure that data is arriving from memory?
Speaker A:And I am absolutely convinced that's the key and only problem in this future computing, especially with AI, meaning everything else has been done, has been solved.
Speaker A:It's engineering, it's difficult, it's fun to build, but we all know how to do it.
Speaker A:Let me say an out of order core, a lot of quite a few people know how to do it.
Speaker A:But combining that with how do we make sure the real compute engines are fed with data is our specialty and I want to say is the learning.
Speaker A:I took from all those previous projects that I worked on, which I really enjoyed.
Speaker A:All of them combining pieces from each project, the learning or the goal was how do we keep these compute units busy.
Speaker A:Right.
Speaker A:And that's the key because not to sound like a peach, but you can buy GPUs from very famous companies, very, very expensive and you get out of them a fraction of what you paid for.
Speaker A:Right.
Speaker A:So that's not fair to the consumer.
Speaker A:Certainly it's great products.
Speaker A:They work, you know, they're going to work well.
Speaker A:But you pay for eggs and you get 15% of X.
Speaker A:Right.
Speaker A:So that's not what we want, right?
Speaker A:So we want to help people that are putting our technology in their silicon to not have that effect.
Speaker A:So we're helping them reduce costs and we're helping their customer reduce costs.
Speaker B:I think I'm going to make a comment to the audience that the GPU cores you're, you're referring to and being expensive, last time I checked from that vendor, they're $30,000, which is, you just have to pause.
Speaker B:You're just like, oh my God.
Speaker A:Exactly.
Speaker A:And if you think, you know, since this free flowing conversation, we're all in agreement.
Speaker A:I think that AI is the revolution.
Speaker A:That AI is here and coming.
Speaker A:I think it's starting, but there's a long, long and impressive road ahead of us.
Speaker A:And for AI to really permeate everything we do.
Speaker A:And some people hate that idea, but it's going to happen whether you like it or not.
Speaker A:The prices have to come down.
Speaker A:We can't be paying the number, you said $30,000 to run a personal AI assistant.
Speaker A:That's just not in the hands of many people.
Speaker A:Right.
Speaker A:It has to come down to.
Speaker A:And in order to do that you need technology that will do the same thing or better than we're doing today at a much lower cost.
Speaker B:Was it really fortuitous as you were thinking about starting Semi Dynamics that the RISC V revolution itself was happening?
Speaker B:Because if it didn't, if it wasn't happening, you would have had to do something like get a license from arm.
Speaker B:But you know, to get a license from ARM that allowed you to change their core costs about.
Speaker B:Well, I don't, I don't know anymore.
Speaker B:But it was like millions of dollars, Millions of dollars to get an architecture license.
Speaker A:No, you're right, it, it was, how would I say, I can't remember was.
Speaker A:I can't remember.
Speaker A:I, I think I went to the second RISC V summit.
Speaker A:This was 14 or 15.
Speaker A:I apologize to the audience for my bad memory.
Speaker A:But I remember going there because, you know, we had started the Broadcom thing, but I, I usually go to California and I said, hell, I'm gonna pop there.
Speaker A:And honestly, and allow me the joke.
Speaker A:I expected to come into a small room with three old guys like me, with suspenders, if you like the joke and be just a tiny, tiny thing.
Speaker A:And I walk into the room and first of all, I'm totally impressed.
Speaker A:Like there's hundred, two hundred people.
Speaker A:But not only that, I start looking around and I'm like, oh my God, that's that person from that important company.
Speaker A:That's this other Person, people I knew and people I knew, their face and their name is like, what is happening?
Speaker A:I mean, this thing is happening.
Speaker A:Right?
Speaker A:So I, I was just lucky.
Speaker A:So yes, I was lucky to attend that meeting.
Speaker A:And I came out of that meeting saying ris5 is happening.
Speaker A:I wasn't, I was following it, I was interested, I thought it was a great idea.
Speaker A:But I, without attending that one day session, I would have not thought it's happening.
Speaker A:And it was happening.
