Exclusive Interview with Uber’s Incoming CFO, OpenAI-Amazon Weigh Deal, US-China Chip Trade War

0:13Welcome, everyone, to the information's TI-TV. My name is Akash Pasricha. It is Wednesday, February 4th. We have a big day of earnings to unpack. First up, I have an exclusive interview with Uber's incoming CFO, Balaji Krishnamurti. The company announced he will be taking the top finance job later this month in its fourth quarter results. I'll talk to him about how he's thinking about the current and future state of the business. We'll also dig into AMD's results and the chip sector more broadly with government regulation and M &A front and center. I'll be joined by Chris Miller, author of Chip War.

0:50We'll then break some exclusive reporting from the information on a possible commercial deal in the works between OpenAI and Amazon. And finally, software stocks are tumbling left, right, and center. We will talk about what investors are reacting to with our finance editor. It's going to be a fun show, so let's get right on into things. Uber reported fourth quarter results. Revenue grew 20%, the same as last quarter, and faster than each of the first two quarters of the year. But bookings growth is expected to decelerate in the current quarter. Of course, the big news of the day was that the company is getting a new CFO.

1:27Balaji Krishnamurthy is taking over that job on February 16th. But today, we have an exclusive interview with him to talk about how he's thinking about the current and future state of the business. Balaji, welcome to the show. It's great to have you here. Thank you, Akash. Congrats on the new job. You must be excited. Thank you. It's been a busy day, but I'm happy we can do this. Great. Well, I'm excited to get into all the different things that are going on with Uber's business. I want to start with the outlook that you guys gave for Bookings growth. Now, bookings grew 22 % in Q4. You're expecting it to grow anywhere between 17 % and 21 % in the current quarter.

2:08And that does incorporate a little bit of a tailwind from the foreign exchange rates. And so my read on it is that the bookings are expected to decelerate in the current quarter. Why is that? So let me start with a clarification. The 17 % to 21 % growth rate that you refer to, it is constant currency. So the FX tailwind would be incremental to that. We said we expect at least four points of FX tailwind. So the reported growth rate would be 21 to 25%. That would be the way to think about that. But even from that, just stepping back, as you think about the outlook philosophy we've had, in Q3 and Q4, we guided to a similar 17 to 21 % range.

2:55and for Q3, we delivered 21 % growth. For Q4, we delivered 22 % growth and now our outlook is consistent. So I wouldn't read this as a big deceleration. If anything, in Q4, we got surprised a little bit positively because the holiday season was so strong. It exceeded our expectations, but otherwise the underlying business momentum is exactly unchanged from the second half and mobility and delivery both are firing on all cylinders and we're very excited about the year. I want to talk about mobility and delivery. I mean, these are two businesses that delivery is growing very quickly. Dara, the CEO of the company, has talked a little bit about the long-term ambitions for these two segments and the relative sizing of how they could play out.

3:47How do you see these two businesses three years from now in terms of relative sizing? Oh, that's a great question. That's a crystal ball gazing exercise that we have done many times and come out wrong. So I will not endeavor to go there. I'll say that both of these businesses are growing at scale, 100 billion gross bookings run rate plus, and they're both growing 20 plus percent, right? It's very rare to be sitting with those kind of global platforms which can deliver that growth. So we'll make the appropriate investments in both of those businesses as long as the returns on those investments meet our hurdle rates.

4:31And then we'll see which one comes out on top. It's a friendly competition. Friendly competition. Now, the other side of the business that, of course, the company is investing a lot into is your autonomous vehicles. ambitions. And you talked a lot about that opportunity in the shareholder letter this quarter. And you talked about some of the misconceptions that you think exist around Uber and what the autonomous vehicles opportunity could mean for the company. What was the overall message that you were trying to send to investors with that? Yeah. So that's great. So I would say, first of all, Uber is very, very bullish about autonomy.

