Listen: Bullish on AI, Bullish on Self-Driving, But Apprehensive About OpenAI & Tesla (with Jon McNeill)

[00:00:00] Hi, I'm Eric Newcomer, and this is the Newcomer Podcast. Madeline is out this week. I have a fun co-host special guest with me, Jon McNeill.

[00:00:12] Hey everybody, nice to be with you, Eric.

[00:00:14] You were on an earlier podcast. Thanks for coming back to our new iteration. Jon and I go way back. He was the chief operating officer at Lyft where I got to know him.

[00:00:24] I followed you to Las Vegas and wrote a profile, and obviously we talked a lot over the years about the ride hailing industry.

[00:00:33] You were also top executive at Tesla for a long time. What was your title at Tesla?

[00:00:37] Jon McNeill It started out as president of sales and then became president of sales and service, and then it was a bunch of stuff after that. So at the end, I think I was just president.

[00:00:47] Jon McNeill One of the people behind the scenes doing a lot of work. Well, Elon is the face of

[00:00:54] the company. And now you're on the board of GM, which obviously puts you in sort of cruise world a lot. And then your day job as an investor helping to what incubate? Do you use incubator? I never know. Accelerate, incubate, what word people, what is the word you prefer and how you think about starting these companies?

[00:01:15] Jon McNeill So the name of our firm is DVX Ventures, and we describe it as we invent companies.

[00:01:19] Because literally, we come up with the idea, and we build the initial product, put it out in the world, and start to work on traction. And then once we get traction, we hire a team. So we're kind of inventing and then hopefully accelerating and incubating all at once.

[00:01:36] Jon McNeill Do you miss the operator world? Or like to be at a company like Tesla, where it's so much work to being an investor? Does it feel like a whipsaw? Or how do you experience the life of an investor?

[00:01:51] Jon McNeill It does feel funky. One of the questions when I was starting DVX, Reid Hoffman was super generous with his time and said, hey, come down and let's spend an afternoon together and run me through how you're thinking about doing this.

[00:02:05] Jon McNeill And one of the questions he asked me in that session was, will you ever be a CEO again? And I said, why do you ask that? He said, well, because I actually had the idea to set up one of these idea studios. And I did it with Max Lepchin.

[00:02:22] And the very first company that we invented was a firm and he jumped in as CEO. And he said, we never thought to have a talk about this. And all of a sudden, I'm giving money back to investors because my partner's gone.

[00:02:34] Jon McNeill And he said, you really need to make this decision. And I said, I've been a CEO seven times. Like I've checked that box and I really like working with other CEOs and helping them.

[00:02:46] And so I said, I think I can affirmatively say I'm not going to be a CEO again. And that's been really helpful. But I got to say, like, I do miss the energy and being energized by being day to day, neck deep in a business

[00:03:03] with a team solving problems. And being an investor is several orders removed from that. And it's hard for me some days. I have to admit, like I just

[00:03:13] But thankfully for us as an investor, you're much more free to opine on the world than probably once we're working for the company.

[00:03:22] Jon McNeill Do this in an operating role. That's for sure.

[00:03:27] Well, first, yeah, I want to talk about open AI. So for context, you know, literally like a year ago, I did a draft with my friends, Max Shaw and James Wilsterman, who I co-host through Royal Valley with on one of our podcasts.

[00:03:40] And in the draft, I basically insisted that I be able to pay for the right to go first because they wanted to sort of randomize it.

[00:03:48] I was like, no, the whole bet is how much you're willing to pay for open AI. And I took I think it was a minus seventy five billion dollar handicap.

[00:03:56] And, you know, Databricks was the second rank company. So that is a real cost.

[00:04:01] So I was extremely bullish in this one hundred fifty seven billion dollar valuation would seemingly validate that.

[00:04:07] So, yes, exactly. Why have I become bearish? I think number one is just the commoditization of foundation models.

[00:04:15] I feel like when I made the bet, it was, you know, at Cerebral Valley, it felt like we're all pivoting off chat GPT and open AI.

