Deirdre Bosa on Vibe Coding & Vibe Shifts In Silicon Valley
30:35 Share

To be invested with Betterment is to believe in better. Whether it's saving for today or building wealth for tomorrow, they help people and small businesses put their money to work. They automate to make savings simpler. They optimize to make investing smarter. They build innovative technology backed by financial experts for anyone who's ever said, I think I can do better. So be invested in yourself, be invested in your business, be invested in better with Betterment. Get started at Betterment.com.

Welcome to the Risk Reversal Podcast. I'm Dan Nathan, joined by Deirdre Bosa. She is the CNBC host of Tech Check. Debo. Hi, I've missed you. It's been a little bit of time. It's been way too long. It felt like Q4 earnings season lasted three months here. And so I feel like we're just about to get into Q1 season. So you and I are going to have plenty to talk about. But I got to start here.

All right. I just got a link to CNBC's tech check this week. The title is AI's vibe coding era, how the shift to apps changed the race. Now, this is 40 minutes long. Most of your tech checks are what, 10 to 15 minutes, maybe as high as 20. Well, so we shift it because they're now longer, 30 to 40 minutes, because we're interviewing some of the most interesting people building in the space. We tack that on the sort of the

piece, the analysis piece is about 10 to 15 minutes. And then we just have all these great interviews that are worth a listen. This was so good. Okay. Like literally you talked about 20 different companies. You had interviews with like a dozen of the CEO founders of these companies. I literally learned so much. These are companies that I guess, unless I'm like you out there and SF and in the Valley, I just don't know about a lot of these companies yet. And they are raising money at huge levels and there's tons of enthusiasm about it. All right. So let's take a step

back and let's talk a little bit before we get to what vibe coding is and everything like that really the crux of this story is that we started talking about ai rappers after these foundation models that everyone got really excited about after chat gpt in late 2022 and there was a bunch of competitors right anthropic and a whole host of other ones obviously google's been working on this

but then they moved on to wrappers, right? So this was like companies that were building on top of these models. Let's talk about that really quickly. And then the evolution from wrappers, right? To this whole idea of apps.

Well, so let me start by this. There's sort of different stages whenever you see a big platform change, be that the internet, be that the mobile era, and now we're in AI. And it seems to be following a similar pattern. So first, it's the chips, right? The semiconductors, in AI's case, the GPUs that get a lot of the attention and a lot of the investment. So that's NVIDIA, AMD, some of the others. And then it's the infrastructure layer, right? So in AI's case, it's

Google, Amazon, Azure, Microsoft Azure, that's building the data centers. And then it's the applications that sit on top of all of it. So we looked at this topic and we said, why isn't the AI trade really working this year in public markets? I know you've been following it very closely. We've seen NVIDIA, I think it's down now 20% or something on the year. We've seen Microsoft, Google, Broadcom,

Amazon struggle and it feels like those first few phases have tired out a little bit. But what is absolutely booming right now from my vantage point in the Bay Area is the app layer. And these are real world use cases. And in many of these cases,

They're more monetizable than the app layer, right? Which requires, excuse me, the infrastructure layer and the GPU layer, which requires billions and billions and billions. And the model layer, by the way, billions of dollars in investment. These are wrappers. They were kind of called disparagingly wrappers at the beginning of the AI race because people would say, oh, they don't actually do anything. They don't create the models. They just, you know, leech off of the models from open AI and anthropic, etc.,

and put something on top, there is no moat. But what has happened over the last few months and particularly because of DeepSeek making models so much cheaper and more efficient to build, they have got a lot of the attention and you're exactly right. They've gotten a lot of the investment as well. So while the AI trade kind of stalls out in the public markets, we've seen these app AI

AI native app companies just take off in the private market. All right. So before we get to some of those and how they make money and who they disrupt and the like, let's take a step back to what you just mentioned about the AI trade stalling out a little bit in the public markets. And when you think about the market cap that was accrued in just Microsoft, Google, Nvidia, Amazon, Broadcom, as you just mentioned, even Tesla could throw that in there. Apple,

at one point last year, there's a lot of enthusiasm about Apple intelligence. I mean, this was literally trillions of dollars in market cap that was gained up, you know, and it was about the, the picks and shovels about the infrastructure build about the excitement about how, you know, Google cloud and AWS and Azure, we're going to monetize, right. All this compute in the light. And so to your point, like the fateful eight, as I like to call them, um, you know, they've really hit, um,

you know, a bit of a buzzsaw here, right? And so if I think about this transformation, like you've talked about where all the innovation now is happening in the private markets, let's say on the app layer and the like here, it's like, okay, well, at what point does the rubber hit the road? Do these companies start making money? Do like investors start saying to themselves, I can see a path towards two things.