5:14We think this is a technological disruption that is absolutely playing out. It's going to be a huge portion of mobility in urban cities in the coming years. And as we go through the next decade, it's going to become a big contributor to expansion of our time as well. And, you know, I think the simple reality that we think about here is that there's an opportunity to get to personal car displacement over the long run. And Uber Stam then becomes a trillion dollar plus if we can have autonomy unlock in the right way. It's going to be incremental. Now, what we see, though, is when investors think about autonomy and Uber's role to play here, there are a lot of misconceptions.

6:00and we've heard enough of them that we thought we would take a moment to just correct the record on a lot of these issues and put the nuance that's missing in the conversations that we hear. You know, the first and most important one that we hear about is autonomy is going to be cannibalistic to ride hailing. That could not be further from the truth. Where we have autonomous deployments happening in Austin and Atlanta, our business is growing at the fastest rate in the US of any markets. Even in markets like SF, where we don't have autonomous on our network yet, we will by the end of this year with Nuro and Lucid.

6:38Even in a market like that, SF is growing for Uber faster than the rest of the large markets in the country, and it accelerated last year. And the underlying premise behind that is supply growth is always incremental for ride-hailing demand. Another one that is commonly heard by us is Uber's profits are very concentrated in top 20 markets. And all you need to do is to launch in those top 20 markets to make a great business. And that is, again, very, very wrong. because as we think about our business, we operate in 8 ,000 plus markets in the US. Consumers travel around the country, right? 40 % of people's trips happen outside of their home cities and they expect this utility service to be available everywhere.

7:28So I think what that has done and given our focus on driving that long tail of markets in our portfolio, 70 % of our business now in the US comes from outside of those top 20 markets. 75 % of our profits come from outside of those stock trading markets. And those markets are going to be unadressable by AVs for a long time to come. I would encourage everyone to read everything else we said. I wouldn't go into everything here, though. So let me go back to, you know, you talked about San Francisco as sort of a case study. And you talked about, you know, people thinking about the competition that comes from platforms like Waymo.

8:06And I know that Dara has talked a lot about this idea that, look, you take something like a Starbucks. Starbucks has a direct-to-consumer relationship. They also use Uber Eats to deliver food and stuff like that. And so there's sort of this hybrid system. But, you know, I do want to get your read a little bit on Tesla as sort of the wild card here. I mean, Waymo has set themselves up. Do you think that Tesla will really come into the fold here as a, do you think we'll pull it off is my question? Sure. It's a good question. So I would start with the premise of camera only AV deployments, right?

8:49And today the jury is still out on that question. And Tesla is making a big bet that camera only, vision only can work on autonomy. So we'll see whether it works or not. we would never bet against Elon. And we need to watch that. But I mean, let's say they put the sensors on. I mean, he has the model of the whole, you know, you can rent your Tesla out, right, as part of the fleet model. So, I mean, you know, as a competitive threat, I mean, what do you think of it? Look, I think if they change their approach to this and moved in that direction, I think there is some time that they will take to make that adjustment.

9:32I would just come back to where they are today, which is based on their publicly reported data, their incident rate is effectively where Wamos incidence rates were in Q1 of 2023. So on that basis, it looks like they're about three years behind. And we are not saying that Tesla is not going to get there. It's entirely likely and possible that they would be there. But in that time, there are going to be other players who will also deliver autonomous, commercially deployable autonomous solutions. And we have the broadest set of partnerships, right? We have 20 partners. We are looking forward to bringing them to the 15 markets we've talked about this year.

10:12And that number is only going to go up in the coming years. And on our platform, the utilization and revenue generation capacity is going to be higher than anyone else's. So we'll compete on that. Now, one of the interesting parts about Uber strategy that you've talked about is the idea that in the near term, while the company has never been in the business of owning fleets and operating it and really sticking to this asset light model, you've talked about how in the near term with partnerships like with Lucid, you may own some amount of vehicles in the short term as you look to test out the economics of having autonomous vehicles.

