Why Alphabet is Doubling AI CapEx, Nvidia Delays Gaming Chip, OpenAI Hires AI Consultants

0:13Welcome everyone to the information's TI TV my name is Akash Pasricha it is Thursday February 5th First up this morning, Alphabet reported earnings. The company reported strong revenue growth figures, but its extraordinary capital expenditures are what investors seem to be reacting to this morning. We will break it all down with a venture capitalist. We'll then turn to some exclusive reporting the information published that NVIDIA is delaying its new gaming chip due to a deepening global shortage of memory chips prompted by the AI boom. I'm then sitting down with David Marcus, co-founder and CEO of Bitcoin infrastructure company LightSpark and formerly president of PayPal.

0:53We'll discuss the sell-off in crypto, what David is building, and also an essay that he recently posted about where PayPal has gone wrong over the past few years. And finally, we will wrap the show with some more exclusive reporting from the information about OpenAI's hiring push to boost its enterprise sales. It is a big show, so let's get right on into it. Google reported its quarterly results last night. Revenue grew 18%, driven by acceleration in the company's search and cloud businesses. But the company's extraordinary CapEx, which it says will double this year to at least$175 billion, seem to have spooked investors a bit.

1:33Shares were trading down at the open today. To break it all down, I want to bring on Sonia Oja, partner at Bain Capital Ventures and author of the popular Change Constant newsletter on Substack. To break it all down, Sonia, welcome back to the TITV. It's great to have you here. Yeah, gosh, great to be here, especially today. What a bloodbath in the markets, huh? It's been a story all week. Gosh. Truly. And I think you can accuse the public markets of being many things, but patient has never been one of them. And this is definitely the week to feel it. I agree with you. I think the Google earnings were very positive.

2:09Obviously, the top line was, but if you look under the hood, there's so many interesting data points. Google and Gemini Enterprise launched four months ago, and they've already sold 8 million paid seats. They signed more deals, more than a billion dollars this year than the prior three years combined. The stat that really stuck with me is that their AI customers use about two times more Google products than non-AI customers. So search is accelerating. Cloud is accelerating. This is clearly a business that has a lot of tailwinds behind its back. That's really regained the AI narrative. It's just all of these big tech companies are in this sort of damned if you do and damned if you don't situation.

2:53You know, if they don't spend on CapEx, they're going to be seeding market at a very crucial moment in this transition to their competitors that are definitely not going to slow down. So it's a prisoner's dilemma sort of thing. And from the public market perspective, if you believe that we're seeing a paradigm shift, which given how investors are trading SaaS stocks these days, clearly they do, then if you don't invest to meet this future, that's negligence. That's not prudence. Being early and wrong here can cost you money, but being late and wrong can cost you the company. So I think I'm positive on what Google and Microsoft are doing to set themselves up for the future.

3:34And I think once these quarterly issues are over, we'll be back on plan, hopefully. So if I hear you correctly, basically what you're saying is, and loosely, the stock market doesn't make any sense. I mean, that's known to just about everybody in some ways. But you're saying what we're seeing with the drawdown enterprise software stocks, that is the fear of AI. And here you have companies like Google that are very clearly investing in that same AI, similar to what Microsoft did last week. and you have opposite reactions, which is not making a whole lot of sense to you. Yeah, these companies are extremely well-placed to capitalize on this trend and they're putting the money behind it and it's not ahead of demand.

4:15The demand is here. What's also very interesting is that the cost of serving is going down a lot. So the unit economics on the spend are getting better with time. Sundar mentioned that I think the serving costs are down 78 % per unit of intelligence. So I think the economics will catch up and get there, but patience needed in the interim seems to be severely lacking in the market at this point. Right. What about the TPU story here? That was not as much a focus in this quarter as it was in prior quarters on the call. I think they did talk about it as being one reason why we saw that tremendous growth in Google Cloud and a big value proposition.

4:54But the answer that executives gave to the question around the long-term strategy around using TPUs, outside of Google Cloud, outside of the company. We didn't get a lot of clarity there. What did you think of that? I think they maybe just didn't talk about it on the call, but in the ecosystem, we hear a lot of great things about the performance that TPUs give and the fact that they seem to be stepping out of just Google Cloud and selling it to the Neo Clouds, and people are beginning to use it and really like it. I think Anthropic obviously is the case study of having used TPUs and really benefited from it.