Speaker A:When you see the companies represented at that meeting, it was like, wow, these people are nervous.
Speaker A:Or not nervous.
Speaker A:Later, digging more, I discovered this was a time about ARM was raising prices and people were very uncomfortable with ARM being owned by SoftBank and where they were very concerned.
Speaker A:And you know, later we had the episode where Nvidia tried to buy arm.
Speaker A:So that added more uncertainty to the people using arm.
Speaker A:And you know, those things fueled the revolution that RISC V is bringing to the table.
Speaker A:So, yeah, partially luck, yes.
Speaker B:I don't think we met at that meeting, but I was at that meeting too.
Speaker A:Okay, that's bad.
Speaker A:I didn't know you then, but yeah.
Speaker A:Was it Google?
Speaker A:I think Google had the room.
Speaker B:Yes, it was at Google.
Speaker B:And in fact, what happened at that meeting was the folks from NXP had decided that they wanted to have another core other than ARM that they could start to work with, but they needed security.
Speaker B:So they asked the head of the RISC V Foundation if there was any work being done on security and he pointed them to me.
Speaker B:So they, they found, they found me.
Speaker B:And that was why they became.
Speaker B:They actually were my Lighthouse customer.
Speaker B:And I don't think I've ever heard of somebody being sought out by their future Lighthouse customer.
Speaker B:It was really good.
Speaker A:Yeah, yeah, no, no, that was an amazing meeting.
Speaker A:I remember what it was.
Speaker A:Yeah.
Speaker B:And then right after that meeting, I joined the board of the RISC V Foundation.
Speaker B:So for the next, I guess two years, that was the term of a board.
Speaker A:Yeah.
Speaker A:It took me a while to get back to work the Risk five, but that was the obvious choice.
Speaker A:When we, I started Semi Dynamics, the first thing we did was a contract work.
Speaker A:Right.
Speaker A:So I met, well, I met Dave Ditzel, a famous founder from the Transmed days in the 90s at Intel.
Speaker A:And when I left intel and he had already left intel, he said, hey, I have a startup called Esperanto Technologies.
Speaker A:You know, do you want to work for me?
Speaker A:Right.
Speaker A:I said, I want to work with you, but not for you, because I've already worked two times for American companies.
Speaker A:So that's why I set up Semi Dynamics.
Speaker A:Right, because at the very beginning we just provided, let's say, engineering services.
Speaker A:Right.
Speaker A:We designed the CPU for Esperanto.
Speaker A:It's owned by them, it's their technology.
Speaker A:But we did it in Barcelona.
Speaker A:So and it was a fantastic project.
Speaker A:And at that project, you know, this, this really was close.
Speaker A:It's like, okay, we got to do RISC V.
Speaker A:There's no question.
Speaker A:I mean we're not doing any, any other crazy stuff or inventing our own ISO.
Speaker A:This is completely pointless with the point that where RISC V was.
Speaker A:And that's, that's why Esperanto is also a RISC V implementation and member of RISC 5 and so on.
Speaker B:So have there been sort of extra challenges in, in terms of getting a startup going because you did it in Barcelona versus, you know, if you'd picked Boston or Silicon Valley.
Speaker A:Absolutely.
Speaker A:I mean the, the vc, the VC crowd in California and also Boston and Austin and other region of us are understand Silicon, know what it takes to get a silicon project off the ground, know the time it takes and know the amount of cash it takes.
Speaker A:So, so you know, they'll fund you based on whether they think you have a good idea or not.
Speaker A:But they're not going to be initially scared of what number did you just say you need?
Speaker A:What?
Speaker A:And that's what I get from the European VCs, which is any number, about 5 million.
Speaker A:They're already out the door.
Speaker A:Right.
Speaker A:So.
Speaker A:And that's unfair.
Speaker A:It's a too general comment.
Speaker A:Of course there's a few that have done a few silicon investments, but it's only a few.
Speaker A:So you only have five doors to go knock.
Speaker A:And if those five don't agree that your idea is good, then that's it, you're done.
Speaker A:Right.