IPO for a lot of these companies or strategic M&A, which might be a thing in this new administration with deregulation. Because like if you can't express these views in the public markets, you know what I mean? And now we're starting to see these valuations skip up. Like it seems like there's a disconnect here. But the model is totally different. So in the public markets, you have the infrastructure companies and the model builders, right? Like how much money is

are amazon azure amazon aws azure google right we talk about this all the time the capex numbers right they just have to invest so much money to host all of this compute power um they're not bringing it in in the same sort of any kind of ratio that's you know justifying all that spend for investors

The app companies, they don't have those costs. They have much smaller costs. And as the cost of using AI compute comes down, as we saw through DeepSeek, they can just bring money in basically, right? Of course, they have to get scale. They have to spend money to get customers. But they're monetizing much quicker because they don't have those upfront capital costs like the public companies do.

So, for example, I know we'll talk about this, but one of the hottest companies in Silicon Valley right now in all of tech is a company called Cursor. Sorry, the parent company is called AnySphere and the product is called Cursor. It's an AI first coding company. It just sits on top of other models, right, made by I don't know which ones exactly, but, you know, by a chat by an open AI or by an anthropic. They don't have to build those models. They just charge a subscription.

So it's not like open AI, which is losing billions and billions of dollars a year because they're creating the models, which are so expensive to do on the back end. They're just bringing in subscription revenue. And sure, they're offering a freemium model, but they're monetizable. So if and when you see these companies go public, it's going to be very, very different. But that's a huge if and when.

Yeah. And I guess as the fast money guy who likes to kind of trade these stocks and the options on them and kind of think about the stories and the like here, it's just getting to a point where, you know, a lot of those names that we just discussed in the public markets just aren't that much that they're not interesting anymore. Right. And so. OK, so they may be interesting for another reason, though, because they may be getting displaced by these AI native app companies. Look at a Microsoft and GitHub copilot. This is exactly who Cursor confronts.

confronts, and by the way, Replit and Windsurfer, these AI first coding companies.

So it may not be sort of a bullish scenario, but you might look at these app companies to think maybe I should get out of Microsoft. So if you look at what just Google did on the acquisition front with Wiz, they paid $32 billion for this company that is basically a cross platform security company and enables, you know, customers who actually have workloads on multiple clouds to kind of, you know, scan their security needs and the like here. And it seems like a great acquisition for GCP because

In general, they are a third player behind AWS and Azure, and it might give them more optionality, right? But if you think about the potential for a deal like that to happen, then you can say to yourself, well, if Microsoft thinks that they are losing share to a cursor, might they pay? I don't even know what the last valuation is, $10, $20?

$20 billion or something like that. And again, you know, and in your tech check video, you kind of highlighted, I think you were highlighting in an information article, just some of the multiple sales of these private companies, how they've been skipping up higher, specifically in the app.

layer. So I wonder if all these folks now start lining up, who do we need to better compete? And they start looking to kind of buy these companies. Do you get any sense out there that there is an anticipation of a wave of strategic M&A? Because a lot of these sorts of founders, they're looking at some of the valuations that are skipping up higher, five, $10 billion at a time. They're like, you know what, why would I sell out now when I have the opportunity to become the next open AI?

And that's really the mentality. I mean, the founders here are working so hard. You hear stories about well into the night, sleeping at the offices, vibe coding through the weekend. And, you know,

Doing an IPO isn't what it used to be to here in the Bay Area. I've had so many discussions about this lately. But there's sort of the idea that you build for the long term and that's easier to do in private markets. M&A is interesting. I mean, everyone has a price, right? Actually, you know, I was at a J.P. Morgan's Tech 100 event.

event last week in Yellowstone and the Wiz co-founders were there. And it was interesting to see them with all of these public company CEOs as well and sort of talk to them about the decision to go with Google. And I mean, in some cases going with a mega cap, right, it hyper charges your growth as a startup if you are allowed to be in there and operate somewhat independently. So maybe it's a path for some of these. Everyone has a price.