10:52And I'm wondering, as the incoming CFO, the first question is, how expensive is that? And what is sort of the upper bound of what you're willing to invest in owning those fleets? And how do you think about a return on investment on that over time? Yeah. As you think about the context of our scale and what we are looking to build here, we have nearly 10 million human drivers around the world. When you think about the autonomous deployments today, the largest deployment, whether in the U.S. or in China, is in the 2 ,000 to 3 ,000 vehicle range. What we will need for autonomy to really work at scale is tens of thousands or hundreds of thousands of cars deployed on Uber's network over the coming years.

11:37And to get there, the first step is for us to get OEMs excited about producing these vehicles, right? And they need to know that if they're producing these vehicles, they are going to be consumed by somebody. And the first step for that is we are working with our OEM partners on offtake commitments. So we may go to a Lucid and say, build us cars. We will guarantee that at least 20 ,000 of the cars you're building over the coming years will be moving off of your assembly lines. And we'll ensure some of our partners buy those cars. But in the early years, we may buy some of those cars ourselves.

12:14Just because we want to learn what this model looks like, we want to demonstrate the economics of these models, and then our partners can get third-party financing to deploy it at a much more massive scale. And then the last point on this topic, as you think through that, is, again, we are so early in this deployment ramp that we want to be flexible and we want to think through what evolution these models take. and it behooves us to remain nimble. And we will do so with discipline. We are not looking to expose a lot of our balance sheet, but to some extent we can do that. And at our 15 billion trip scale, this is going to be pretty small.

12:56Do you have a budget for buying these fleets? I'm not going to set a number there. I would still say that if you think through the capital allocation priorities we set out today, I would say we have the ability to make investments in our core business, to make investments in AVs, to potentially look at bolt-on acquisitions, and still buy back a ton of our own stock, which remains very, very cheap. And today, the highest return on investment for us is to buy our stock, right? So we will be able to do all of those things in parallel. I want to come back to the core operations of the business today.

13:38I mean, recruiting drivers is one of the biggest success factors to having a network like Uber does. Even today in 2026, what is the biggest challenge for Uber in recruiting drivers and keeping that growth going? Yeah, I think it is the fundamental motion that we need to have. We are a supply led business. as we bring on drivers, we have to make sure that we are doing so in a way where driver onboarding is streamlined, drivers can come onto our platform with as little friction as possible. So there's a lot of tech investment we have had to make in improving that onboarding funnel. The second thing we have to ensure is that earnings are predictable and they remain high relative to their alternatives.

14:29And we have done a very good job of growing our top line, our profitability while delivering really healthy outcomes to our driver partners. And then finally, I think there's incremental pain points that drivers have always surfaced up to us, such as they want to see trip transferency, which we addressed a few years ago on where they're going, how much they would earn. They want to be able to, you know, there's a long list of things that our product team has to go through to ensure that we are meeting drivers' expectations because this is a competitive labor market. And if we don't meet those expectations, they can go elsewhere.

15:08Great. Well, Balaji, I want to thank you for coming on the show and congrats with the new role again. That is Balaji Christian Murthy, the incoming CFO for Uber here on TI TV. AMD reported its quarterly results. Revenue grew 34 % in the fourth quarter. The company is expecting top line to grow about 32 % at the midpoint of its range this quarter. Growth, for the most part, has hovered around this ballpark for the past year. Meanwhile, the chip sector has never been more dynamic, with government regulations and M &A being front and center. I want to bring on someone who has gone deep on this sector.

15:44Chris Miller is the author of Chip War. Chris, welcome back to the show. It's great to have you here. Thanks for having me. So there's so much that I want to cover with you. We have the AMD news this morning. We, of course, have the H200s, the trade rules, the NVIDIA Grok deal. So we'll see how much of it we can get to. But I want to start with AMD here. I mean, shares are down today because of the earnings. But I am curious from you on what you think of their place in the broader chip story right now and what you make of their competitive positioning. I think the challenge that AMD faces is that they're not only competing with NVIDIA, they're also competing with all of the ASIC projects that Meta and Microsoft and Google have in-house.