5:35So I think TPUs will continue to be a very big part of the Google story and we'll continue to hear a lot about it. So if everything is going right here for Google and it seems like a lot is going right for the company, what do you think are the big risks now facing the company going forward? I mean, the thing is, AI is a marathon and the laps are very short. So I think the sentiment shifts like a pendulum every few weeks where for any given company, we go from like, it's so over to we're so back. Like right now, OpenAI and Microsoft are in the doghouse. Antropic and Google seem to be doing really well.

6:14I think we're one model job or one product launch away from that shifting. So I think it's too early to call it. Consumers are trying a bunch of things. Their interests are fickle right now. I think everyone just needs to keep their head down, execute. The dust is going to take a while to settle. So I think the biggest risk to Google is that their next few models don't land. And OpenAI, again, regains the narrative momentum and the customer interest. So it's no one's game yet. Every leadership position is borrowed, and we'll see how things shake out. But I think Google and Microsoft, the reason I'm very positive on these big tech companies is one, they're moving aggressively.

6:57But two, they have established data advantages. They have distribution advantages. And in a world where you think software is going to become easier than ever to create, distribution is going to become ever more important. Because I'm already seeing this in enterprise buying right now. Vendor fatigue is very real. You have five companies that are selling you very similar sounding things. Unless it's a step change function, you default back to whoever your existing vendor is and ask them, hey, what are you doing with regards to AI? So I think, yeah, a lot of roads lead back to the incumbents. And this is something that you wrote in your newsletter recapping the Microsoft earnings as well.

7:37You wrote here that the market is underweighting the application layer where Microsoft has a right to win. And the point you're making there is that where people have that user base already and have those applications installed, that will really give them a leg up with future applications. Yeah, I agree. I think incumbency has always had an advantage if you don't fumble the bag. And I think distribution matters even more now in a world where software is easier to create. And these incumbents are being extremely aggressive. So that's not something I'm very worried about, particularly for big tech.

8:10So let me ask you about the broader SaaSpocalypse now at large that we are seeing, because it's not just these hyperscalers that are walking a tight leash. It's really everybody. Look, we know the narrative here. Will AI replace software? That's the narrative that people seem to have suggested is bringing the market down. My question for you is really, is this fundamentally a new normal with respect to what the median multiple for what a software company is? I mean, at various points in history, I mean, look, when Snowflake went public, right? I mean, there was a multiple normal then. It seems to have come down.

8:53It's coming down more. I don't know. I'm sort of starting to think that in a world where these AI companies will go public, maybe in the next two to three years, is this making space for what people deem to be a higher multiple company and an old school SaaS company maybe isn't worth what it was, you know, a couple of years ago? I think right now the market is trading fear, not reality. Reality will diverge and we'll see the, you know, SaaS camp sort into winners and losers. And then there will be a re-rating of multiples in both directions. But I think right now it's too early to tell. Like, you know, again, I go back to you can't underestimate distribution and data as advantages.

9:35And yes, some people in the SaaS complex will ship a bad co-pilot on a 2015 architecture and call it their AI strategy, and they're not going to do well. But some people are going to move aggressively and foundationally change the product, and they will be in a very prime spot to retain that customer interest. So I think we're overreacting right now. This is a fear-based reaction, And that's not to say that there won't be losers and there won't be winners. But I think once that becomes clear, you'll diverge into those camps. I think this feels temporary. Great. Well, Sonia, I want to thank you for coming on the show.

10:20That is Sonia Oja, a partner at Bain Capital Ventures here on TITV. The information is exclusively reporting a major shift at NVIDIA. For the first time in three decades, the company is expected to skip its annual release of a new gaming chip. The culprit is the deepening global shortage of memory chips fueled by the AI boom. NVIDIA, of course, built its empire on gaming graphics in the 90s, but now we are reporting that it is slashing production of its current gaming line to pivot resources. Memory chips are kind of like the warehouses of data. They're essential for training AI, but they're also the backbone of your smartphone and your laptop.

11:00all. NVIDIA is using its limited supply of memory chips to fulfill demand for AI chips. As AI chips have taken off, the importance of gaming chips to NVIDIA's bottom line has dwindled. Now, the company could change their mind and release a gaming chip if the market improves. It is known for its flexibility, and moving quickly, we will see how this all plays out and keep you posted of any changes. Crypto markets have had it rough the past few months, and the sell-off has stumped even Bitcoin's loudest proponents. Meanwhile, there are still companies that are charging ahead to bridge the gap between the traditional financial system and cryptocurrencies.