Yeah, no doubt. You know, it's interesting when you think about just, you know, some of these folks like an NVIDIA and, you know, kind of the investment they made in CoreWeave, you know, from the get go and the IPO, which happened on Friday afternoon. I think we could all agree there's no real innovation there. Right. It is really an infrastructure company and it's meant to kind of bridge a gap. But it's also one of these situations where you start to wonder to yourself is like,

if NVIDIA was not willing to back this company, might they even exist, right? Might they have access to the GPUs that they used within these data centers that they lease and they had the ability to kind of, you know, take advantage from a pricing standpoint of a lot of these companies that needed the compute. And so, you know, when I look at a company like that go public and we can all agree there's no innovation, it really is about access to, you know, some of the hardware and then building it out in what was a breakneck speed

of adoption of this technology. But then on the flip side of it, and you've been reading all these stories and you've been reporting on it, is that without the kind of, I don't know, use cases and the return on investment, we're starting to see a slowing of demand. We're seeing a collapse in pricing and the like. And that's, I guess, where you go from these cloud-based situations, these infrastructure names to the app layer. So talk to me about what are the needs of

a lot of these like a replet or a cursor or whatever, well,

Who are they partnering with? And are there some similar sort of arrangements like NVIDIA backing like a CoreWeave? And NVIDIA have made dozens of investments, probably in some of these companies too. Jevons Law stands, right? The idea that Satya Nadella brought up right after sort of the deep seek panic. Are we going to need less compute power? And I don't think that's true at all. You're going to need more as more people get attuned to AI first coding or vibe coding, more people are going to use it and you're going to need more of that. So, I mean,

But you said it, the costs are kind of, or the price of compute is collapsing if you're okay to use a deep seek or you're okay to use a model that, you know, can operate more efficiently, which a lot of these models can. I mean, a perplexity is another great wrapper company that at its beginning was sort of looked down on because it was a wrapper, but it's

It's now creating its own models, being able to fine tune open source models, and that is creating a moat for itself. Yes, these companies still need to figure out a way of how to find profitability and monetize, but I think that path is going to be easier for them than it is a frontier model company. The other thing I would say is that the AI trade is fatigued in US public markets, but

It's really taken off at the app layer and in China. And these are public companies that people can invest in. Alibaba has just gone hyperbolic this year. And there is sort of they've captured it maybe a little bit late to the game. But I think that that's somewhere else as well that investors are very interested in. You know, it still has room to run.

Yeah, you know, it's funny. The Chinese angle of this is pretty interesting because I think before DeepSeek in late January, there were not a lot of folks that were focused on this. And then when you think about, you know, Alibaba, you think about Baidu, you think about Ant, which just released some models that all kind of have a little bit of the same...

tenor, if you will, of the deep-seek situation. There's a lot of open source. There's a lot of using Chinese-owned technology. If deep-seek brought us anything in the export bands of our high-end GPUs and the like here, is that we're forcing them to innovate in some way, shape, or form. And then you start thinking about

you know, the trade war that we're going to be in and the kind of difficult stance we're going to take with China. And you think about not just our difficult stance with China, but with some of our biggest allies, whether it be the EU, Canada, Mexico, might China be setting up at a time where President Xi has just kind of brought back into the fold a lot of these kind of tech national champions and the like, might they be creating a

digital belt and road for their own cheaper models that are open source. And I'm just curious, cause I know you spent a lot of time in Asia before you came back to North America. Is that something that you are hearing about by folks like outside of the U S.

Absolutely. I mean, go back to what we heard a million times post DeepSeek, which is necessity is the mother of invention. And by the way, sort of the Chinese approach to developing AI, it's similar to that wrapper model. They're not trying to push frontier models. They're using distillation to just get to the frontier and then build use cases on top of it, which is why it's become a compelling story. I mean, when you look at Apple versus the

smartphone AI companies in China. They already have AI embedded and there's been surveys done that show that Chinese consumers care way more deeply. They're prepared to buy new phones if it has AI features more so than American consumers. So again, it goes back to this idea of use cases. They don't need GPT 4.5, which is beating all the benchmarks on math and coding. They need usable AI and something else

R1 from DeepSeek and soon to be R2 is good enough and gives them everything they need. And that's probably true for like 90% of people using AI. You know what's so interesting? You just brought up Apple and I've kind of been making this point for a couple years. In the first time in a long time, I think Apple...

