Generative Quarterly with Semil Shah: AI Expectations vs Reality, Tech Trends, and 2025 Predictions

0:05Hey, everyone, and welcome to Generative Now. I am Michael McDonough. I am a partner at Lightspeed. We are in the final stretch of the year. And so I am back with my friend and fellow investor, Samil Shah. We have another generative quarterly conversation where we ask each other if 2024 lived up to our outsized expectations for AI. Samil and I also dissect all the trends of 2024. Driverless cars, AI devices, Jensen Huang's rock star status, and the tech takeover of DC that we expect in 2025. So take a listen to this conversation with Samil Shah.

1:09it yeah well it's been it's been a big year obviously 2024 ton of ton of momentum continued hype around all things ai but i think a lot of people when you really dig in feel like it was a little bit of a letdown like 2023 was the year of like was the breakout year right and then 2024 was a little bit of reality setting in i don't know i mean i think uh i don't know what people were expecting or maybe their expectations were too high. But it seemed like there was a lot of technical advancements and a lot of really smart people building in and around AI and continuing to build. And it's sort of up and down the stack.

1:50You see people starting new things that people like us or around us are really excited by. And you see large companies making huge CapEx investments. in it. So it feels like it's just taken over the entire conversation. So from my view,

2:14it seems like tech advancement in and around AI up and down the stack is still moving. If it's a letdown, it may be that some of that could be pricing related or just momentum related, but those things kind of wax and wane, I think. Well, so pricing related, that's actually another thing I want to talk about, like fundraisers. But I think progress seemed to stall a little bit. I think a lot of people were expecting GBD-5. I mean, how many times were we hearing, oh, GBD-5 is right around the corner? Like if you think about how the models sort of act today as opposed to a year ago, I don't know.

2:54It doesn't feel that different. even like some of like in the media formats you know i mean i think a lot of people probably just assumed they you know like i think folks that we work with and in our industry this is a good skill because it the following skill they sort of extrapolate very quickly of what can go right and then they conflate that with it should happen soon and i think the reality is that like you have to take the good with the bad. I think the good is that like people would just assume like, you know, if you listen to Friedberg on All In Pod, he'll just say like, yeah, like you'll be able to just narrate a script and a background and audio and everything and create your own personalized movie and content.

3:40And like, yeah, you will be able to do that. The question is when and for how much. And so when you add time and money or compute in this case as a vector, you know, I think people's expectations might just be a little bit out of out of place yeah yeah maybe it's like we we had our minds like so blown in 2023 just like absolutely blown and then you start thinking about what's possible and believing that everything you're thinking is possible and by the way it probably is but given the rate of change in 23 maybe it was unrealistic to assume that rate would continue in 24. I also just think like the cost will catch up in people.

4:22So I think I view it as all like very positive and people would compare it to like iPhone moment and like putting a computer in a bunch of people's pockets. I think this is bigger. I'm more of like the Peter Thiel sort of, you know, what he said kind of, I think he said on Rogan, And, you know, it feels more like the advent of the internet, you know, and potentially bigger in terms of capture, value capture. So, you know, I think the question is when and how much will it cost to get there? And I think those are just unclear. When did you start thinking like this is bigger than the iPhone? This is the internet level.

5:11Like when did that click for you? Was that this year? Was that last year? I'd say kind of last year and clicking more. I just think of it as like, you know, I'm sure you've talked with Nabil about this, right? There's a horizontal thing. I think Jeff Bezos just talked about this a few weeks ago at the New York Times Dealbook Summit.

5:33So, yeah, probably about a year of just thinking like the Internet connected different people and created networks. and created an excuse for different graphs, including e-commerce, et cetera, et cetera, et cetera. This seems much more, even more horizontal in the ability for more and more networks of computers rather than just people and agents working within those networks to do things 24-7. And to go into different fields, to pull together research, make research a reality, all those sorts of things that we can pontificate about, like to me, it's just a matter of when. Yeah. No, I agree with that.

6:21I agree with that as well. Maybe one difference is that the tough part about it, and Teal goes into it on the Rogan podcast from the summer, but like the internet sort of leveled an economic playing field for folks who had access to the internet and could read, write content into the internet. I think the value capture here may go to fewer people and so could be more disruptive economically that way. And so I don't really know what the implications are of that. But I would say listening to Lex Friedman, Mark Andreessen podcast or the Teal, Joe Rogan podcast, those kind of things, when they talk about AI and value capture, those are pretty solid arguments.