11:38LightSpark is one of those companies. David Marcus is the co-founder and CEO of that business. He previously led Meta's major blockchain and digital currency efforts. David also wrote an essay recently on PayPal, the company he helped lead before Meta. I want to bring him on to talk about all of that. David, welcome back. Welcome to TITV. It's great to have you here. Great to be here. I say welcome back because you and I spoke years and years ago when you actually launched LightSpark. That was the big reveal. And here we are years and years later. I want to talk to you about the company that you're building.

12:11But broadly speaking, look, we have this crypto sell-off right now. You know, Bitcoin, the numbers are, what, 20 % down this year, down 40 % in the past six months. Mike Novogratz is out here saying Bitcoin wasn't supposed to act like this. What's going on in your opinion? Well, there are a number of things that are clearly happening. As you know, there was a massive margin pressure on 10.10 last year, and that had cascading effects across the market. It's unclear really how much the Binance failure actually had an impact on the market, but clearly, there's some of that at play. And then, you know, Bitcoin has become so big over the last year that it's now a place where most investors find liquidity, especially when the markets are closed at the scale of other very large assets.

13:07And as such, we're seeing it move with the market and sometimes even more so because of all of the margin that has been used to leverage buy basically more Bitcoin and more crypto in general over the last couple of years. So I think we're seeing a changing of hands. I remain very bullish on Bitcoin. I think, you know, I've always been a long term thinker when it comes to Bitcoin. And if you have that mindset, you should be long and buying more right now. Tell us about the company that you're building. And I want to ask you to explain it as if you were explaining it to someone who wasn't in the tech sector, what the company does.

13:47Yeah, so think of it as an internet for money, an open, neutral internet that moves money in real time, 24-7, irrespective of which payment system you're on, which country you're on, which day of the week or time of the day it is. And we're building this on top of Bitcoin and stable coins. And the way it works is that it enables us to connect domestic payment systems that have now become, for the most part, real time and free to one another. And those domestic payment systems that work really, really well, like you can send money in the US with Zelle, in Europe with SEPA, Instant, in Brazil with PIX, in India with UPI, all of these systems are real-time 24-7, but they don't talk to one another.

14:37The way that they talk to one another is through a web of correspondent banks and Swift, a combo that was invented over 53 years ago. And that still works like 53 years ago. It takes three days to clear. It doesn't work Friday after 5 p.m. or on weekends. That's what we call bank holidays, bank holidays. And so what we're doing is really enabling money to move in real time globally for businesses and people all around the world. Well, and you've got a number of different use cases and products for connecting cryptocurrencies to the traditional financial system. How big has the company gotten now in terms of revenue?

15:17So it's gotten really, really big in terms of payment volume and network effects. We launched Spark, our Bitcoin L2 last year. and I think you know it's on track for being the the largest volume on top of Bitcoin as an L2 even if you include Lightning and others in the next few months and and so we're seeing a lot of success with banks all around the world that are using the network to offer their customers a way to actually move money around like you know you look at SoFi most recently like if you're a SoFi client. You can now send money to over 2 ,000 countries that they've already turned on in real time 24-7.

15:58We have more and more large businesses that are platform businesses that are doing payouts to creators to endpoints all around the world, either in stable coins or in fiat. So the company has grown exponentially. I think our total payment volume grew 500 % plus last year and do you make money only is it is it uh based on the the payment volume only are the are these sort of a uh you know a software fee that the banks are paying you what's the business model yeah the business model is mainly a transactional uh business model where we take a few basis points on every transaction uh that uh that are on the network but there are also other fees uh and gradually we're we're packaging up more of the the different services that we sell more as a service.

16:45But the core revenue is really a BIPs on transactions business. And so as crypto volatility does what it does, and we've seen this drawdown in the last couple months, I wonder, does that affect the interest that you are seeing from banks and from institutional investors in bridging their systems to digital currencies? No, really not. I mean, we're fortunate enough that we're in a business that's completely independent of the price of Bitcoin. As long as Bitcoin is around and the networks that support stablecoins are around, we'll be fine. And the reason for that is that we're solving a real problem for real businesses that need to move money in real time or need to enable their customers to move money in real time all around the world.