16:28And what we've seen, I think, over the last couple of months is these ASICs are, at least some of them, looking pretty good. Google's TPUs in particular are getting a lot of traction, impressive on their specs. And so that is creating a much more crowded market that AMD has to compete in. Now, the other thing that came up on the call was the trade dynamics with China right now. And AMD is, like NVIDIA, waiting for all sorts of green lights from multiple different parties on getting its chips to customers in China. What do you make of the current state of that dynamic right now? We had some reporting in the last 24 hours from the FT that we're sort of back to a bit of a stalemate here between U.S.

17:12and China. Even though the U.S. has given a green light, it's more of strategy as opposed to execution right now. And then, of course, you have China that's waiting to see whether the customers will get their green light. What do you make of this? Yeah, I think we know that the Chinese customers, like Dan, Alibaba, and others, they want the chips, no doubt about that. They don't have access to domestic alternatives at the scale that they need. The Chinese government, though, is less certain. The Chinese government wants to make sure its companies have the computing resources that they need, but China also wants to build up its own domestic ecosystem to support Huawei and the other companies that are designing AI accelerators.

17:53And I think the Chinese government has been going back and forth on how many GPUs it's going to allow in because it wants to make sure that there's plenty of space in the market for Huawei. In the long term, the strategy is to supplant foreign providers of GPUs with Huawei's own domestic production. And then the other uncertainty is indeed in Washington, where there's a debate ongoing as to whether and what volume of H200s should be sold to which customers. And so we've seen a formal green light from the White House, but that just kicked off a security review, which is currently working its way through the U.S.

18:28government. And one of the questions is, what are the customers that you can sell chips to without triggering any concerns about diversion to the Chinese government or Chinese military? And what are the customers that you need to be worried about? And that debate, as I understand, is still ongoing. One of the questions that we try to get to the heart of on this show is you have the issue that will play out over the next few months. and this decision will go one way or another, depending on what week we're talking about. But what I'm really interested in is the two, three-year roadmap for how you think this issue plays out.

19:00Where do you think it goes on that time horizon? Well, I think what we know with pretty high confidence over the next couple of years is that China's not going to have the domestic manufacturing capacity to produce the chips that it needs. Right now, best estimates are that the US plus Taiwan are going to produce something like 30 times the number of AI accelerators quality adjusted as China is. That's a huge gap, which means that for now, China's reliant on whatever imports it can either smuggle in or get its hands on legally. So I think Chinese tech companies are going to keep trying to buy whatever chips they're allowed to buy from the outside.

19:35And then the question is on the Washington side, what are we going to be willing to sell or not sell? And I think here you've seen the president signal a willingness to sell some, but you've also seen a strong reaction in Congress and elsewhere in Washington that is really cautious about giving China any additional computing resources on the grounds that the more chips China has, the stronger its AI models and the stronger its cloud computing firms are likely to be. So basically, you're saying that even though this is a story that changes week by week, three years from now, you think we'll actually be in the same position where we still have U.S.

20:11chips that are better and China trying to catch up, and we still have this dynamic. Yeah, I mean, the key challenge that China faces is that TSMC is in Taiwan, and China doesn't have anything comparable. And so they can't currently make domestically the tools that you need to manufacture advanced chips, especially the lithography tools that ASML is the sole source of. They can't buy those tools because we don't let them. And so China is trying to make advanced ships using second-rate equipment, and it's very, very hard to scale up. They've been trying to scale up for a couple of years with some success, but TSMC is scaling up faster, building more fabs, racing forwards on its process technology.

20:49And that's how you get this 30 to 1 gap, which is really pretty extraordinary. And it's not something China is going to solve in a couple of years. And one of the things I wanted to ask you, because you're, you know, you, you've studied the history of, of the chip sector so closely is, are there any historical parallels, maybe even outside of chips? You know, I'm thinking about electric vehicles, for example, are there any historical parallels where China has not had the capacity, the technology initially, you know, maybe they've come from behind in some ways? I mean, I'm just trying to sort of leave opened the possibility that maybe there could be a breakthrough.