Apple's innovation, a lot of folks have been saying this for years and years. The iPhone really hasn't been innovative whatsoever. Some of the tweaks they do year after year, it's not one of the reasons why you'd upgrade the phone. The upgrade cycles have been extended. That's one of the reasons why people were so excited about the promise of Apple intelligence, but there was no there there, right? And so now you think about this, we're like almost a year on from WWDC where they introduced

apple intelligence just last month they kind of pushed out indefinitely the upgrades there's nothing there siri has been laid bare you know what i mean their in their inability to kind of integrate this software technology as they say on the edge and stuff like that is really disappointing and i kept on hearing last summer there's 200 300 maybe even more iphones in china that are two three years old that need to be upgraded

But the Chinese, if you just think of it from a nationalistic standpoint, the way that they might be choosing to make these sorts of purchases, their use of things like WeChat, these super apps that exist, the way in which they're integrating AI across those platforms, you don't need a really expensive piece of hardware anymore to get the uses that you like. And I just wonder...

If this is a huge moment for Apple that they basically missed the boat. And the last thing I'll just say about Apple is unless they partner with an Alibaba, and I think they've signaled that. I don't know if they've made that decision outright yet, but they will not have Apple intelligence inside the Chinese firewall. So Apple seems to be totally left out in the cold here.

I believe they have signed on to a partnership with Alibaba, but it needs to be approved. But it's also like, what's the holdup? They have to rethink their strategy. If they insert Alibaba's QWEN model in the same way that they're leaning on

OpenAI's GPT in the iPhone here, which is to say it's not a good experience. There's a lot of friction. You have to wait until Siri can't answer your question and then prompt you if you want to go to chat GPT. People don't want to use it like that. So, you know, I'm with you. As long as Apple takes to figure this out, there are fierce, fierce competitors in China, like a Huawei.

that is integrating AI in a far more seamless way. So, you know, even months can do damage. We've seen some of that. Apple, you know, is losing market share in China. It gained some back a few quarters ago, but this is, you know, one of the fiercest races there is for this mega cap.

Imagining a better future is the first step. Investing that future with Betterment Advisor Solutions is the next. Whether you're launching your own practice, looking to streamline client onboarding, or just searching for efficient ways to scale your firm, Betterment is here to help. They automate to make tax optimization simpler. They provide support to make administrative tasks easier. At

Betterment, they're building innovative technology all for anyone who's ever said, I think I can do better. So grow your RIA your way with Betterment Advisor Solutions. Learn more at betterment.com slash advisors. Investing involves risk, performance not guaranteed.

You know, I want to talk about Huawei for one second because I know this is probably a company that you're familiar with. And it wasn't just on the smartphone space. And they've been kind of introducing some high-end phones that have really been taking it to Apple, especially when Apple is not innovating and they don't have this kind of AI, you know, kind of built within the fabric of the device. But they also announced an electric vehicle.

that is coming right at Tesla. And I'm not going to ask you too much on that. I don't expect you to kind of be all over that, but I just feel like there's a moment here for Chinese innovation coming at some of the things that, you know, US companies had been at the forefront as it relates to innovation. And we've really gotten pretty stale, which is maybe one of the reasons why all of these tech CEOs here in the US have kind of lined up behind the Trump administration thinking about, wait, we've been doing all this innovation here.

And yet the government, which, by the way, started under the first Trump administration, when you think of the FTC and the DOJ investigations, might they be angling for a little more room to kind of do some of the things, I guess, move fast, break things and ask questions later where the hell they used to say about it. I'm just curious if that's something that is kind of going to be something that Washington champions in a way, because as they think about the innovation that's going on in China.

