Robotaxis & the Future of Autonomous Mobility

Sriram Viswanathan: Hello, everyone. This is the TechSurge Deep Tech Podcast presented by Celesta Capital. Each episode, we spotlight issues and voices at the intersection of emerging technologies, company building, and venture investments. My name is Sriram Viswanathan, founding managing partner at Celesta Capital. Welcome, Aicha Evans, CEO of RoboTaxi developer Zoox.

Aicha is a very dear friend. I've known her for many, many, many years. We have lots of shared history and we're going to talk about that and all the fun things that she's doing in her current job as the CEO of Zoox, which is an Amazon company. And she started at Zoox in 2019. And prior to that, she's had a very long and exciting career at Intel, which is been a topic on many people's minds in Silicon Valley. But Aicha was the chief strategy officer at Intel until 2019. I'm truly lucky to call Aicha as a dear friend. I've seen her grow in the technology area for 20 years. I, as I tell her, I'm one of our biggest fans and I keep rooting for her on everything that she does.

And I'm sure she's going to be doing something very exciting at this company. So, Aicha. Welcome. We have much to talk about. Thank you for joining us in this podcast.

Aicha Evans: Well, hi. So good to see you. First of all, as usual, we are dear friends, with so much shared history.

Sriram Viswanathan: So let's, let's, let's talk about your background because I think, it's just remarkable, for someone who's had such an amazing career.

In the technology area, you actually started in a very modest setting in Senegal in West Africa. And then you spent time in Paris. And of course you had a remarkable father who was in telecommunications. if I recall he was an alcatel.

Aicha Evans: Yep.

Sriram Viswanathan: And so you were exposed to technology early on. Talk about your early life, and how you sort of got into technology.

Aicha Evans: Yeah. So I was, look, I mean, You Senegal is a smallish country, back then I think it was seven, seven million people. right now it's around maybe 15, 17. So it's tiny. We don't have gold or I think they just found some natural gas right now, but at least at the time, no, basically we had peanuts and seafood like fish.

Now we were the capital of the colonial French empire. So there was an emphasis on education and also a gateway into, into Africa. So I was privileged in a way, because obviously I had a long lineage of parents that were highly educated and family. but it was, you know, I was a Senegalese girl.

I don't know how else to put it. And so now I'm shuffling back and forth between Paris and, and a car. And when you're in Paris, you see what's possible. A lot of it because of technology, whether it's hopping on the Metro to go do stuff. But the most. Pressing thing is that I had friends in Dakar and I had friends in Paris.

So when I was in Dakar, I wanted to talk to the friends in Paris and vice versa. And gosh darn it, my dad was in telecommunications. So I was in a lot of meetings and under the table listening to things. And I was like, yeah, I can crack that.

Sriram Viswanathan: So that is truly your entry into wireless.

Aicha Evans: That is my entry into wireless, the virtual transportation system of basically information and knowledge and context.

I was a good STEM student. I was also good at philosophy and literature. Even to this day, I'm pretty good at languages. in Israel and picked up some decent Hebrew. And my dad, it's not just that he said, you're going to be an engineer. It was just like setting expectations. Like I don't know that I had any other choice.

And when I said that, I also liked philosophy. So yeah, a lot of the mathematicians like philosophy. So that's normal. And so did well in math and physics, but also enjoyed building things, repairing VCRs, breaking things down, putting them together. And so that's how I ended up in engineering. As far as Paris though, my dear father, who now has passed away and I, I owe him a lot.

It's funny as you get older and you have your own kids, how you reflect. I think he wanted me to be STEM on one side of the brain, but still be the traditional Senegalese girl. And that was a little too late. I spent too much time abroad. But what it came down to is I was longing for freedom, and the ability to have more agency in what I was doing, number one.

Number two, computers were coming up. And France had the Minitel because, you know, in France, sometimes we always have to fight the trend and we're smart enough to, but it was obvious that the U. S. was leading and so I said I wanted to go to the U. S. and I wanted to study computers, and, which is what I ended up doing.

But then, see, I didn't call the universities like you did. It is expensive to go to school here. So he had two conditions, if he was going to pay for it. One condition was that it was in Washington, D. C. I said, no problem. And the second condition was that I was in touch with more Senegalese people when I got here.

Uh, I think both kind of happens, but, but I also met my husband, who is, I went to GW and he was at University of Maryland and we ended up meeting at a Senegalese concert and then obviously were still married and have our kids. At GW I was lucky. It's a very small engineering program and so, fell in love with hardware.

Uh, we were working with Motorola and US agencies on special projects. Some professors like Professor Meltzer and Professor Corman really took interest in me, and chips were the thing. So, went that direction and basically the rest is history.

Sriram Viswanathan: So you got into VLSI very early on and in fact, I saw you and you and I worked together at Intel.

I was, I saw you in Israel.

Sriram Viswanathan: I think I forget it was maybe 2005, so 2007, yes, you're in Israel and I was running the wireless program office at Intel and then you were part of my group and then we got to know each other and you just kept going. So, let's briefly touch on your wireless program.

Part of your, your history, because you, you really have been a pioneer in looking at the transition from WiFi to things moving beyond WiFi to cellular and all of that. And you were, you were, you were natural at that. Talk about, What made that decision for you, to jump into Intel specifically, because you were not at Intel when you started the wireless journey, but what attracted you to Intel at that time?

And what did you, how did, what was that experience, initial experience?