17:30And so as long as Bitcoin works and it exists and other networks like Solana and others that support stablecoins are around, we're fine. Right. Well, I want to ask you about an essay that you published a few days ago about PayPal, which is a company you worked for more than a decade ago. The company, of course, appointed a new CEO earlier this week, and the stock has fallen dramatically since that announcement. And broadly speaking, it's been on a bit of a drawdown for the last couple of years since its pandemic highs, at least. You published this essay and you ended it by saying, the company that had every advantage and could have become the most consequential and relevant payments company of our time lost its mojo, its product edge, and its ability to compete.

18:17Why did you decide to write this essay now? Look, I mean, this is a company that I gave a lot to and that I care deeply about. I took the helm in 2012, early 2012. And at the time, the company was already like in that trajectory of having a lot of consultants, having offshore development, having lost its technology edge and its innovation juices. And, you know, I found a number of really great, talented people within the organization. We completely changed the culture. We were really on a tear. We acquired Braintree and Venmo. and the growth really accelerated at the time and it felt like the company was on the right trajectory.

19:04And after that, I feel like you had a succession of different leaders that actually just decided to just ride the wave that the company had and the network effects basically and try to optimize the company financially rather than continue to innovate and continue to push for products in the face of a tremendous amount of innovation, right? It's like, you know, this was the time where Apple Pay was really growing at a rapid pace, taking away PayPal's most valuable customers from checkout experiences. They decided to focus on unbranded checkout, which is basically just card processing, which is super commoditized and low margin versus the branded flows that could have been made better, faster, you know, more mobile friendly.

19:57And on lending, they completely missed the mark at the time where you really had Affirm and Klarna and Afterpay being built. And then, you know, they acquired a bunch of companies that made no sense to acquire or to be part of PayPal. And so all of this basically just eroded the company's place in a market that was being reinvented. And, you know, I sat on this feeling, being frustrated by the sideline for years, and felt like I really didn't want to do armchair commentary. And, you know, I ejected myself out of this company, and I felt like, you know, it wasn't my place to comment. But like, after 12 years, I felt like, you know, that was enough time, and I felt frustrated enough to speak up.

20:44I want to go back to the start, though. You know, the initial advantage that you think PayPal had that it didn't really make use of, what was that advantage at the start? Well, look, I mean, if you think about two-sided payments networks, there are really only a few, right? I mean, there's definitely, you know, Visa, there's MasterCard. Visa sits at the center between acquiring and issuing banks, issuing banks that touch consumers, acquiring banks that touch the merchants. Amex has the two sides of the network themselves as a first party. And then, you know, since those networks were built, PayPal was really the first and last shot at a company at building a two-sided payments network at a global scale.

21:26It had consumers, it had merchants, it could clear transactions between consumers and merchants in real time on its ledger. And so it had every advantage to build a closed payment network that was the most efficient. and you know Elon Musk always talked about about x and his vision for PayPal at the time x money at the time as like a giant database where basically if you have the largest database and you can move value on your own database then you're more efficient than everyone else and then you win that was kind of the original premise for PayPal and and they completely missed the mark Now, you talk about how over the course of the company's history, they prioritize payment volume as opposed to prioritizing product and the experience that customers go through.

22:18I wonder, was it your experience? Was that clearly the focus on the ground? And you weren't at the company over this time, so this is all reflection. but i guess was that explicitly the focus in the moment or is that something that in hindsight that appears to have been what happened i mean i i think you know i i don't know for sure because i wasn't inside the company but i think it's really clear that they focused on trying to do um more card processing uh and more unbranded volume as a way to show growth to the street and you know report earnings quarter after quarter and optimize for short-term uh upside uh and then of course The online shopping boom that COVID fueled clearly helped them hide the realities of losing share of checkouts to Apple Pay and others and also losing share to BNPL products and Buy Now Pay Later products and others.

23:16And so it looked better than it actually was under the hood for a while. right now if you you mentioned a few opportunities in your essay branded checkout lending buy now pay later which of those do you think could have been the biggest opportunity for paypal well i mean look and it's like really branded checkout like the two-sided network is really where where all of the all of the money was like and and you know i mean we had like a healthy a healthy tension between the card networks and paypal for the longest time because paypal used to to do something that card networks hated, which was called steering.

23:53Basically, you have a bank account that's connected to your PayPal account, and you have credit cards, and then you try to steer as many transactions as possible to bank, because then you can arbitrage the cost of an ACH pool that you underwrite for instant payments, and the payment that a merchant is paying you 300 bips or so for. And that was where all of the money PayPal was for a very long time. and I think the team at Visa masterfully played PayPal at the time to actually get them on a sort of gravy train for a few years for that to look good on the street but then force them to stop trying to get a better funding mix from PayPal's economics, which was also long-term, very, very bad, but short-term maybe felt good.