Yeah, I mean, there's two sides of this, right? I think you could say that the U.S. leads in terms of

software and software technology, right? We're pushing the frontier in AI. But when it comes to manufacturing, right, you look at sort of the robotics race as well, a unitary out of China, the cars that, by the way, not just Huawei, Xiaomi a while ago, you know, released a car that was pretty well reviewed considering it's a smartphone company. And you contrast that with, you know, Apple's secret car product and how it led to nothing. I mean, there's a huge gap here. And, you know,

Yeah, it's going to be hard. The mega capsule, what I will say, still ahead. Our American tech champions ahead in terms of the most advanced technology, even in the very critical chip space, right? Huawei is also, in addition to its cars and smartphones, getting a lot of attention for its...

it's chips, it's AI chips, which, you know, you know, a year ago was thought of as very, very much far behind. Still, I can't tell these days, but seems like it's, you know, catching up quicker than it, than it ever has. Yeah. And, and going back to some of these, um, you know, kind of, uh, app layer names, I'm not going to call them rappers here. Um,

Are folks getting a little kind of nervous about some of the valuations? Because it just seems like this is stuff that we've never seen before in the private markets. And you mentioned earlier, there's kind of less excitement about going public, let these companies innovate in the private markets. They're not having a hard time raising equity. And if it's debt they need, then they can go and get that too and not be within the public eye.

if you like. And so when I think about that sort of mantra, you could see this kind of private for longer get to a point where maybe these valuations are really on, you know, they're just they're not able to continue at a pace where they'll ever get to public markets because sooner or later you will have to do that.

You can only raise so much equity and so much debt. And a lot of these VC funds are going to need to recycle it. A lot of these early founders and the like here. So I'm just curious, like how folks out there are thinking about the fact that we have hundreds of billions of dollars in valuations that

are not going to be tested. They're not marked to market. You know what I mean? You can go through a difficult economic patch and you can see like in the last three months, all of the hyperscalers down at least 20% in the markets. And that, you know, that is something that can be sobering to the investment environment. If you're not paying attention to it in the private markets, you might end up with a scenario like we had from late 21 into kind of early 2023, really until chat GPT changed the game again for the private markets.

And set everything off again. I mean, this has been the story of my 10 years reporting on technology in the Bay Area. It's always been frothy. I came here when Masayoshi-san first released $100 billion into the startup ecosystem and bid up companies like WeWork. We know what happened there. I mean...

OpenAI just raised $40 billion round, the largest ever venture round in history. Yes, some of these app companies, even if they have $100 million in ARR, annual recurring revenue, are getting valuations that are just massive. And by the way, OpenAI's valuation, even with this latest fundraise, looks like nothing compared to ex-AI Elon Musk's company. But this has always been the case. Even in 2022, when you saw that big correction in the public markets,

Venture capital doesn't have to mark down its portfolio. They can just sit on it. And these startups don't have to raise another equity round and hold on to their valuations. Do people believe it? Not really. But some of them will go bust and some of them will last and pivot to AI or the next trend. So it's always been this way. And I haven't seen any change.

Certainly in this environment, there's just money being thrown at some of the buzziest companies like OpenAI, like Anthropic, like Perplexity, like Cursor. - You know, it's interesting, and again, I don't have a lot of experience with some of those other apps, but when I think about Perplexity as a name that we've kind of,

You know, I've been tracking for a while here and it's one of the companies that I actually use their product and I look at it and I look at the way that they've innovated on the user interface front. I look at it compared to like a Google and the experience there. It's just kind of night and day and they're adding new features. It feels like on a weekly basis and I start saying to myself, oh man, well, if Perplexity had a browser, if

perplexity had maps built into it. If there was a whole host of other things, which I'm assuming are all coming. I mean, you think to yourself, you know, what the hell is Google doing? Because, you know, Gemini might be fine. It might not be, but if I'm not using it, if I'm not going to Google and doing the searches and I say to myself, okay, like this is the first time in a long time, people can really consider Google losing, you know, some serious market share as it relates to search and advertising.

there was a perfect example of Google's sort of inability to tell its narrative or give its products to users just, you know, over the last week or two. Released Gemini Pro 2.5 first in experimental mode. And, you know, because I'm

So in the tech community, I was seeing people play around with it and it looked really cool. I went in and tried it myself. Just a terrible user experience. First, you got to go to AI labs from Google and then you got to set it to the right model. And then every time you do something, you have to reset it and it wouldn't take kids' photos when I just wanted to make a photo of my kids with a new puppy to kind of

play a joke on them. Couldn't do that. And just, you know, just so many like points along the way where I got fed up and said, this is like not worth it. And then what was it like a few days or a week later, OpenAI comes out with its new image generator and totally sweeps the internet.