Speaker A:So it is certainly more challenging in Europe.
Speaker A:Right.
Speaker A:Also the amount of technology experience that the ecosystem in California and Boston and Austin and North Carolina and Portland and many other places have means you can recruit people.
Speaker A:Frankly, there's not much happening in Barcelona or Spain and a little bit happening in Europe.
Speaker A:So recruiting is very complicated.
Speaker B:So you are building a very high end core.
Speaker B:It's not, not, it's not going to be in my washer and dryer.
Speaker B:It's not an Iot type of.
Speaker B:You are building a very high end core.
Speaker B:It's not, not, it's not going to be in my washer and dryer.
Speaker B:It's not an IoT type of or Internet of things type of part, is it?
Speaker B:Mostly.
Speaker B:Are you thinking it's Mostly for like data centers or what sort of, you know, what's the situation you expect your part to be used in?
Speaker B:Most.
Speaker A:Yeah, I mean, just for the audience.
Speaker A:The word high power or high end core varies across people.
Speaker A:Right.
Speaker A:So certain people would say we're not even high end because there is even higher end.
Speaker A:Right.
Speaker A:But you, you're correct.
Speaker A:I mean, the answer to that question is not going to be very precise.
Speaker A:It's where our customers take us.
Speaker A:We, we can go, maybe not a watch, I agree, not a watch on batteries, but we can go onto a small camera where somebody wants to process the feed and run some AI on the camera local before sending the data out.
Speaker A:Typical use case would be, you know, I want to run some AI, is the person in this factory floor has fallen on the floor or is injured.
Speaker A:Right.
Speaker A:Just send out an alert.
Speaker A:There's no point in sending the whole video stream out when the only thing you want is to check that nobody's been injured on the floor of the factory or similar for is there somebody in my house?
Speaker A:Right.
Speaker A:So that would be a low end case where maybe one core with one tensor unit is enough to run those algorithms.
Speaker A:Midway we've had people interested in, okay, we have 5G, we're doing a 5G XG chip and we also want to run AI with the data we're receiving.
Speaker A:And this would be medium, medium performance, medium power.
Speaker A:And then you're correct, you mentioned data center, like edge and data center.
Speaker A:An example I always give is imagine a hotel that wants to have a virtual concierge where you're in your room and you talk to a microphone and say, please set my alarm for 7am Right?
Speaker A:Or please send me a burger 10 minutes.
Speaker A:Right.
Speaker A:And this actually goes from the microphone to the hotel basement.
Speaker A:And down in the hotel basement we have a bunch of big, big compute.
Speaker A:But not, not a data center, but what we call edge AI, which means it's in the data, it's in the basement, it's big, it's bulky, it's noisy.
Speaker A:But this thing is doing all the processing, getting your voice, doing the voice to text translation, then doing text to understanding, and then finally really ordering your burger for 9:00pm, right.
Speaker A:So that would be AGI.
Speaker A:So we really go again, you said it perfectly at the beginning.
Speaker A:It's our customer that says, I want to do this chip for this goal.
Speaker A:And we're there to help them get the AI they want inside their chip.
Speaker A:So they tell us the requirements and we say, yeah, possible or not possible.
Speaker B:Hi, the podcast you are listening to is a companion to my recent book Tech Startup Toolkit how to Launch Strong and Exit Big.
Speaker B:This is the book I wish I'd had as I was founding and running eight startups over 35 years.
Speaker B:I tell the unvarnished truth about what went right and especially about what went wrong.
Speaker B:You could get it from all the usual booksellers.
Speaker B:I hope you like it.
Speaker B:It's a true labor of love.
Speaker B:Now back to the show.
Speaker B:Speaking of AI, the, the models for AI are becoming enormous and they're kind of breaking the assumptions or this standard amount of memory that chips would have.
Speaker B:So have you done anything about that that's not secret that you could talk about that, you know, these multi gigabyte, maybe even you know, larger than, than that models that have been created through a machine learning process.