[00:04:23] And like everything is a reaction to the magic there and people trying to find the magic for their own companies.

[00:04:29] And I just feel like that's really changed where people are much more excited about applications and much less excited about any of the foundation models.

[00:04:38] And then the second is just sort of the the weaknesses that I knew about Sam Altman now really manifesting in open AI.

[00:04:46] You know, it wasn't clear. Maybe sort of my my view on some of his shortcomings wouldn't translate into open AI.

[00:04:52] But then you see so many people leaving, you know, the board trying to oust him and just like overall dysfunction and all these really wild, you know, ideas like a seven trillion dollar project to build chip infrastructure.

[00:05:06] And just, you know, sort of sort of a crazy idea.

[00:05:09] So so I think it was those it was the sort of weakness of foundation models as a business and and the weaknesses of the CEO leader who so much of the value was was hung on.

[00:05:21] Yeah, like I think, you know, the framework that's dominated tech for the past 50, 60 years is Moore's Law, where you've got essentially a every 18 month doubling in the capacity processing capacity of the chip.

[00:05:37] And that starts to define all kinds of businesses and business models around that, because the chips get more and more powerful and you can do more and more with software.

[00:05:46] And it's enabled things like obviously the PC and the mobile phone.

[00:05:51] But underlying that is kind of a business framework, which is, OK, you've got 18 months.

[00:05:57] If you've got a an advancement happening every 18 months is a business.

[00:06:01] You've got 18 months to make your profits.

[00:06:03] And then that gets commoditized and sort of gets lower placement on the shelf.

[00:06:08] And you've got another product that gets top shelf placement and you get to make money off of that.

[00:06:13] To your point on the foundational models, these things are leapfrogging each other roughly every 30 days with new releases.

[00:06:19] And what I haven't understood.

[00:06:20] You need 18 months.

[00:06:21] You have 30 days, right?

[00:06:23] You have 30 days.

[00:06:24] So now you have 30 days to collect rents.

[00:06:27] And I don't understand how you can get a margin profile that's attractive in a business that commoditizes that fast.

[00:06:34] And so I have, you know, I understand that roughly 15 times revenue doesn't seem like a crazy valuation.

[00:06:44] I understand that.

[00:06:45] But it also, it lacks an understanding of what the quality of the margins are.

[00:06:52] The multiple, if you just look at revenue, seems reasonable compared to Facebook and Google IPOs.

[00:06:57] This is a fast growing company with a lot of revenue.

[00:06:59] But then if you try to think of, you know, both Facebook and Google were profitable.

[00:07:03] Facebook was making like a billion dollars in profit when it went public.

[00:07:07] And, you know, the OpenAI is losing what?

[00:07:10] Five billion dollars probably this year.

[00:07:12] More probably next year.

[00:07:13] Yeah.

[00:07:14] It's spending a lot on a product that is way less differentiated than Facebook and Google were.

[00:07:19] Exactly.

[00:07:20] And I think you've got a couple of very big kind of questions to ask, which is what's the margin profile in the business that would justify that valuation?

[00:07:30] And the second is you've got a CapEx angle or issue in this business that you don't have in a lot of other businesses where you've got to keep building clusters, rolling out clusters.

[00:07:41] We saw Sam said last week as he was kind of poking Microsoft a little bit to say, hey, are you guys going to build my next cluster or not?

[00:07:50] Right.

[00:07:51] Right.

[00:07:51] And if they don't, of course, he's got now, you know, six billion dollars in the bank plus another four billion dollar credit line.

[00:07:57] And so he could probably take a pretty good stab at building that cluster.

[00:08:01] But it just speaks to the CapEx that is necessary in this business to keep maintaining your foundational model lead.

[00:08:09] And I think this race is going to skinny down to a few players fairly quickly.

[00:08:17] But again, coming back to your valuation question, I think all of these would be really interesting to start to get some visibility into.

[00:08:25] But investors today don't have that.

[00:08:27] We don't have visibility into margins and what the future margin profile looks like.