7:10you know i i wouldn't take credit myself who do you like who is who is the small group of people like that is is that saying like because it's expensive because there's um there's like a finite resource that you need to be able to access this stuff and like who who has that resource well you can think about it right there's like the chips the data centers the energy like people have already started peeling back um all the layers of what's needed to go and so like you You can read the tea leaves from the CapEx expenditures. And so people want it to be there, and it's just a railroad time. So people are buying the land, building out the data centers, storing chips, building new chips.

7:55So the innovation piece and the economic activity associated with it is fantastic. I mean, it seems like it'd be hard to imagine anything like this in our lifetimes, frankly. Don't you think like we could also get to a place where this stuff is just as accessible as the internet is? And in that respect, like does it then become more democratized than just like a select few having access? I think this is like the open source argument that you and I have talked about before. And I do think there's good arguments for having open source capabilities. these. There's good arguments for having closed source networks for security reasons, things like that.

8:39I don't really know which way it's going to go. And maybe it's both. Maybe it's open source with managed services on top. I don't know. I mean, it feels like it's all moving so, so fast. But I wonder if the costs are catching up with all of us. That to me seems like a big unknown And, you know, the line I've been using with other investor friends is like people in our industry have conflated the sort of inevitability and the intoxication of the possibility of what this horizontal AI shift can bring with like at any price. And so, you know, again, it's sort of cliche now for VCs to talk about that.

9:26um i i've just never known how to like price growth deals to begin with before ai and now it just seems like um totally it's there's no rhyme or reason for it yeah yeah that makes sense all right so that's so that's like um what i was what i was calling stalled progress but maybe maybe that's actually not right i i opened that last one up with sort of like hey there's been a letdown here, but now let's hit sort of the other side of the coin, which you alluded to at the beginning of that topic, which was sort of pricing funding. We saw some of the biggest fundraisers ever in the history of venture over the past year.

10:10Two XAI rounds, or sorry, one confirmed, another there's rumors of that I've seen in the media. Maybe by the time this episode drops, maybe that will or will not be confirmed. We'll see. Obviously the OpenAI funding round, what was that 150 160 billion um valuation these are these are just crazy numbers right these are big big numbers uh to your point of of maybe at any price um yeah i don't know what do you what do you think about this i mean we were basically in um we were in a trough of a trough of sorrow for nvc for a couple of years and now we're literally seeing the biggest funding ever what stayed the same as that.

10:52There's just these compounding effects. You combine VC excitement and the huge funds with this scale and opportunity and intoxication of what the open AIs and X.AIs of the world can do. And you get a lot of intoxicated behavior, but it may lead to good times. You don't know. I think the one thing that's different now is that maybe from the last couple times we talk about this topic is that every quarter it feels more and more likely that like oh hey SpaceX may stay private forever and like OpenAI may like stay private forever and like maybe perplexity stays private forever like that feels like an unknown and I don't know if you know this guy Mark Rowan he's um I think he's one of the co-founders or like co-CEOs of Blackstone um he's someone i follow online because he's just when he gives an interview it's like extremely well reasoned like he's i i don't know like half of what he talks about that he's been on yeah i mean if you if you search on him on youtube i mean he definitely gives interviews but he's selective about the interviews you know um but he he he was saying now i'm parroting what he's saying i this is not my own thought but like he had something that stuck with me which was you know he said for like the last 30-ish years that he's been in the business the conventional wisdom has been that like hey if you're a large investor public markets are safe and private markets are risky okay and like that's probably how you and I grew up his view now is like that's actually flipped there's been a realignment where like potentially for long-term large institutional investors, private markets offer more stability and actually public markets are riskier.

12:52And I'm watering down his argument, but I think what he's saying is that the intense liquidity and algorithms running these public market trades distort value and create a lot of problems. So it's the old and recent argument of like, you know, 10 years ago, you know, why would you go public when you're, you're basically beaten, you can be beaten down any day or any week by an activist or an algorithm basically. And so now that technology is also there. So like, if you read Matt Levine from Bloomberg, you know, it's basically a solved technology problem that you can run these tenders now and, and do that.