Speaker B:And now for the, for the, you know, the operational part of AI which is called an inference engine to you know, uses that model to make decisions based on some, some inputs.
Speaker B:That model has to be stored somewhere that's easy to access and isn't going to be have a bottleneck for accessing the data.
Speaker B:Anyway.
Speaker B:I've probably said enough to.
Speaker A:It's an excellent point.
Speaker A:You are 100% correct.
Speaker A:Those models are getting gigantic and I don't want to.
Speaker A:Have we done something about it?
Speaker A:Yes, but the problem is divided between us and our customer.
Speaker A:So it is our customer building the chip and building the board and deciding that they put 3 gigabytes of memory or 20 or half.
Speaker A:That's their decision.
Speaker A:What we help them is we understand this problem very well and we say okay, what model are you running?
Speaker A:We say that memory is not going to be enough.
Speaker A:What bandwidth are you going to put?
Speaker A:Are you going to put really old fashioned LPDDR3 or are you going to put super fancy HBM?
Speaker A:And we help them realize that guys, you're not getting enough bandwidth or enough capacity for what you want to achieve.
Speaker A:Once we do that, what is the part we contribute?
Speaker A:We say, okay, once you tell us the bandwidth you're going to have, what we guarantee or we help them realize is that our technology with this thing called gazillion really is going to use that bandwidth.
Speaker A:They're not going to put, I don't know, 400 gigabytes a second of bandwidth and extract 20 out of it.
Speaker A:We're going to be very, very close to that peak thanks to our technology.
Speaker A:That's the part we contribute and we work with our customer to understand what you just explained, which is happening.
Speaker A:And it is a bit of a let's call it a limiter or a price limiter.
Speaker A:It's not a, it's not that we need to understand it.
Speaker A:It's like, wow, it's expensive.
Speaker A:You want to put one of these big, big models on a watch it' today because the.
Speaker A:Just the memory rams are just bigger than you watch.
Speaker A:Right.
Speaker A:If you count.
Speaker A:I need this number of rams and I put, you put them on the surface.
Speaker A:They're bigger than, you know, my laptop.
Speaker A:So.
Speaker A:So it is a problem.
Speaker A:It's a very interesting problem that, that we will together as an industry will, will be facing and solving.
Speaker B:So from the, from the standpoint of just your company, you know, your startup, what have been some of the scary moments, what have been some of the struggles that you've had with Semi Dynamics?
Speaker A:Well, there's so many that if your audience is future founders, be ready to be scared.
Speaker A:Okay, let's classify them.
Speaker A:I mean, we have been particularly lucky in that we've always been cash flow positive.
Speaker A:So we haven't had one of those moments, oh my God, I can't pay my employees.
Speaker A:And that's.
Speaker A:And we have plenty of cash going forward.
Speaker A:And this is partly because we EU has been quite generous in grants, so we've been lucky in that.
Speaker A:But certainly watching cash flow diminish and cash reserves go down and the anxiety that this generates is important.
Speaker A:Hiring and losing people in the sense that they decide to move on or they weren't fit for the job, you ask, is also a lot of work.
Speaker A:Right.
Speaker A:You invest a tremendous amount of time trying to get people and when they leave or you then realize, oh my God, this is not the guy I needed or the girl, you know, the expertise of that person was one and I was looking for something else.
Speaker A:That's, that's a difficult moment.
Speaker A:And then raising money is complicated.
Speaker A:Right.
Speaker A:There's just a tremendous amount of hours, you know, words, missing slides that you think are going to make all the difference.
Speaker A:And maybe the difference is something else that, that the investor had.
Speaker B:Oh, you just use AI and it'll.
Speaker B:It'll do a perfect job on your slides.
Speaker A:Exactly.
Speaker A:But then it loops on somebody else's speech.
Speaker A:Right.
Speaker A:So, yeah.
Speaker A:So.
Speaker A:And then there's little operational things every day, right.
Speaker A:Every day there's something that there's a fire.
Speaker A:Right.
Speaker A:That you got to put, put, put down one way or another.
Speaker A:Sorry.