[00:08:31] And we don't have visibility into how much capital is going to be necessary to create those margins.

[00:08:36] And sort of the exact, you know, the regular we don't have access to the exact terms of like the Microsoft deal, which means that Microsoft gets to like gobble up a lot of profits that certainly Google and Facebook didn't didn't have those types of deals.

[00:08:50] I wanted to say, you know, in the piece I talk about Brad Gerstner at Altimeters, his bull case for OpenAI, which he delivered at Madrona's IA Summit in Seattle recently.

[00:09:03] Right.

[00:09:03] And, you know, his case is very much tied to ChatGPT.

[00:09:07] Just like forget foundation models.

[00:09:09] Yes, they're going to be commoditized.

[00:09:11] But ChatGPT is the iconic product.

[00:09:15] You know, Google had AltaVista and Yahoo and like that didn't matter.

[00:09:18] It was Google search.

[00:09:20] And this is ChatGPT.

[00:09:21] It's in the lead.

[00:09:22] It's making billions of dollars growing quickly.

[00:09:26] This is, you know, I sort of joked in the piece, ChatGPT is what every student has learned to cheat on their essays with.

[00:09:32] And they're going to take it with them when they go to the workplace.

[00:09:35] I mean, people are very dependent.

[00:09:36] It's funny.

[00:09:37] I was looking at Google Trends.

[00:09:38] If you look at ChatGPT, it's like during the workday, people are using it all the time and sort of columns on the weekends.

[00:09:43] Like people are really using ChatGPT.

[00:09:46] So what do you, do you have a take on that argument, whether it's a big enough consumer product to justify how people are thinking about open AI?

[00:09:54] I think you described Brad's case really well.

[00:09:57] I mean, he describes open AI, ChatGPT and open AI as having reached consumer escape velocity.

[00:10:04] And so you've got a multi-billion dollar business.

[00:10:06] You've got a $5 billion business heading to an $11 billion business over the next year.

[00:10:12] I see his analysts kind of built the slides that he used in the presentation at Madrona.

[00:10:18] But also last week in his investor, his LP summit, his argument is this is growing way faster than Facebook grew, than Google grew, et cetera.

[00:10:27] And so I hear that.

[00:10:29] Like I do think there is on the consumer side, there are folks that reach the breakout that become the verb.

[00:10:35] And open AI has become the verb for AI.

[00:10:38] Right.

[00:10:38] And then they do get positioned within the consumer markets of significant share and they get to collect rents.

[00:10:45] We've certainly watched Google do this in search.

[00:10:47] And I got to put the mirror on myself.

[00:10:50] My own consumer behavior is that I am searching a whole lot less and I'm asking Google and Anthropic and perplexity questions a whole lot more.

[00:11:00] Because I just don't have to waste time poking in to link.

[00:11:03] Wait, sorry.

[00:11:03] Google?

[00:11:04] You're not asking ChatGPT or are you asking Chat?

[00:11:07] Yeah, I'm asking ChatGPT and I'm not asking Google because I just am tired of clicking through links to try to find the right answer.

[00:11:16] And I get the right answer served up most of the time now.

[00:11:19] Very few hallucinations with either ChatGPT.

[00:11:22] Interesting.

[00:11:23] I don't feel that way.

[00:11:24] I want to be able to – once they – I like to perplex the idea that I don't use it regularly where I at least want to know why it's saying what it is.

[00:11:31] And it's very hard for me just like, okay, OpenAI has given me some factual answer.

[00:11:35] But now I'm sort of lost.

[00:11:36] I'm like, was it right or like what was it not?

[00:11:40] You know, I think I'm someone on Google search where I want to find the original link.

[00:11:44] You know, like so much – not even like the Business Insider rewrite of the original link.

[00:11:48] You know, I want to find the original original.

[00:11:50] And I feel like to me the chatbots sort of get you so far away from that.

[00:11:56] All right, Nat, Nat, at $157 billion, are you a bear or are you a bull?