24:41But definitely Visa played that way better than PayPal did back in the day. so what would you do now i mean look i think that it's really clear that there's a governance problem at the company um and um and the board is a is you know it's it's more or less a you know a collection of professional board members some of them who've been there for a long time uh and so i think you know for for for anyone taking the helm there first of all you need someone who understands payments deeply i think that's like really critical and then like that person needs to have enough skin in the game and uh and have like close to dictatorship type of of uh authority to actually really revamp the company and rewire it from from the inside out which is kind of an open heart surgery um but uh yeah i mean it's not it won't be easy to to actually get someone who who has the skill and uh and and the right structure from a governance and incentive standpoint to actually go do that work did they ask you to come back at all no any interest no no i'm good like look light spark is on a tear light spark is we're doing great like it's amazing it's going to be an amazing company that's going to be foundational for how money moves in the long term i'm having a blast it's a it's a you know it's still a startup three and a half years in i'm having the best time so no interest let me let me ask you one more question before you go which is that look i mean the paypal's performance over the past few years i I mean, you look at the performance of recently IPO'd fintech companies like Chime, like Klarna, even Wealthfront.

26:17I mean, this is a group, too, that has not performed as well as I think many would have hoped since their debuts. What do you make of this broader fintech drawdown in stocks? Well, look, I think just fintech in general. Like for years, it used to be, I like to say that it's not the most glorious way to describe it, but it used to be what I think is lipstick on a pig, right? It's like it's still the old, bad, non-realtime infrastructure, and you're just putting a veneer of a cool-looking app on top of it without rewiring the underlying infrastructure. And so I think you can do that and get customers and decent economics, but if you don't actually have better rails for all of these things that are being rebuilt right now on top of crypto rails and blockchains and real-time settlement engines.

27:08If you don't do that, then your advantages are not going to be long-lived. And I think there are a few other examples of very successful companies like Revolut that has done really well, of course, and more than all of those, Nubank that I have the honor of sitting on the board of that has really core technologies and capabilities that I think very few other players have and have been able to really, really take advantage of. And so I think it's kind of a tale of two cities. You have those that have really the core technologies to have a competitive, long-lasting advantage, and those who've built on top of legacy infrastructure and are just offering a better consumer or business experience.

27:51Great. Well, David, I want to thank you for coming on. That is David Marcus, co-founder and CEO of LightSpark here on TI TV. Great to be back. Thank you. The Information published exclusive reporting on OpenAI's big hiring push to boost its enterprise sales. I want to bring on Shree Mupiti, who broke that story. She covers OpenAI and Anthropic. Shree, welcome back to the show. It's great to have you here. Excited to be here. What is this big hiring push that OpenAI is on specifically this time? OpenAI has been on a push to hire hundreds of what they call our forward deployment engineers. Essentially, these engineers are embedded technical staff that sit alongside large enterprise customers to ensure that when you're deploying an AI product, whether that's, for example, an AI agent, which OpenAI launched its frontier platform this morning, or other fine-tuned models based on its API models, it could actually support to ensure that these customers are using AI correctly.

28:55now forward deployed engineers i was just at dinner last night and we and and i i asked the group the question i said aren't these just consultants i mean you know like it's a it's a fancy term it's the people that go in and help you implement the software that you've bought help you integrate it into your systems you know is that just what it is really what it is essentially is it is consulting in a way, especially in that it's very services oriented. It takes a lot of sort of like handholding of the customer to ensure that the AI is correctly being deployed. What I would distinguish of like just a general consultant, for example, from Bain or BCG is that a forward deployed engineer has very specific AI technical expertise.

29:40And so they're the expert on OpenAI's products can ensure that, for example, that a customer knows how to use this API correctly and fine tune the models to their data. So they're the technical folks. But then OpenAI is also partnering with traditional consultants like those from BCG, Accenture, et cetera, to be able to actually provide the industry expertise that one might need to be able to go into these customers. And so this forward deployed engineering model has actually been pioneered by Palantir, which obviously had earnings earlier this week. It has done fairly well. So I think it's a model that AI companies are now starting to realize that they actually need people to help sell and ensure it gets into customers' hands versus just sort of expecting it to be purchased right off on their own and customers can, like, know how to do it right away on their own.