Speaker A:Not.
Speaker A:No, I know LA is suffering.
Speaker A:That's a bad example at this point.
Speaker A:But there's always something that's like needed for yesterday or the bank is complaining that you didn't pay something.
Speaker A:There's, you know, it's a never ending.
Speaker A:You never, you really never sleep, right?
Speaker A:You do sleep, but you, you don't sleep peacefully.
Speaker A:So let me say.
Speaker A:So that's true.
Speaker B:That's true.
Speaker B:And you may not have predicted all these things when you were starting it up.
Speaker A:Most of I worked in very, very successful companies where I had all the resources I needed.
Speaker A:And I always complained I needed more, but I actually had all the things I needed.
Speaker A:I had really good expert in the company, lots of money, and all the operational stuff was taken care of.
Speaker A:Of course, then as spoiled engineers, we complained, oh, I don't like why they do this and I don't like why the company does that.
Speaker A:And then when you change sides, so to speak, and you're the one actually having to decide who gets into this door and why is this door locked and why do we have more security or less security, you realize, well, all those things are needed and somebody has to set up that.
Speaker A:That's what part of a successful company is, right?
Speaker A:Because an investor comes and looks at your data room and says, what?
Speaker A:You don't even have a log there.
Speaker A:I'm not investing in this company.
Speaker A:You're not focus on data security.
Speaker A:And you think, oh my God, why didn't we put a lock in the damn door?
Speaker A:Right?
Speaker A:And it's not that you're not focused on security.
Speaker A:It's the God of the nine million things you have to do per day.
Speaker B:So the other thing that you may not have thought about when you were starting is the fact that everybody that starts a startup has to have grit.
Speaker B:And, you know, I ask about grit every one of these episodes.
Speaker B:So I'm asking, obviously you have grit.
Speaker B:You've been able to fight through all these things.
Speaker B:And frankly, by the way, when you said, oh, we've always been cashflow positive, it's like, okay, you haven't really had catastrophic problems.
Speaker A:Correct.
Speaker B:But seriously, tell us where your, you think your grit comes from.
Speaker A:Boy, that's.
Speaker A:I don't know what to say.
Speaker A:I mean, you know, you start this thing and you think, okay, it's not a lot of work, you know, I can manage it.
Speaker A:And you work more and you work more, more, and you work more and you work many hours and your weeks get even longer and then your Saturday disappears, and then your Sunday morning disappears.
Speaker A:You know, I guess I don't know where it comes.
Speaker A:It's a good question.
Speaker A:I never, never had a minute to think about it, but I think it's like you have Invested so much that what is the next reasonable action is to keep pushing.
Speaker A:Right.
Speaker A:I don't see it as an option.
Speaker A:Maybe I should consider it after talking to you, but I don't see it as an option to just say, hey, guys, you know, I'm done to take the keys of the.
Speaker A:Of the house, and I'm going to go fishing in Menorca and the Balearic Islands.
Speaker A:Right.
Speaker A:So I think it comes from that, you know, you started something as an engineer.
Speaker A:You enjoy seeing things working.
Speaker A:You enjoy things working correctly.
Speaker A:And I have to say also that seeing customers interested is the positive side of everything.
Speaker A:Right.
Speaker A:In the end, when a customer looks at your technology and says, I would like to buy it, that's the ultimate satisfaction.
Speaker A:All this work culminates in, okay, I'm helping someone, and they really think my technology, the one we've worked so hard to, not me, the whole team.
Speaker A:It's really useful.
Speaker A:It's how it works.
Speaker A:I guess it's that.
Speaker A:But I'll have to analyze myself tonight and maybe come back to your.
Speaker A:To your podcast.
Speaker A:Give me a better answer.
Speaker B:In a way, it's probably better that I didn't tell you I was going to ask you this ahead of time because it's.
Speaker B:I think it's very useful to just sort of hear what you have to say, kind of, you know, just off the cuff.
Speaker B:Yeah.
Speaker B:I mean, sometimes it comes from how you were raised, sometimes it comes from your experiences.