[00:12:03] I may end up eating my words.

[00:12:05] But if I'm an investor charged with finding 5 to 10x opportunities, I just don't see the rent story here that creates a trillion-dollar company.

[00:12:18] And so I don't know how to get 5 or 10x my money at that valuation.

[00:12:21] And so I guess, you know, we voted with our feet.

[00:12:23] We were asked and flattered to be asked by a number of investors in this round if we wanted to participate.

[00:12:32] And we didn't because –

[00:12:33] Lots of SPVs running around.

[00:12:34] Lots of SPVs running around.

[00:12:36] You know, Coastal was real today, almost a half a billion-dollar SPV that went in.

[00:12:42] And so – but we didn't because I couldn't figure out how are you going to get 5 or 10x return on this.

[00:12:47] And I still understand.

[00:12:49] I know how Apple collects rents and it's a margin story.

[00:12:52] And I know how NVIDIA collects rents.

[00:12:54] And that's a very similar margin story.

[00:12:56] In fact, double the gross margins of Apple.

[00:12:58] So I understand why those companies are valued the way they are.

[00:13:01] I just can't – I don't have the visibility to understand that now with OpenAI.

[00:13:06] And I could be wrong over time.

[00:13:07] You just can't ignore the executive turnover.

[00:13:10] When I see turnover like that in – amongst executive teams, like this much, this deep, this rapidly, that points to something else going on that I'd really want to understand as an investor.

[00:13:24] Is that cultural?

[00:13:25] Is that just merely opportunities that people are seeing on the outside?

[00:13:29] Is it disagreement, a strategy?

[00:13:30] Whatever that is, I'd like to understand that because there are very few companies that over time have that level of exec turnover and are successful.

[00:13:37] I'm not saying there aren't any.

[00:13:39] Yeah.

[00:13:40] I'm curious what you have to think about this.

[00:13:42] This sort of fits both honestly like some of the Elon world stuff and this OpenAI situation.

[00:13:47] You know, there was this founder mode discussion where Paul Graham wrote the essay saying, you know, you need a founder who's really in charge of their company and talks a lot about Brian Chesky.

[00:13:57] And he compares that with manager mode.

[00:13:59] I made a snarky tweet, you know, when that was all happening comparing Uber stock price and Airbnb, you know, Uber being the manager mode and Airbnb being the founder mode.

[00:14:10] And Dara has performed better over the last few years.

[00:14:13] Anyway, but a subtle point that I've been making is that a lot of times you want these sort of founder mode execs right below, you know, the CEO.

[00:14:24] And that we place so much focus on the CEO and not a lot of the execs they have out running the business.

[00:14:30] And that in the Uber case, there were these sort of founder mode execs who understood the business and been there pretty early on.

[00:14:37] Like, how do you think about this sort of founder mode discussion as someone who has been sort of a number two at a lot of these companies?

[00:14:44] Well, like I was I was a six time founder before I met Elon.

[00:14:50] And so and so the whole discussion of founder mode totally resonates with me because I because I got to experience it firsthand.

[00:14:58] But then like to answer this question, when I got to when I got to know Elon and Elon asked me to join the team at Tesla, I said to myself, like electric vehicles wasn't my thing.

[00:15:11] Like, I didn't invent this. This was I wasn't the founder.

[00:15:13] But I found myself like just literally a few days in having absorbed that this the whole EV opportunity in space and Tesla's a company and literally adopted it and made it my own.

[00:15:27] And so I pursued I found I was pursuing that on a daily basis with all the fervor that I pursued my own companies as a founder.

[00:15:36] And I watched Gwen doing this at SpaceX and we might want to talk about that.

[00:15:41] But I yeah, because I think she's a good example of this, too.

[00:15:45] And again, I had this like similar conversation with Reid Hoffman where he described he had founded LinkedIn and he was operating as the founder.

[00:15:53] And then and then Jeff Weiner came in five years after the founding, roughly, if I have the story right.

[00:16:00] Right.

[00:16:00] And Jeff took LinkedIn to a whole other level.