13:39So open AI is doing that. So I think that's the one thing that's really changed, which is like, if there's more and more money going into private markets, more and more companies wanting to stay private, this is the 10 decade ago and recent argument coming to fruition today, and enough liquidity where like people inside open AI, they can leave as like, you know, with eight figures, you know, before an IPO. go. It seems like in that chain, it's better for everybody. Now, I don't know how many companies will reach that level of demand. That's probably the bottleneck where it's like you have these premium, premium assets.

14:24Yeah. What does that do to venture? I mean, I think we're starting to see they have to become multi-strategy and part of the multi - strategy, quote unquote, is to like rotate in and out of things before IPO. So like the traditional conventional wisdom is, you know, the benchmark model, you join Mike McDonald, starts a company, it's starting to scale, you join at the A, you take 20%, you join the board, you're with Mike and the co-founders for the rest of their journey, you train them to go public, and then, you know, you book a return. Maybe you trade in and out more of these things. A lot harder though, right?

15:06To trade in and out, especially if it's not one of those like super premium companies that you mentioned. Yeah. So then it becomes like, if you go IPO, it'll sort of be like, oh, if you're going IPO, you're actually not one of those top 10 premium companies. Right. Okay. Right. Yeah. I guess in that world, does the bar get lower to IPO? This is out of my zone of competence for sure. But what I've heard from other people And I always go to the Goldman Sachs annual tech event right before Thanksgiving in Vegas. So this is actually pretty topical because I talk to a bunch of people about these things just to learn.

15:46And essentially, like the way the math has worked, because basically if you're an investment banker the last three or four years, you haven't had IPO revenue. Right. You've had no M &A revenue. Right. Right. So it's just super dry. And even before that, the bar used to be like when you had Anchor or I started in venture, can your startup get to 100 million of annual recurring revenue with some growth? And then you can be public. Now it feels like it's 300, 400 million. And there's like, technically you can, I think, do a smaller cap listing, but there aren't banks out there willing to do that because it's basically all the banks have gotten really big.

16:32yeah and then the other thing that i think friedberg i can't remember what episode it was on all in but friedberg had a good review of like the coinbase and slack listing where like generally there's two types of ipos if i'm paraphrasing him there's ones where you can actually go ipo and also raise capital as part of it and that has lockup requirements or you can do these direct listings where you listen on exchange oh what spotify was yeah then you would know more than me and um you're not raising capital and everyone's liquid as soon as you hit the exchange yeah i think there is still a lockup i could be wrong but um yeah different idea though different idea and so maybe we see more people doing that i just don't know like um i don't know how it changes vc because if the funds keep getting bigger and these companies start growing faster and there's fewer of them there'll be a race flight to quality um and you know the current generation of entrepreneurs, I don't mean this pejoratively, like the majority of them, even if they reach scale, like don't know how to take that next step and sort of say, Hey, we have to go public or Hey, we have to sell or Hey, can we stay private longer?

17:41It's very difficult. Even some of the best CEOs I work with who have scaled a lot of level still don't know exactly what to do. Yeah. Super interesting. Yeah. I wonder if we'll see more opportunities or options, um, for private companies to seek liquidity or find you know or or enable investors to sort of like swap in or swap out like you're saying i don't know what that looks like i'm also this is not my area of expertise but that could be super interesting i mean spacex is probably the hbs case study on that yeah you know um let's talk about these these sort of like ai device flops you know there were a lot of like startups this year that tried to bring AI to physical devices.

18:28We'll talk about incumbents too, but at least from the startups or these very hyped products, Humane, the Rabbit to a certain extent, a couple other small ones here and there, they all just kind of flopped. I don't know. I don't know if that's necessarily surprising, but that's certain to be a little bit of a, it seemed to be a little bit of a theme this year. Well, I want to learn from you because I know you spend a lot of time on this. So like your take will be 100x better than mine. I think that in the consumer device, let's say consumer AI device lane, you're adding a lot of like risk factors already to something that's risky.