Speaker B:And, And.
Speaker B:And I wouldn't be surprised if, if some of the things that, you know, you said early on about, well, this project got canceled, and then this one, and then this one, and you kind of threw your hands up in disgust and flew back to Barcelona and said, I'm going to start my own company.
Speaker A:It could be.
Speaker A:And if you go back, I would say, I don't like not finishing things right so.
Speaker A:Or in double positive, I like finishing things right so.
Speaker A:So this thing, this company is not finished.
Speaker A:We're not done.
Speaker A:We're not, you know, making billions or millions or.
Speaker A:We're not even.
Speaker A:We're not there yet.
Speaker A:Right.
Speaker A:We're doing great, but we're not there yet.
Speaker A:There's a ton of work still in front of us.
Speaker A:And, you know, if you go.
Speaker A:If you like hiking, you tend to enjoy getting to the peak of the mountain.
Speaker A:Do you go hiking and midway, you say, oh, it's been a great day.
Speaker A:It's enough.
Speaker A:We go down.
Speaker A:No, you.
Speaker A:You keep walking until you hit the top.
Speaker A:Right.
Speaker A:So I guess it's the same.
Speaker A:But, yeah, it's hard to say.
Speaker B:Yeah, I think, I think that without using these words, I think one of the things you were kind of saying is that you wanted not to have some vice president in control of your destiny.
Speaker B:You wanted to be in control of your destiny.
Speaker B:Now there, there's a downside to that.
Speaker B:It's really hard.
Speaker B:Yes, of course.
Speaker B:But the, for me, even in a startup that ultimately didn't succeed, and I, I don't mean to, you know, imply in any way that you won't, because I, I think you're, you're doing great.
Speaker B:But on some of my startups, that, that, that didn't succeed, I look back and I say that's not a, you know, that's not a disaster because it was a very joyful experience all that time.
Speaker B:And yes, it didn't, it didn't have an exit, but we still had fun because we were in charge of our own destiny and that was by itself a great thing.
Speaker B:And I mean, you know, I'm, as, you know, I, I'm right now trying to sell the IP that we spent so many years developing.
Speaker B:And, and, and it was, you know, and this, these, I, these patents are very, very innovative and powerful.
Speaker B:And.
Speaker B:But it won't be a good exit.
Speaker B:It'll, it'll be, you know, a minimal exit, but it'll still be still okay.
Speaker B:I mean, I look back on the, on the time of developing Dover, and there were some very hard times.
Speaker B:The pandemic was extremely hard for us.
Speaker B:We, we never even were able to raise a Series A because we got caught just, like, at the worst time for trying to do that.
Speaker B:And, and, you know, the whole thing has just taken, you know, too long.
Speaker B:Startup didn't have enough cash to, to withstand the, the time frames of getting our part in somebody's chip and then, you know, to start getting royalties paid.
Speaker B:It was just.
Speaker B:But we'll still look back on it and say that was a good experience.
Speaker A:I would concur.
Speaker A:Right.
Speaker A:I mean, you know, you were asking a little bit like, hey, why did you do this?
Speaker A:It's good to be in control of your destiny.
Speaker A:Or that's not, I don't phrase it like that.
Speaker A:I don't think I'm in control, but I'm in the know.
Speaker A:Right?
Speaker A:So I know what's going to come.
Speaker A:I know what needs to be done, and some things I hope I can solve.
Speaker A:And the day you encounter one thing that you don't solve, then the company is done and that's it.
Speaker A:Right.
Speaker A:But the difference between being in A gigantic company.
Speaker A:And being in your own startup is, you know what's coming and you know where the problems are.
Speaker A:You know what's going to happen.
Speaker A:In big companies, you're told design is cheap and other things are happening in other layers.
Speaker A:And sometimes you can influence a strategy, sometimes you cannot.
Speaker A:And although you see that the strategy, there's something not working with the strategy, you don't have enough visibility to just change it.
Speaker A:Here at least you can't blame anyone if you get the wrong strategy.