Aicha Evans: So What happened is I was working on cellular technology, for Skyworks and in the middle of all that, I had a difficult experience, trying to become a mom. I had yeah, a couple of, I had a very difficult miscarriage.

And so I decided I wanted to, we decided we wanted to have kids. So I actually left Skyworks thinking, you know, I'm just going to take a break when I was pregnant with my daughter. And Intel started calling, because I lived, we happened to live in Portland, Oregon. My husband was at Mentor Graphics and we liked the lifestyle in Oregon.

And so, Intel, I think it was Paola, Paola Moreto, started calling me and I was like, you guys are memory, microprocessor and process company. Why are you calling a wireless person? And Paola sat, talked to me about WiMAX and I said, yeah, that sounds, I mean, the technology is going to be okay and is magical, but from an ecosystem, this is going to be expensive.

And she's like, well, we need wireless people and you're important. So that's how I joined. And then I met you, I met Sean Maloney, I met daddy and there was an all, I think it was a, one of those things where they say a few engineers will meet with daddy and Sean. And I asked the question, I said, Hey, this is a very hard technology.

It will work. But from an ecosystem and a sort of incumbency standpoint with 3GPP, what exactly are we trying to do? This is going to be very expensive. That landed me in a meeting with Dadi.

Sriram Viswanathan: And for those of the audience that doesn't know Dadi, he was the head of engineering and ran pretty much all of processor and wireless development and architecture and all of that at Intel.

And Sean Maloney was the guy who was actually. And I cannot believe that was a time when Paul Otellini was the CEO, who unfortunately is no more. And I recently read an article. that talked about the four CEO curse. I don't know if you saw that, which talked about the number of CEOs that Intel has gone through since then.

So, so talk about that period that you were, you were doing wireless and I, and I know this and it's great to, to hear somebody else, talk about it because both of us have battle scars out of that period. Absolutely.

Aicha Evans: Absolutely. One of the important things is to understand. You know, the why, what are we trying to do and why, and that led back to essentially the convergence of wireless and compute, starting with unlicensed technology, wifi, and then, with, having all compute devices, connected, to the, to the internet essentially, including cellular.

So, that, that was the, that was the story and the, and the Genesis. And so I first worked on the engineering side, I guess the first text test was, being in Israel, with the wifi team. and, that was a lot of fun, learned a ton, and then eventually coming back to the U S and picking up the, licensed, part and building, building modems.

Sriram Viswanathan: So this is, this is, you know, we're talking about circa, you know, early 2000, and it was a great time until it was. still sort of the king of the hill. you know, it was launching new laptops and people forget, you know, laptops and Wi Fi didn't go together. In fact, Wi Fi didn't exist. it was actually HomeRF or 802.

11 as the two candidates for Wi Fi, for wireless access and Intel made the decision. And you were very much part of those early days in driving that and, and Intel did a fantastic job. So if you fast forward from that period, Where Intel actually was really at the top, and then subsequently you and I left, and then you were the chief strategy officer.

You were running the wireless group, all of that. Something changed in the maybe post 2013, 14 timeframe, because Intel gradually started losing its advantage. Can you talk about from your vantage point as being the strategy officer, what were those discussions? Some you obviously, I'm sure you are very much in the center of driving Intel to move in that direction and probably some they did and some they didn't.

So talk about the missteps that Intel may have, you know, gone into, which probably has resulted in where Intel is today. So talk about that.

Aicha Evans: Yeah. Look, I mean, I don't know that I have the monopoly on truth or understanding, so I'll talk about some general principles. It turns out, first of all, that big I'll talk about American companies because it's easier that way.

Big American companies, when they are very successful, staying successful is actually a very hard problem. And so I would say, issue number one is probably, what got you here won't get you there. To quote Marshall Goldsmith, I think Intel, the combination of X86 process technology and the systems around that, and then the ecosystem was extremely successful.

Um, there's a famous quote, I think it's by Jeff Bezos that your, your margin is my opportunity. So when you have a very high margin business, you actually have to be careful because you are creating an opportunity for others to go after that margin, and you need to be strategic about that. And then I would say that, even though engineers and technologists, we love talking about the technology, you have to talk about the culture and the people, because doing violence to your own set of beliefs is very hard.

And I think that's another component. Last but not least, it turns out that software rules. And you can even see that today in the AI transition that's happening. And I think there were some misses from that standpoint. So I think the combination of those four things during some difficult CEO transitions,

Sriram Viswanathan: Right.

Aicha Evans: Really created the situation we have now.

Sriram Viswanathan: Yeah. I, you know, I don't want to obviously make this podcast about Intel, notwithstanding the fact that you probably have just a wealth of knowledge. insights and perspectives on what ails Intel and what may be fixed. Let's just, let's just talk about the evolution of AI itself, because arguably, as you rightly point out, you know, companies like Intel or, For that matter, perhaps even NVIDIA.

You know, the saying goes that you have a, you know, you have a hammer and the world is a nail of opportunities, right? Because everything that is out there, you think that you can solve with the product and technologies that you have. And Intel may have suffered from that. But if you, if you sort of take that point of view with NVIDIA, who happens to be your supplier, we can talk about NVIDIA and Zoox and how integral their technology is.

But before we do that, Talk about the importance of AI, not just for the advanced, you know, LLMs that are coming out there, you know, vision and perception and every walks of life, AI is going to play a big role. Talk about how you see AI as an important aspect of any technology company, especially the likes of the ones that we're talking about, you know, NVIDIA and Intel.