19:12So it's like, do I want to purchase new hardware? Do I want to connect it to my phone? Do I want to like charge a new thing and all these sort of things? So when you add all those layers or barriers into consumer behavior, the benefit has to 10x outweigh all the work and overhead you have to do. And I'm sure something will pop up. I just don't know what it is. And then the other vector is like, I can't remember if you and I talked about these Ray-Ban, Meta. you know that that's something i think that's just super interesting and i use those a lot in the summer oh really oh yeah yeah the ones with the ai in them like the overlay and stuff no i have the last version so like i'll get the new version but i'm i'm just a huge fan of that with like the music and like not having your airpods in so i don't know how that'll change all of that too like i'm not really sure but but i think maybe maybe to to wrap on that like i do think something like a key fob or something that you could just talk to or ask questions to that's like just voice you know i do think that will happen at some point but but again it could be the watch too yeah right i think a lot about the watch i i feel like the apple watch is in very theoretically i mean i also think apple's made some mistakes here with ai so far but like in theory the apple watch seems like a very very important uh opportunity for and the airpods right and the airpods and the airpods absolutely yeah i completely agree with that um what's your view isn't what's your view as an investor though like obviously you have to track all these new consumer ams coming you probably want one you probably want one for yourself you want it to work i buy all these things like as soon as they come out i buy you know i pre-order them i've got one coming i think within the next couple of weeks um yeah partially because it's my job partially because i'm you know i'm just i don't know i'm i'm i'm a nerd for this stuff i remember getting like remember the pebble watch back in the day like i think i got like the first pebble watch like i'm just always into this stuff but um no i think as an investor you know when i've looked at these it's been it's been hard to sort of build conviction i i'm not i actually am an investor that sort of um i try not to have too many hard and fast rules because nobody knows anything and so just buy and try everything and see what clicks no i i guess i'm just saying i'm never gonna be like i will never invest in a hardware product i just i don't know maybe but i i do think as the year has gone on i've i've it's it feels hard for me to um rationalize some of these things i just think there's so much risk as a startup in you know the the supply chain for hardware getting the technology right and then going up against obviously the incumbents um and isn't open ai like working on a phone open ai guys working on stuff yeah i mean i saw i saw something recently a startup product technology super cool like almost nothing bad i can say about the product the technology but it's like yeah it's gonna take us three years to get this thing to market because of you know the supply chain and stuff and three years just feels like an eternity especially in ai land especially when we already know these incumbents are building these things you know when all these these smart speakers came out kindle google home you know and then to some extent siri plays in this world with the watch and the phone it was a problem that these things are all on different platforms but the problem was somewhat mitigated by the fact that none of this stuff was actually that useful right they were they were they were just setting you know setting timers and playing spotify and whatever but i think the moment this stuff becomes really good i think that fragmentation becomes kind of a problem for the user um yeah i totally agree with that i think that's a great insight i'm actually wondering if the the best positioned company here is is the company that can do this horizontally well that's what like um that's like xiaomi in china right yeah they basically took over your home, which Apple tried to do, and they have a good position with their mobile devices, Apple TV, but they've really, they really haven't invested in Apple TV.

24:01Yeah. I think, I know you talked to, oh, by the way, I listened to your MG pod. Yeah. That was great. He's a legend. Great to hear his. Larger than life, MG. Totally a legend. I think he's been talking about just them actually doing a TV, you know, Apple. Yeah. Yeah. Yeah, he's talked about that before. I'm thinking very specifically about OpenAI. I think if OpenAI can become indispensable to the consumer, they will be in a really strong position to integrate with all of these players and be that horizontal glue that kind of connects all the devices, agnostic of who made them, you know? I see.

24:42So like they would go to the hyperscalers. Yeah. You basically say, give us premium access And then that could be the control point to like telling it what to do and like learning your preferences and doing it. Totally. I mean, you can already see this. They have the integration with the iPhone, right? Imagine that makes its way into the HomePod. I don't even know if people still buy that. But like, and they do something similar with Google. I think that's a bit of a long shot because Google obviously has Gemini and they're going to lean really hard into this. But I don't know, if you fast forward five years into the future and OpenAI is the dominant AI and sort of voice interface.

25:18I don't know, maybe they're in a position where they can do this. This was actually a huge part of Spotify's strategy. And it's fairly well known that Spotify, you know, from very early on, really focused on what they call ubiquity, which is being ubiquitous across all platforms and devices and just being present and being there. And that gave them a big leg up and leverage over, say, Apple Music, which wasn't available on Google devices or in your car or on your smart speaker or on your Google Home or your Alexa. Right. So I wonder if OpenAI can can do something similar. Yeah, I hadn't thought about that way.