with Studio Gilby, which is kind of like a silly example, but it does say a lot. Like it went absolutely viral. And now you're seeing these interior decoration use cases and so on and so on. And Google had it first. And there's been so many instances of this over the last year or so of Google being first, having great technology and just not hitting the moment with it only for OpenAI or someone else to sweep in. And it's like,

It happens so often that it's hard to see because Google has the technology. They have the talent. They have Demis Hassabis, who is a Nobel Prize winner, but they just can't get product from lab to user. That is, you know, the major struggle. And why, you know, maybe the valuation is justified, lowest of the Meg 7.

Yeah. And you just mentioned, you know, as long as you've been covering tech for 10 years and, you know, I go back to the late 90s when I started covering tech and trading, I was at a hedge fund. And, you know, that was just an amazing time and the optimism in and around the Internet. And there was no end to the valuations, you know, given valuation.

Like these companies that you're talking about, the way in which they are bringing on revenue. And I know that was a skepticism maybe like a year ago, but some of them now are really doing it, right? And that ARR is reaching numbers where a lot of bankers would say, hey, you should go public right here. I mean, it's getting to that sort of scale and they can see a path to profitability much sooner, let's say, than you could 20 years ago for some of these models. But then I

think about, again, and I just want to kind of maybe bookend this a little bit, how we started in the public markets is like, when you think about NVIDIA, it went from an early 23, I want to say a $300 billion market cap to at its highs last year, three and a half trillion dollars, right? If you think about Microsoft, if you think about Google, you think about Amazon, you think about Meta, and even Apple, like we just mentioned, you know, these were the perceived

early winners. Apple got there late to the game, but if we just, you just went through Google. All right, let's talk about Microsoft. They had the catbird seat as it went to open AI and their inability to kind of the co-pilot thing to kind of take hold and GitLab now having a difficult time with like cursor and some of these other competitors.

Meta, who's been able to use this technology to serve better ads and it's clearly been something that's helped their own business, but they have Meta AI, which they are not monetizing in any way, shape or form. Amazon, right, AWS beneficiary, no innovation there. They have to kind of license Anthropic. I mean, the list kind of goes on and on here. So the biggest winners to date, which have all stalled out, are not innovative.

whatsoever. So it leads you to believe there has to be strategic M&A. And I couldn't think of a better time to do that than right now under the current political environment.

Well, I will tell you, there's a lot of very interesting companies for the taking. Another one, too, is Sesame, right? And you just think, why can't Apple get Siri? Why can't it improve Siri? Why doesn't it just acquire Sesame, which, by the way, is just a great voice assistant? So natural. I don't know. We'll see. I mean, I'm not totally convinced that...

the current trump administration is going to you know just let m a fly right now wiz will be a really interesting indicator obviously they're confident three billion dollar breakup fee by the way for that one so i'm optimistic about that deal i think it's a really smart deal it's not a sexy deal for google um you know what i mean but it's one of those things that if they're really thinking critically ahead and how they're going to compete in this space i think if gcp is going to become a catch-up play to aws and azure i think this is an important point

piece about it. All right, Dee, I got to say this. I really appreciate you being here. We had a whole rundown set up, but then until I saw your tech check, I'm like, we got to dig into this thing. We're going to put it. No, I loved it. We got to put it in the show. Thank you. It's most interesting to me. And it feels like that's what everyone in tech is talking about right now. So thank you for

I learned a lot. I hope you guys keep pumping these out. It's AI's vibe coding area. We did not explain what vibe coding is. There might be some people in the audience that are like, what have you been talking about? So how do I explain this? Vibe coding is when you focus on the vibes and you let

AI do the coding. I love it. I love it. It sounds ridiculous, but it's the hot thing. Actually, my nine-year-old, I put him on a course now on Replit, so he's going to be vibe coding soon. I love it. I love it. I love it. All right. It's AI vibe coding era, how to shift to apps, change the race. This is Deirdre Bosa, CNBC's Tech Check. I really appreciate it, Dee. We'll see you really soon. Thanks, Sam.