And Intel and others.

Aicha Evans: I think as soon as the transition from single threaded to multi threaded happened. the ability, the simplest way I can say is if you have a thread of codes that was running sequentially, and you're now breaking it up to run in parallel, and you expect to see essentially an explosion in types, in terms of volume, speed, latency, and so on, software becomes critical.

And with AI, the way I look at it is the math actually has been here for a long time. What's now happened is the chips, the software and the access are very available, but there are also other problems. It takes a lot of power. And one of the things that actually scares me is it's going to be an Every aspect of society.

I tell young engineers now that I'm jealous because they are going to have the same feeling we had at the beginning of wireless. They are going to touch the medical industry. They are going to touch commercial. They are going to touch everything because we are going to have the luxury of being able to see the wealth of data and patterns that in the past we didn't think we could get to.

Now, what that means though, is that, integration of hardware software. power envelope, power efficiency, and prioritization of the data and cleanliness of the data are going to become essential. And I truly believe we're still at the beginning of that wave. And it's a wave that's too big to be controlled by one company only.

I think it's going to be quite broad and we're only scratching the surface.

Sriram Viswanathan: Yeah.

Aicha Evans: Where it's going to have, I think, and this is my own personal opinion, I think computers are good in general at patterns and rules. There is this thing called judgment that we human beings have. And I think true winners are people who are able to codify or add a layer that will somehow codify judgment.

I have no idea how one would do that, but that's going to be really the, sort of the cream of the cream.

Sriram Viswanathan: No, but I think there's a lot, lots to unpack in what you just said. I think power. Performance software, all of those things are, are crucial and maybe there's a perfect sort of inflection that's happening in the computer industry, especially with Moore's law delivering this sort of performance so far.

Um, but I think. If you just double click on that from an AI standpoint, our brains don't partition, learning from inference. So architecturally, do you see that as something very specific to the way AI is getting implemented today versus how human beings function? You know, we don't, you know, you, a three year old doesn't have, you know, a partition in his or her head that this is the period I'm going to be.

training, and then I'm going to infer if this is a ball or a toy or something else. They do it sort of in C2, in C2. And so architecturally, do you see that as an issue with AI implementations today?

Aicha Evans: That's why I think it's only the beginning. I think today it's separate, but I think the blend and how you blend it, because it's, it's going to have to be a continuous blend.

Um, I think that's the only thing that's going to lead us towards true general intelligence. I think if you split the two, first of all, there's an inherent latency. Second of all, the data changes. but it's not an unnatural beginning, kind of like we didn't get to what we have on, on LT and 5g overnight.

Uh, I don't know if you remember GSM, GPRS, and edge. Of course. We see. We sent, you know, we celebrated a few characters getting through. so I, I, I think you're going to see a blend and that blend is going to be continuous, not just, at all phases and the trigger points. And then the judgment calls is what will create the actual breakthrough.

Last but not least for AI engines to know what it is they don't know. Right. That is actually a power that human beings have.

Sriram Viswanathan: That's right.

Aicha Evans: Because when you don't know what you don't know, you consciously go seek it out if you need it.

Sriram Viswanathan: But AI hallucinates. And it's, you know…

Aicha Evans: And that’s a big problem!

Sriram Viswanathan: Not to say the humans don't, because you know, you see Washington, a lot of people are hallucinating about the truth.

Right. But anyway, we're not going to do a podcast on that. We're not going there. We're not going there. But, but, but I want to, I want to just go back to what you said, on this comparison that you made with GPRS and edge and other stuff. So if you were to sort of think about the compute industry in general, you could make the argument, it was an exponential industry, thanks to Moore's law, and you could argue that, you know, Metcalfe's law drove networking in a pretty dramatic way.

Uh, And that you could argue is sort of applicable in the telecommunications area. So all of these things were almost step functions. They were not really exponential in the telecommunications area, but in AI, it seems like the capacity to actually infer and the capacity to train is truly exponential in what has happened.

The last three years, why do you think that is?

Aicha Evans: Well, I think that first of all, I mean, the compute allowed it. Second of all, I'm going to give credit to Jensen a little bit in some ways. I think that having a hardware provider that has software capability, but that went into application space to show what was possible and stimulate the industry.

Uh, third of all, I think that the democratization of, almost, some of these algorithms and some of these learnings being. core to computer science for young students. I think all of those things allowed that. But at the end of the day, it's still the speed and performance of, because it's not like the ideas weren't here.

I mean, people, I remember, Jesse, a friend of mine always tells me about the amount of time that used to be spent in computer vision to recognize a chair. Turns out that was a hard problem, but that's also because it took a lot of cycles. So I think that, the general availability plus the fact that computers can do that.

And then this Simplification around hardware and software to experiment made that happen.

Sriram Viswanathan: Yeah. But, but, you know, to your point earlier about you know, maybe Jensen was an NVIDIA was at the right place at the right time, because, you know, we know at the time when you, you know, NVIDIA was really a gaming accelerator company and discrete graphics, they were struggling, right.

And they've built on top of that. But would you say that their commitment to software, as you, as you alluded to, You know, specifically, you know, if, if I wonder if NVIDIA could be as successful as it is, had it not been for CUDA. No. and so, so talk about what did CUDA do and why can't that be replicated by others?