25:57I mean, your insight is right. Like if I were to summarize what you just said, which I had never thought about before, the fragmentation of all these home devices didn't really pose a problem to folks like you and I or people at home because the service capability was limited. But if it gets to a point where the AI gets really good, the fragmentation across these platforms will be a problem. because then it's like, I want to just connect my Spotify to my Sonos, to my Amazon. And I wanted to know my preferences and know what I ate for dinner last night. Yeah, it's too much of a loop. So like, yeah, what's the glue, the personalized glue that brings it together?

26:43That future that you articulated feels more likely than one of those platforms dominating it because they haven't shown, they've shown weakness in a lot of these categories. Yeah. I think Google could win that. I put more stock in Google to win that than Apple. Actually, I don't know. Well, they won't. I'm going to hedge on that a little bit. But like Pixel, they need to flood the market with Pixel. They need to flood. Yeah, I was just going to say that. Apple's got the hardware advantage. Google's got the software advantage, I feel like. Yeah, they would need to make Pixel as popular as the iPhone.

27:17Yeah, that's fascinating. I hadn't thought about any of that. I mean, I've kind of given up on the smart home view. So I think that's a super interesting thesis. Do you use it? You don't have like smartphone home stuff? Not really. Really? Like you've stopped over the years or? I just was never really interested in it. Like it feels like the juice isn't worth the squeeze. Yeah, it's pain in the ass. I have invested in it quite a bit and I have a very connected setup. I've got cameras. I've got thermostats. I've got hubs. like we we have that but it's all it's all just different platforms and it's like a point solution so i meant like trying to connect everything in some kind of harmonious way just felt like a mile too far yeah so so i did that and i would say it's it's pain in the ass i think to your point and it's brittle brittle yes yeah it's like doing an auth connection to like linkedin right it's worse than that yeah it um i actually had a huge problem with it recently where like i was trying to add all these cameras but they weren't working properly and so i like removed some cameras as a way to almost like reset it and it just blew up my whole setup there was like a bug and it just zapped my whole setup meanwhile i have cameras like hanging off the roof that you have to physically scan to reconnect them like it's yeah i had to get on a ladder like it was you're a better you're a better person than me i just would give up it's just not worth it next thing we should cover 2024 sort of surprises jensen is a rock star like like possibly we've never seen a tech ceo be a rock star before this is a guy who's literally signing uh autographs like on on women's physical bodies like i mean i've never seen a tech ceo be a celebrity or a rock star like this guy i mean even zuck you know zuck zuck has become very very uh very what do they call uh base base zuck um but jensen arguably is even more of a celebrity than zuck is this this This is out of nowhere for me.

29:35Yeah. I mean, I think it's a great story. And I think when he gives his interviews, they're so consistent and authentic. And, you know, if you listen to any of those interviews, like they're very, like his philosophy just seems very consistent. He's been doing this a long time. And I think he leans into the, it appears like, I mean, obviously he's got publicists now and people helping him but it seems like he leans into the fun of it but doesn't get i think so yeah yeah i mean but like you know like you said these kind of rock star viral moments where he's had he you know he's pictures of him flying around the internet whether it's his jacket or you know signing people's bodies or whatever but i also don't think he's he gets enwrapped in the in the fame part of it where where like his humility doesn't get um overtaken you know yeah do you think how much do you think is intentional about his persona now his public persona yeah i don't think you get up there and do all these interviews and like be calm and say these things that are your philosophy without it rooted in something deeper yeah you know i think people would see through that.

30:55And it's hard to find other CEOs. Like, I think when you hear like a Bezos interview and he goes into like how he designs a team meeting or something, you're like, okay, this guy's been through a lot of team meetings and really thought about how to do it. Like, can't really fake that, you know, in terms of like the leather jacket or the rock star moments, I think of it, maybe it's charitable that he's just like leaning or accenting into what he's already doing and just kind of layering in a little more mustard. But that that's sort of how he is. I mean, how much is this the sign of some sort of.