Speaker A:There's nobody else to blame but me.
Speaker A:So it's good and bad.
Speaker A:Right.
Speaker A:I actually enjoy it.
Speaker A:Right.
Speaker A:In the sense of if somebody presents a problem, okay, let's sit down.
Speaker A:Let's work one problem at a time.
Speaker A:Let's try to solve it.
Speaker A:Let's try to solve it in the best way for everyone and the best way to keep going forward.
Speaker A:I do enjoy that.
Speaker A:And I have a good friend that always says people think that because they're working at Google, they're not.
Speaker A:Their next day is guaranteed not true.
Speaker A:They can be fired the next day.
Speaker A:They just don't know.
Speaker A:The difference is you as the founder, you know, okay, I'm firing everyone in 10 days.
Speaker A:Right.
Speaker A:And that's not going to happen for us in the next 10 days.
Speaker A:But, and as an employee, you just don't know.
Speaker A:One day you get a message or a meeting or communication and that's it.
Speaker A:Yes, it's nice to have that extra info.
Speaker A:Having more info is always good.
Speaker A:But, but having control of your destiny sounds for me a little bit too arrogant.
Speaker A:There's so many things moving that is not that I control them.
Speaker A:Right.
Speaker A:There's, there's major, major companies moving right and left and in there.
Speaker A:As they move, they move the floor or the waters.
Speaker A:And you are a little boat affected by those waters.
Speaker A:Right.
Speaker B:So, yeah, well, I'll have to come up with it.
Speaker B:I'll think I have to think of a different way to express it, but it's, it's kind of how I think about it, is that as a founder, you're more in control of your destiny than any other work experience I've ever had.
Speaker B:So it's relevant.
Speaker A:Fair enough, fair enough, fair enough, fair enough.
Speaker B:Well, Roger, thank you very much.
Speaker B:I, I think this has been great and, and in fact, I think this is especially great for people that are, that are not yet founder of a startup and better thinking about it because you've done a really nice job of, of comparing and contrasting your experiences before the heartbreak of having your project that you poured your heart into for three or two or three years or whatever it was and have that happen multiple times.
Speaker B:That's really depressing.
Speaker A:And now you're one time per company, right?
Speaker A:Well, I didn't tell.
Speaker A:We did finish a bunch of projects, but the last one obviously got canceled.
Speaker A:And the same with broadcast.
Speaker A:Yeah, yeah, you're right.
Speaker B:And you've.
Speaker B:You've built this from nothing.
Speaker B:You had, you know, had ideas in your head.
Speaker B:You went back to, you know, your home, Barcelona.
Speaker B:Beautiful city, by the way.
Speaker B:My wife's.
Speaker B:In my mind, there are two cities in all of Europe tied for being our favorite.
Speaker B:Paris and Barcelona.
Speaker A:There are more, but I agree.
Speaker A:And it's a beautiful weather and you have the sea and you have a snow wealth with climate change.
Speaker A:Snow is going away, so I should take that.
Speaker A:Back when I was young, we had snow in the mountains.
Speaker A:No, no longer.
Speaker B:Yeah, no, I love Barcelona except for one thing.
Speaker B:You have to eat dinner so late.
Speaker A:Yes, yes, yes, you do.
Speaker A:That was my problem when I was in Boston or nearby working is that I would really like to come in late and I would get to the office and people say, do we go lunch?
Speaker A:And you say, what are you talking about?
Speaker A:It's breakfast time.
Speaker A:So, yes, we are shifted towards US timetable.
Speaker B:Well, this has been great.
Speaker B:Thank you very much.
Speaker A:Appreciate your very much for having me.
Speaker B:That's a wrap.
Speaker B:Thanks for tuning into this week's episode of the Designing Successful Startups podcast.
Speaker B:Check out the show notes for resources and links.
Speaker B:Please follow and rate us@podchaser.com designing successful startups.
Speaker B:Also please share and like us on your social media channels.
Speaker B:This is Jothi Rosenberg saying TTFN Tata for now.