Aicha Evans: Well, I'll answer the second part. I think it can be replicated in terms of new fields. I think it's hard because of just install base, right? I mean, I take our case. We have a lot of code running. If you tell me that you want me to run it on something totally different from the get go, I don't know that the switching cost and so on.

So, but if I'm standing, starting from a clean sheet of paper, maybe. So look, I think it was brilliant because, yeah. As soon as you're going to, like the simple principle of taking something that's sequential and breaking it up and having it run in parallel so that you can get gains of efficiency, speed, and latency to get to agency and outcomes, something needs to make that transition.

And I think that's essentially one of the things at the highest, simplest level that CUDA does. And I think that was brilliant in the sense that it provided the API and access that did that middle work so that you could come in with your inputs, feed it in, and it did the work for you. And the fact that it had to be done for gaming and was done in a generalized enough way that that middleware became something that's essentially a kit is, was brilliant.

And he should get credit for that.

Sriram Viswanathan: But that's not new though. I mean, you remember the whole, you know, multimedia evolution with MMX, architecture extensions to x86. You could talk about the, you know, internet explorer time when the extensions to the browser itself did lots of things. Yet, I think, you know, Chrome came in, and took the lead ahead of Microsoft.

Why is it not possible for, you know, something else, you know, Fred, to take advantage of this established, understanding of what AI can do and have a completely new, you know, set of developers going after, you know, the next generation of CUDA, but it's not coming from, I mean, I understand it is a barrier to entry, but no barrier to entry sustains for, you know, All times, as we know, none do.

Yeah.

Aicha Evans: So look, it's about, the next one is going to have to deliver value that you can't easily get, or I hate to say this, the margins are going to be annoying enough that people put the firepower to be able to have something that disrupts it. Yeah. I don't think that it's a technology impossibility.

I think it's more of a commitment. it's, you also have to be a little bit lucky with the application because the application has to be, that you apply it to, has to be broad enough and generalized enough that you can transport it to other things. I hear people all the time say, Oh, well, they were lucky.

Well, first graphics and then gaming and crypto and now AI, and there is some truth to that, but I've told you before luck favors the prepared and lady luck knocks on the doors where people have done the right foundational work. So I believe that, with this, blended architecture of inference and training with this, push to control hallucinations with this thirst to go towards more generalized intelligence, that's not just based on essentially awesome pattern record, parallel pattern recognition, something might come out of that, that will deliver such value around a specific application that if done right, will then be generalized.

Sriram Viswanathan: Sure. But, but, you know, at the end of the day, you know, it's, it's great to be in the, In the business of picks and shovels, but at the end of the day, the picks and shovels have got to be able to dig up some gold. So where's the, where's the gold? Where's the golden AI?

Aicha Evans: So that. Is, I'm going to give you a very, surprising answer.

I am very skeptical. I think that from a usage standpoint and install base, you will see AI all over, all over the place. In terms of digging the gold at the end of the day, if you do not accelerate the business, the fundamental business parameters of an industry, or you don't deliver some.

customer value that they didn't even know was possible. There is no gold, bottom line.

Sriram Viswanathan: But I can see some of the things that you alluded to the applications of healthcare, you know, maybe in, in other areas of retail where AI driven systems could make. More intelligent choices and decision making, but, you know, in general, I think, would you agree that people are still not clear?

I mean, open AI is probably an exception, perhaps, all these large language models that are getting created, you know, Lama and Gemini and others, you know, where does this all go? And where, where, where is the money in, in monetizing, you know, billions and billions of dollars of infrastructure that these hyperscalers.

If you have to put in place, how do they monetize that with the AI? I can see how NVIDIA monetize. I say, but what about the others?

Aicha Evans: Well, first of all buyer beware. If you're the enabler of an ecosystem, but the ecosystem is not figuring out a way to make money. We all know problems arise. Right. And so, I don't know that people know, I don't know the answer to that at this point in time.

I do know you have to deliver value and that value has to be quantifiable in a way that either grows a town or makes a town more efficient. And I, I, you know, healthcare, banking, retail, you know, Analytics, lots of industries could benefit, but the benefit has to bring kind of a derivative value. Right.

Otherwise, that doesn't make sense. So I'm not sure the answer is clear. Even with open AI. Yeah, I understand it's hard, the subscription and this and that and the other, but I'm not sure there's clarity yet. I believe it will come though, because the capability is so clear in terms of what it could make available.

And we know that humans, even from the time, the analog time that we are, the more information we have, the more knowledge we have, the more access we have, the more reach we have, good things get created. But I can't sit here and tell you for sure. What it is going to be. 

Sriram Viswanathan: It’s kind of like, you know, build it, they'll come, which is a, which is a good segue to talk about the business that you're in right now, which is, you know, transportation, and Zoox, which is really a robot taxi business.

Uh, you could argue the taxi business, you know, New York medallion, the, the price of all taxi businesses, which is New York it has suffered tremendously when Uber came on. So Uber disrupted the taxi business and now you created RoboTaxi. So help me understand what is the problem that you're trying to solve in RoboTaxi?

What's the RoboTaxi problem statement?

Aicha Evans: Okay. good one. So I would put it this way. Right now, even at the current levels of, macroeconomics and socioeconomics is expensive. It's unsafe. There's a lot of pollution. I mean, hopefully we're done arguing about the climate problem. And when you fast forward 10 years from now, 20 years from now, 30 years from now, as even more people come up the economic ladder, frankly, especially in dense urban environment, it's not tenable.