31:35You know, bubble or, you know, stock price sort of inflated kind of mania. I mean, a couple of years ago. There were there were there were crypto celebrities that are now in jail, right? You know what I mean? Yeah. Yeah. I think because Jensen's been doing this for 30 plus years, like one of my mentors just said, like, he's been very successful, you know, like multiple companies he started, some in public, some were acquired for huge amounts. So he's just been around the block. And so he said he's seen people get obscenely wealthy at different ages. And he says, like, when people get obscenely wealthy, when they're really young, they're more likely to be a jerk.

32:23But if someone gets it late 40s, 50s, 60s, they're a little more mature. So I do think some of that is at play here where it's just compounding at the end. I also think of Jensen, and again, maybe I'm giving him too much credit, as someone who's trying to make sure the mailroom guys and the secretaries and the people that work the grounds in NVIDIA go home with eight figures. and I think he takes that pretty seriously given his roots I also think we may have done a pod like early this summer where like the bloom was coming off the rose of AI just like six months ago a little bit where people were starting to like slow down and back up and they just blast the estimates every quarter you know um so so they are the straw that stirs the drink now in terms of this industry and you know if you're thinking about the risk like i don't think of it as so much a bubble i think even more is like the risks are greater given the size of the mountain so like there's the china monopoly you know anti-monopoly sort of anti-trading risk there's the there's a risk around like hey if everyone you know you have this advantage in in chips like are other people going to be building chips um the data center wars right um government risk like maybe they become so big and powerful that the government sets eyes on them like that's how i'd be thinking about it um but but the chips part feels far away the China stuff I'm not qualified to opine on.

34:18I don't know how that will work. I don't see the government going after NVIDIA because I think we're past that era now. I think people want to let this stuff loose. And if one company becomes a$10 trillion company, so be it. Right. Amazon will be a$10 trillion company. NVIDIA probably will become a$10 trillion company. Google will probably be a$10 trillion company. So that's the trajectory we're on where the overall, the flood of money around the world is coming here. Markets have basically survived into a soft landing. Like two, three years ago, we wouldn't be talking about all this stuff. We were thinking the world economic engine would die.

35:04And then basically, you know, we're entering, I think people after a number of years of like regulatory crackdown are sort of the gloves are off now. And so it just feels like there's going to be a lot of room to run. So maybe a new kind of bubble. I don't know. But that's where fortunes are made. Good opportunity to transition to a little bit on the politics side. You know, Trump won. We don't need to necessarily get into whether or not that was or wasn't a surprise or like we don't need to get in like personal politics. But maybe sort of the implication of that on kind of tech and the tech takeover of Washington, D.C.

35:50Was this a surprise? And by the way, you know, obviously lots of people from our industry now now moving into V.C. We've got David Sachs, new A.I. czar. We've got other folks from the industry moving in. Congrats to all of them. But yeah, was this a surprise to you, this sort of tech takeover of DC? And what do you think we have to look forward to? I would characterize it a little bit different. I don't think of it as like a tech takeover. I just think of it as like a reorientation around how the government, from the viewpoint of the United States government, their stance towards enabling, supporting, accelerating technologies, both for the domestic economy and increased opportunities for folks and also internationally as a competitive advantage and security posture.

36:46Okay. So it feels pretty clear and like an apolitical statement almost to say that like that has changed, felt like it's changed 180. And I think that it has unleashed a lot of enthusiasm around tech from when before it was sort of like, I don't know what the right word is, but like gun shy, you know, not really sure how to interact with the government. So will there be negative effects of this shift? unsure unintended consequences. But like, overall, it seems very positive for the economy. It seems really positive for proliferation of technologies. Again, some of those could have, you know, in the short term, adverse effects like robots and AI and labor force stuff, or call centers, etc.

37:44Or crypto, right? Who knows? But then also, I think it's a real, It's a more real politic view of how the world is interacting and how states are interacting with each other and the type of warfare that is now being conducted or the type of espionage that is now being conducted. It seems a more real approach to those situations. So I think that those markets for entrepreneurs, for investors like us, will have a lot of room to run if that kind of openness stays now. Because it's actually very tough, regardless of what political candidate you're for, which party you're for. Or it feels like that genie is now out of the bottle and it would be very hard, especially from that international competitive point of view and the security posture point of view to actually go back.