[00:16:04] And Reid started almost immediately to call him a co-founder.

[00:16:07] And when I asked Reid why, he said because he acts like a founder.

[00:16:10] He acts like the founder I am.

[00:16:13] And there are cases where, like you said, Dara came in and really took Uber on a valuation journey that is a significant departure from Lyft.

[00:16:24] Satya has come in post Bill Gates and Steve Ballmer.

[00:16:28] And you could argue like the value that he has created is incredible, just absolutely incredible.

[00:16:36] And there are others.

[00:16:37] Brian Nichols, who was at Chipotle is now at Starbucks.

[00:16:42] You see these CEOs who are able to come in and make the mission their own and therefore operate like a founder.

[00:16:48] But they also have operator chops so they can scale these businesses.

[00:16:53] And and I think those those leaders are super interesting.

[00:16:56] So I think it's it's just a more nuanced framework than just saying there's founder mode and manager mode.

[00:17:04] And and those are two things like I don't think that that really works out in the real world.

[00:17:09] And as a shareholder, you want to find the Satya's and the Brian's and the Gwen's to come in and the Jeff's to come in and scale your company because they create so much value.

[00:17:20] All right. Well, let's that the Tesla self-driving day.

[00:17:26] Do you know what is he calling a demo day?

[00:17:28] We robot is what it was called.

[00:17:31] We robot.

[00:17:32] A little twist on Isaac Asimov's I robot.

[00:17:37] That famous book.

[00:17:38] But yeah, the event and a lot of sci fi borrowing overall at the event.

[00:17:43] Anyway, yeah.

[00:17:43] Do you want to give us a rundown of what what stood out to you?

[00:17:46] This was late last week.

[00:17:49] Well, I think what stood out to me was, number one.

[00:17:53] Like I will I have learned not to discount Elon, not to bet against Elon because he finds a way to win.

[00:18:00] And there has been kind of this multi-year buildup to this cyber cyber cab robo taxi event and full self-driving getting to really it's it's the potential of its name full self-driving.

[00:18:15] And so I like a lot of people was really looking forward to this to say, OK, I wonder what he's got up his sleeve now, because it's been in tech circles pretty well known.

[00:18:25] People move around these companies a lot between Tesla and Waymo and Cruise.

[00:18:30] And so you sort of get a sense of what everybody else is doing because of this employee transit transition between the three.

[00:18:39] And, you know, it was pretty well known that about a year ago they ripped up their current script of of deterministic neural nets and went to an end to end LLM.

[00:18:52] And you could see in the Tesla tracker, this is this website that tracks each release and self-reported from Tesla owners how many miles between takeovers.

[00:19:02] So, in other words, the car is doing something you're not comfortable with, you take it over.

[00:19:06] And that's it's called an intervention.

[00:19:08] And so the key metric in the business is miles between intervention.

[00:19:11] And just to set the set the story, like if you're if you've got a permit from the state of California, you've got to report this data to them.

[00:19:20] And granted, there's a little bit of noise around the definition of an intervention.

[00:19:26] But all this data is published for Waymo, Cruise, Zoox, Aurora, you name it that are on the roads in California.

[00:19:35] And you'll see that over the last year, Waymo and Cruise have tens of thousands of miles between interventions.

[00:19:44] And Tesla was at like 12.

[00:19:47] Right.

[00:19:48] We're talking you've got to get on a logarithmic scale to even plot those companies on the same graph.

[00:19:55] And Tesla is now up to 14.

[00:19:59] And and they're making progress.

[00:20:01] You can see it with every release.

[00:20:03] But that product, that progress has to be orders of magnitude to be able to get a job.

[00:20:09] And so I thought we were going to start to hear the technical side of that story on the robot taxi night.

[00:20:19] And we didn't hear any of it.

[00:20:21] It was a 19 minute presentation, half of which was robots dancing around.

[00:20:26] And it was almost like they had lost the script.

[00:20:30] And and, you know, I so I think I like a lot of investors was was disappointed.