Number one. Number two. There's a whole segment of the population that would like to enjoy mobility on demand, whether it's through taxis or what have you, that doesn't even have access today because they are not comfortable or because they can't afford it. And so it's really thinking years ahead and going with the volume of transportation and rides are the current solutions acceptable.

And maybe in Europe, with, you know, The, the, the economy there is more built around public transportation, but in the United States of America, it is still built around automobiles. What is a better, more efficient way to prepare for that explosion? And we think that robo taxis are the way to do that.

Sriram Viswanathan: Yeah. But you know, the US in general is, is, you know, everybody owns a car and I think, you know, Elon is trying to do exactly, you know, he would agree with. A lot of things that you're saying, but he believes in, you know, each individual owning an autonomous car, and maybe they'll get into the robot taxi business also, which he has announced.

And we'll talk about that in a second. But do you make this distinction between a car that is heavily, heavily driven with the foundational AI capability in it that drives you around and looks for you to take control when it gets into a sticky situation? Versus people sort of treating it as a car or as a, as a taxi, but really meant to be with taxi, maybe with more people with human, you know, in the loop, when it gets into a sticky situation, these are two completely different,

Aicha Evans: different business models, different business models

Sriram Viswanathan: and completely different movies.

What is the thing that attracted you to this movie?

Aicha Evans: So 4 percent of the time versus 96 percent of the time. So after your house in the United States or your dwelling, it's the second most expensive purchase a car. And by the way, it also dictates a little bit of your dwelling because you have to park it, garages and so on while we talk about the housing crisis and what have you.

And the reality is. You drive it to work or you drive it to an errand and then it just sits there and it depreciates and it takes real estate away. So to us, or to me, everybody having a car is essentially inefficient. wasteful, bad for the environment. And you look at commercial aviation, I'll take, the extreme and don't think about, you know, yourself and your friends, but for the general public, just because we could build a lot of private jets, even if we make make would make them cheap, it would be very bad for everybody to have their own jet, essentially.

Yeah.

Aicha Evans: Commercial aviation is solved a problem. And we think it's the same thing with cars. Decades on decades, continuing to make more of them, using more stuff, digging for more stuff to make them, polluting more, when really 96 percent of the time they are sitting and depreciating, we can do better.

Sriram Viswanathan: But I get that. I get the fact that the pollution part of it, I get, I get, you know, it's, it's a perfect proxy for, for public transportation, which is more convenient and all of that. Uber came up with a completely different model, obviously, and when Uber becomes, and Uber has a, you know, well established, you know, use case, people understand it, they have trained people on how to summon an Uber and get on an Uber and trust a driver, maybe sometimes they don't, but, isn't it possible for Uber to be a natural robo taxi extension business why do you need another RoboTaxi specific business when existing rental car companies, you know, subscription service car companies.

Could just extend it. Why, why do we need a, you know, as a robotaxi business?

Aicha Evans: Well, so the first thing is we need to have robotaxi technology. We're just starting the journey. It takes a lot of capital to deploy this. You, I remember, taught me a very important lesson, which is, access to the customer.

Now, who talks to the customer, builds the business. And I think what you will see, in the future is, the conundrum of partnering with Uber or not. And at the end of the day, if we're going to spend billions of dollars building this hard technology with a fairly low barrier of entry to get to the customer ourselves, why should we give that up?

Unless it makes sense for a partner. 

Sriram Viswanathan: Yeah. And I think those questions are still open. So the, the, the business model piece of it. Mm-Hmm. is still to be figured out. Mm-Hmm. . And so you're not saying that, you know, Uber or Tesla service or, you know, way more cruise or any of those, can exist and survive.

So we, you know, there'll be lots of people that'll try to do this. So how do you differentiate? I, is it like. Uber versus Lyft. you know, there'll be, you know, Zoox versus Waymo and I can get to choose. How do you build a long term sustainable advantage? Because, you know, it's not about access to technology because whatever technology you have, which is largely powered by the H100s or what NVIDIA gives you, is accessible to everybody.

So what's your unique secret sauce? That Waymo doesn't have, for instance.

Aicha Evans: Well, I mean, I think Waymo is doing quite well. Congratulations to them. I mean, you go in San Francisco. You can't come out without seeing one go by. 

Sriram Viswanathan: Or at least you can't leave San Francisco without hearing them. If you have seen them honking all night.

Aicha Evans: Oh, come on, be nice now. We're friendly, so be nice. I'm sure they'll solve that. Look, it's going to be about the writer. What do you provide the rider? Because obviously if the rider is not driving fully, and I don't mean you drive sometimes because, Oh, now you have to take over because the car, I mean, literally the driver's being transported.

And so what do you do? What do you offer to the driver during that time? What is the customer experience? how that's done. an extension of their own space, the freedoms that you unlock, because again, as you know, at Zoox, we don't talk about it as a car. We talk about it as it's a robo taxi, meaning when you're in it, you can do whatever you want.

Uh, the ability to scale is going to be key. the ability to remain safe and we don't want to compete on safety, but there is an implicit trust and trust. in terms of safety. And if you're not as good as your last ride, you're as good as your next ride. But I think the, the, the, the, the, the customer experience, and some of the, some of the, the folks we will bring on, that's where the differentiation is going to be.

Sriram Viswanathan: Yeah. So, that's an interesting point. Safety is not something that you want to differentiate on. That's, that's a, that's a good line. You know, 40, 000 people die in, in, in the United States. But if you actually zoom out and look at it, you know, there's, you know, one fatality for what? 100 million. Which is, you know, which is.

It's a hard, it's amazing. It's unbelievable. George,

Aicha Evans: It's what we were talking about earlier when we talked about AI, right? The 43, 000 unfortunate fatalities. That's more about. Yeah. people were drunk, people were on their phone, they were distracted, and so on and so forth. And, you know, we are confident as an industry that we will bring that down.

Not to zero, but down. One fatality per hundred million miles is something we don't talk about enough. I mean, that is an incredible number. Bar to meet and beat. And you have to beat.

Sriram Viswanathan: Yeah. I wonder what is it for the airline industry? It must be, you know, significantly order of magnitude different.

Aicha Evans: Exactly. I don't know the exact number, but same thing.

Sriram Viswanathan: Yeah. Yeah. So, going back to the technology, if I'm sitting in a way more. RoboTaxi versus Azus, and you have both, and we should talk about the actual technology components that go into a RoboTaxi. Obviously, you use LiDAR, you have radar, you have 360 degree, scanning all the time.

Uh, and, you know, how many H100s do you have in the car? Lots of them. Lots of them. So can you talk about. What goes into that car from a technology? You just break it down into the subsystems. What are all the pieces?

Aicha Evans: So obviously you, you have the shell itself and the inside that hopefully looks like a moving living room is what I call it.

Uh, now, you have center parts. This is how you see and perceive the worlds. And inside of them, you have long range LiDAR. So to see far in front of you, short range to see around you. you have cameras. you have radar, so you sense movement, in our case, we also have thermal cameras or infrared.

Well, night vision and also for sensing things that are alive.

Sriram Viswanathan: Yeah.

Aicha Evans: Because you never know. and so, you know, that worst case, we sense a fusion. You might not know what it is, but you know that something is alive and you can maneuver around that. So, we have one at each corner, so that we have not just.

360 degrees, but we have overlapping 360 degree circles, all of that data comes in.

Sriram Viswanathan: So there's some redundancy with LIDAR. Redundancy

Aicha Evans: is absolutely key. Look, when you're talking about no manual driver.

Sriram Viswanathan: Yeah.

Aicha Evans: When you're talking about essentially the computer is in charge of the driving at all times, having redundancy from a safety case perspective is key.

Aviation taught us that. Right. And so, and things will fail, things will disagree. So making sure you, you really maximize your chances of having the right perception and the interactions around you is key. all of that data then, yeah, goes to, essentially the perception engine, which runs on on the big box.

And then from that, the perception engine breaks that down into a representation of, of essentially the environment around it, pedestrians, cars, agents, you name it. the network itself of roads and so on that then goes into prediction, which is all of the non static elements in that perception scene.

What do we think their movement and actions are going to be? And then from that we choose and send that to planning and control that actually generates the trajectory for the next level of movement and actuates the vehicle to meet that trajectory.

Sriram Viswanathan: All of this happens. Real time.

Aicha Evans: Real time. We do not do any, all of the driving is done with compute on boards.

Sriram Viswanathan: Yeah.

Aicha Evans: Period. Now we do have a human in the loop. Right. As you would have with aviation, by the way. and obviously having low ratios there is important from a business standpoint.

Sriram Viswanathan: So, so obviously Waymo got ahead, you know, much before you and they are in Phoenix. Phoenix,

Aicha Evans: San Francisco, and Los Angeles, Los Angeles.

Sriram Viswanathan: So is this a business of, you know, who gets to the most number of cities in the shortest time and gets. You know, broad consumer appeal and builds brand loyalty or, or, you know, how, how does this thing evolve? I mean, do you envision this kind of like, you know, I mean, Lyft and Uber seem to be, you know, people's preferences kind of like Coke and Pepsi, right?

So, is that, is that the scenario here too?

Aicha Evans: No, well, I mean, look, I can't predict the future. What I do know is that I expect Zoox and Waymo to be successful. I expect that we probably will be in similar cities because we all start with a better weather, and where, and sort of cities where there are people are already accustomed to, ride hailing.

It's not a new concept. It's just the way you're delivering it is better. So I believe the time is going to be so big that this is really not going to be an issue. And some people, you know, I haven't changed too much in not liking to criticize others, but I do believe the customer experience in a zoox is totally very, very different.

I mean, I think at scale, I mean, I, I look at what happens when my kids are driving me, I want to comment on the driving. I'm like judging the driving. I don't give you that opportunity because you're not, Zoox doesn't give you that opportunity. You're not involved. You don't even know what's going on.

Success equal, you've even forgot that driving was involved. And I remember when you came for a ride, you were like, Oh, I was like, yeah, that's good. I wanted to catch up and be like, how are you doing? And what have you, it's only when we arrived that you were like, goodness gracious. We were on public roads driving.

And I'm like, yep.

Sriram Viswanathan: I was actually blown away by that. And I have to tell you that, you know, just yesterday I tried out the FSD, with Tesla and I was blown away by, you know, obviously I've had different versions of the FSD that I've experienced before. There's the autonomous driving capability, the latest version, the smoothness with which it navigated.