[00:20:36] Some were dismayed.

[00:20:38] The fan the fanboys are definitely still stock market didn't like it.

[00:20:42] Right.

[00:20:42] It was down like 8 percent or something because the market is struggling to see, OK, what's the path?

[00:20:47] As you probably know, like the FSD system in a Tesla today is what's described as level two.

[00:20:52] And that means as a driver, you can't take your eyes off the road.

[00:20:57] And level three is where you can take your eyes off the road.

[00:21:00] So now you're hands free and eyes free.

[00:21:02] And level four is you don't have to pay attention at all.

[00:21:06] And the car can just drive.

[00:21:07] And that's where Waymo and Cruise are.

[00:21:09] So you've got like Waymo and Cruise at level four.

[00:21:11] You've got Tesla down at level two.

[00:21:14] And I think what investors were looking for was show us the roadmap to go from level two to level four to at least get equivalency with these other companies.

[00:21:23] That already have millions of miles of driverless experience.

[00:21:30] And when investors didn't hear that, you watch two things happen.

[00:21:33] One is the Tesla stock went down.

[00:21:35] And the second is the Uber stock went up because I think folks were either shorting and they had to cover their shorts of Uber and Lyft to say, gosh, we thought this was going to be a big announcement with timelines that were imminent.

[00:21:48] And this was really going to damage Uber's business.

[00:21:51] And they didn't hear that.

[00:21:52] So they reinvested into the rideshare market and disinvested from the Tesla story, which I think is really interesting.

[00:22:03] To paint the picture of the day for people.

[00:22:05] I mean, Elon has been hyping for years self-driving.

[00:22:10] They are going to be a self-driving company.

[00:22:11] They're going to replace Uber and Lyft.

[00:22:13] That's why you would have people shorting these companies.

[00:22:16] It's like, OK, this is going to be the big reveal, especially at a time where Elon's focused on the election.

[00:22:21] He's focused on XAI.

[00:22:23] You know, SpaceX, which we'll mention in a little bit, is having great moments.

[00:22:27] Tesla is big public company that his fanboys are heavily invested in is going to have its moment.

[00:22:32] And he's going to present everything.

[00:22:34] And what do we really see?

[00:22:35] We see sort of like prototype.

[00:22:38] You know, we see like a car with two doors and no steering wheel.

[00:22:40] We see these cool robots that are almost certainly operated by human beings.

[00:22:47] But what we don't see is some huge leap in self-driving.

[00:22:50] Tesla is the ultimate sort of gets a lot of mind share relative to where the tech is.

[00:22:57] I would describe this.

[00:22:59] This is up to Elon to confirm this or not.

[00:23:01] But I would describe the rocket business as his first love.

[00:23:05] And I would describe Tesla as a business that he was an investor in, but a reluctant CEO.

[00:23:13] He had to come in to save the company, not only with money, but with leadership.

[00:23:19] But I think his first love is being at Starbase, watching those things go up and watching them come back down.

[00:23:29] And pushing the engineering teams in his role there.

[00:23:33] And Tesla is the cash cow.

[00:23:34] That's the thing that makes this all work.

[00:23:35] That's how he buys Twitter.

[00:23:37] Well, it's certainly the personal asset that allows his personal investment into SpaceX, etc.

[00:23:45] But now you could argue that on its own, SpaceX is a cash cow.

[00:23:48] I mean, they've got dominant market share delivering payloads to space.

[00:23:51] And now they've got Starlink, which is the only product of its kind, and no competitors.

[00:23:58] And so they've got two cash flowing businesses that are really attractive now at SpaceX.

[00:24:03] So I think SpaceX is a fantastic cash flowing business in its own right.

[00:24:08] But Tesla's stock allows Elon to do things like X and other cash consumptive dreams that he's got.

[00:24:16] John McNeil at DVX, thank you so much for coming on the show.

[00:24:19] This has been a lot of fun.

[00:24:20] It has been a lot of fun, Eric.

[00:24:21] Good to talk to you.