Traffic and, you know, right turn on, on stop signs and stuff like that. It was almost, almost, human like. So the industry is obviously, you know, moving, you know, at a very fast pace. some would say, well, it's, it was promised. You know, five years ago, but that notwithstanding, it's here, but I want to go back to your, your technical.

Aicha Evans: I do have to make one comment though, because I also have one, as you know I think you were part of the people who made me discover, you're like, do you know about this? I was like, Whoa. And I've had one for a long time. So I think they are doing an incredible job.

Sriram Viswanathan: Right.

Aicha Evans: I think, if. If everything is around providing the best driving experience, individual driving experience, I'm in chapeau.

I think the question is going to be around a robo taxi because my question to you is, are you ready and willing based on your experience yesterday to go to your car right now and tell it to go to your house without you?

Sriram Viswanathan: Well, I mean, you're, you're, you're asking a very pertinent question because, you know, I spend time between New York and, and, San Francisco and I, I never take Bart and I always drive.

Uh, but in New York, I never drive and I always take public transportation. And I think you, you said it, in one of the conversations before that, the big price for robo taxi is winning New York, but, related to that, a third of this country. for a third of the year is unsafe with snow

and

Sriram Viswanathan: LIDAR and radar and all of these things that you're talking about, as advanced as they are, they're still not as reliable in snow.

So how do you expect to overcome that challenge? If indeed New York is the price, because, you know, if without getting New York, you know, you can be a great business, but you're still a small business.

Aicha Evans: Look the thing with, with snow, first of all, we're all working on it. we've tested our vehicle on snow, so, or our robot on snow and our robot taxi.

And we're, we're not worried about that. Is it LIDAR? No, it's different. The problem with snow is that the environment is constantly changing. A road that you drove yesterday. Same road could drive very differently the next day because the person who plowed it plowed it slightly differently. And so where the lane markings were, where everything was has changed a lot.

So, I mean, I don't want to reveal anything too special, but the, the, what we're focusing on is solving dynamic, rapid world change because that is the Holy Grail for snow. And then being able to validate that to the point that you can deliver those. four to five nines that are needed for a robo taxi.

But it's definitely coming. It's just not the first step. there are still a lot of cities in this country, to, I mean, all of the major metropolitans from, you know, the Southern, like you leave California, go a little North, then go South and turn. So there's a lot of business out there. The other thing is, How often does it snow is important because there are cities like DC, when it snows, the city shuts down.

What's interesting is a city like New York, when it snows, people actually drive more. The demand is higher. So being able to control spikes as well as world change is what we're working

Sriram Viswanathan: on. There's the stability of the vehicle. Change with form factor, or in other words, do you see a model, do you see a scenario where the core IP of Zoox, is used by Metrobus and other people?

Is that, is that part of your strategy or did it?

Aicha Evans: It's not part of our strategy per se, but that's, that's a possibility. one of the thing is we're more of an abundance company. We think that, the same way a brand new ecosystem was born around, ICE cars, The same thing will happen with robo taxis.

There are a lot of things we do ourselves today that I don't think in the future we'll do because companies that are maybe gas station companies today or that are maybe resale companies or maintenance companies or towing companies will graduate with us and evolve with us. And so absolutely we, it's, we don't have this thing where we're going to be like, no, we're not going to share anything with anybody, but we would like to get to market, compete, have a foundation and base, and then.

Sriram Viswanathan: Or in other words, it's conceivable that you could be like a white box supplier of the technology where somebody else could make with your technology. You know, a robo taxi like business

Aicha Evans: someday, very possible. We would have to be like minded, because safety is, as we said, foundational, right back to that one fatality per a hundred million.

And our bar is a lot higher than that, but absolutely conceivable.

Sriram Viswanathan: Yeah. So as you think about expanding this market, I sort of see. In addition to overcoming a lot of these technical challenges, in addition to sort of getting user behavior and acceptance solved, because that's a, you know, that's a, that's a big deal, that's a big deal.

Yep. But you also have a huge challenge in regulatory. You know, obstacles and we're an investor in a, you know, a little simple digital license plate company called Reviver and we know what it takes to get approval. I mean, who would not want to avoid having to go stand in front of DMV, and you would think that.

Every state would say, let's make the DMV more efficient by just going completely digital. And it is a process. And this doesn't include, this doesn't involve functional safety, right? So you are having to deal with it. What is, talk about the process by which you get approvals in, in Nevada and California.

Is it like, you gotta go, you know, one bite of the apple every single time or how does it work?

Aicha Evans: Yeah. I mean, right now, let's, let's leave federal and sort of the, the Robotaxi aside because that's, you know, FMVSS, NHTSA, everybody knows that process, though we use self certification, but let's put that aside.

Right now it's, it's canvassing. You have to talk to the state. you have to then talk to the local municipality and, and get on there. I have to tell you though, at least so far, maybe it's because of the approach. We're very transparent with them. they know that, we don't have any incentive to kind of do anything stupid.

So very measured in our approach. we promise them no surprises before we even, before you see the first robot taxi, there's so much engagement, including with first responders, authorities, and so on. So it's been what I would say. A long one by one process, but it's been a good process because the reality, you know, Shreya, you know me well, we should, there should be a lot of care in this.

Yeah. You are sending, we are sending people, a machine to drive amongst humans. Somebody ought to check what we are up to and how does that work and is that okay and how that, how is that going to evolve? So it's been tedious because it's canvassing, but it's also been rewarding because you can see the evolution and, and I, and I think it's well wanted.

That's the way it should be.

Sriram Viswanathan: But is there not a federal department of transportation level? opportunity to sort of look at, I mean, not just Zoox's needs, but, you know, Waymo and everybody else. Is there an industry level understanding of what's needed?

Aicha Evans: We would love that, but, you know, we have to be realistic.

Uh, so, I think in terms of the vehicles themselves, yes, there is, yeah, NHTSA and FMVSS, but and that's a fair motor vehicle safety standards, but in terms of an AV federal framework, not quite yet, but I am hopeful. I mean, I'm sure the, I've been studying the story of the FAA.

Sriram Viswanathan: Yeah.

Aicha Evans: It didn't happen overnight.

Sriram Viswanathan: Well, you were trying to become a fighter pilot before you got the wireless.

Aicha Evans: I still, and you know what somebody told me, they're like, did you watch Top Gun? I said, yes, I did. They're like, you know, that's propaganda. I said, I don't care. I still watch it.

Sriram Viswanathan: You bought it. You bought it completely.

But let's, let's just, you know, let's just in the final few minutes, let's just talk about it. Amazon, you know, speaking of Top Gun and, you know, people going at much faster speeds than, you know, speed of sound, you know, Bezos has this company, you know, Blue Origin, which is, seems like a, you know, completely separate from, it's not even a startup and I understand, you know, people don't even get equity.

It's really kind of very done very differently, but that all withstanding, where does Zoox fit into the big umbrella of Amazon?

Aicha Evans: We'll bet. we're about growth, we're about innovation, we're about AI and upon the foundational success, there are a lot of potential things we could do together.

Uh, but you know me, we have to earn that first and yeah, the top executives at Amazon talk to us all the time. We have meetings and what are you up to and what are you going to do? And if you did it great, if you didn't do it, what's wrong and what can you do about it?

Sriram Viswanathan:  So you're, you're really running as a separate company within the Amazon framework and you have board meetings and board approvals and all of that. 

Aicha Evans: All of that good stuff.

Sriram Viswanathan: Yeah. So I'm sure that there are times when you have to disagree with your board chairman. how do you, how do you, how do you get over it? Because you're not really a startup.

Aicha Evans: Look, I mean, I've been asked this question a lot. We haven't had a lot of disagreements. We have a lot of debates. But look, there are some things that I will never, the decision belongs to the decision maker and I make my peace with it. but as you know, I've said that many times, if you ask me for my professional opinion, I will always give it to you and tell you the why of it.

Um, there are disagreements that, you know, are non consequential, meaning, you know, it's mini mini mo. I don't pick those fights. If it's something that's foundational, we talk about it. We discuss it. We bring data. we experiment, but it's 

Sriram Viswanathan: So you have your six page lineup.

Aicha Evans: Our six pager, exactly. It's been a great experience being in that company. And I am learning a lot. Like the six pages is a great thing because we

Sriram Viswanathan: are referring to the Amazon culture when you write a doc, no, no PowerPoint, BS, but you're really writing a doc. But is that, is that culturally, how much was that an adjustment?

From Zoox as a, you know, move fast, break things kind of company. So how, how, how are you finding that balance, preserving your nimbleness and operating within that?

Aicha Evans: It's been great. I mean, Zoox operates exactly the way it operated before it was, it was basically a sub of Amazon. I would say the six pager at the beginning, we were offered the opportunity to not do it, but I like the six pager writing in prose and putting thoughts and views on the table.

Uh, As opposed to doing PowerPoint bullets and then in the meeting, kind of like discussing them and maybe people hear them one way or the other. having at the beginning of the meeting, everybody read the doc. So, you know, everybody saw the same information and then ask questions about it. It's something that I've enjoyed.

And I think it's very clarifying. And frankly, I recommend it because I think there's a lot of stuff that put people off. People put on PowerPoint that maybe they remember, maybe they didn't, and maybe this day it meant this. Maybe the next day it meant that. That didn't really that didn't really bother us.

Sriram Viswanathan: That's fantastic. Well, listen, I, I, so what, what, what are some big milestones that you would say we should look for, for Zoox to cross over the, let's say, next, next year? Three, four years.

Aicha Evans: Well, first of all, see you on the strip. see you in San Francisco and, looking forward to having people download the app in, you know, multiple cities.

Sriram Viswanathan: But listen, Aisha, you know, as I started out this podcast, I'm a big fan of yours. I know you take on hard problems. one thing I know is that. You may be wrong, but you're never in doubt, and that is fantastic. And that's kind of what's needed in an entrepreneur. Life is too

Aicha Evans: short. I mean, you should do things that are worthy.

Sriram Viswanathan: That's fantastic. And I wish you all the best. And, and I tell you, it was transformative when I actually went on your vehicle. I think the user experience is what was different. I mean, I've been on the way more, you know, being in your vehicle is different. different. and I wish you great success.

And, hopefully you'll come back again into this podcast and tell us about all the other great things that you've done. Thank you, my friend. Thank you. Thank you, Aisha. You be well.

Aicha Evans: Right back at you and you be well too. Thank you.

Sriram Viswanathan: Thank you for tuning in to the Tech Surge podcast from Celesta Capital. If you enjoyed this episode, feel free to share it, subscribe, or leave us a review on your favorite podcast platform. We'll be back every two weeks with more insights and discussions of all things deep tech. Bye for now

‍