21:25I don't know. Yeah. If you look at the sectors where China's done really well and kind of leapfrogged ahead, like EVs being a good example, what you find is that those are spheres where China's technology, you could debate whether a BYD is better or worse than a Tesla, but there's not a clear metric that says one is twice as good. China's certainly very competitive on price. And so if you match quality and are lower on price, you win the market. But in chips, it's different because you got this enormous quality improvement that drives an enormous quantity improvement in computing power. And so no one in the US is buying four-year-old chips to say nothing of eight-year-old chips, because they're just so much worse than the cutting edge that a company like NVIDIA can offer.

22:06And the need to stay on the cutting edge is just so much more acute in the chip industry, which means that you can't compete simply by getting 80 % the quality and being half the price. You've got to be 100 % the quality to be in the game. Now, you talked about TSMC, and I want to ask you about their work in the United States and their ambitions to expand here. That has been a long journey for them, and they've gone through a number of challenges. What do you make of the challenges they've had and what it tells us about where they'll be two, three years from now in the U.S.? Yeah. You know, I think TSMC is certainly committed to ramping up in the U.S.

22:45They've said publicly that they're going to have something like 25 % or 30 % of their advanced production in the US by the mid-2030s. That's a real big change from a couple of years ago when that number was zero. But it's also true that it's been a complicated journey. This is the first time TSMC has done advanced manufacturing outside of Taiwan. It's had to build up its local workforce, build up its knowledge, and it's confronted both a challenging environment, very different from Taiwan, that's driven higher costs. And so I think we should be optimistic that TSMC is going to build a lot more volume in Arizona, but we should be sober about the cost differentials that I think are going to persist over the long run.

23:24What gives you optimism there? Look, they've executed. They've already got one plant up and running. They've publicly said the yields, the share of chips they manufacture that work as intended are just as good, if not actually slightly better than in Taiwan. I mean, TSMC, I think is second to none in terms of their ability to execute. And when they devote themselves to a project like this, I have no doubt they're going to fulfill it. I think the challenge for the U.S. is just that it's not going to be at the same price point as Taiwan. And so we've got to balance our desire for resilience, which is important, with the cost dynamics.

23:57Now, since the last time we had you on, the big chip deal that we saw was NVIDIA and Grok. And I wonder if you think that that deal was offensive or defensive on NVIDIA's part. Maybe a bit of both. Certainly, the defensive dynamics are clear. But I think it was notable that just a couple weeks later, it was followed by an announcement of a deal between OpenAI and Cerebris, which is another firm that has inference-focused architectures, which speaks to the offensive part, that these companies just need more capabilities that are more efficient at inference. And so for NVIDIA, Grok could deliver some of that as well.

24:38And I think, you know, these two deals really open up a big new space for inference-specific chips, which until recently haven't gotten a whole lot of market traction. Well, and the third deal that, to be clear, I mean, it hasn't, it's been reported to be in the works and there's rumors about it. We haven't seen any news of it formally. But the other deal that we've been watching is Intel perhaps looking at a company like Samba Nova. And we have documented the fast growth of these chip startups here at the information for quite a few years now. I wonder if you think we are at a bit of an inflection point where you have these chip startups that are really coming to terms with reality with how big the competition is.

25:23And I mean, is this the start of, I don't want to call it start of the end, but we're seeing acquisitions. We've seen this play out time and time again in different industries. Yeah. I think you see these acquisitions, these kind of interestingly structured deals, I guess, not fully acquisitions. Right. Yeah. Whatever you want to call them. Whatever you want to call them. Yeah. We also see new startups in this space raising really substantial sums of money. And so I would say the venture space seems willing to keep funding new inference-specific architectures. And so that either speaks to a realization that even if you get acquired, you get acquired at pretty large dollar amounts or that the market is so vast that there's going to be space for both training-specific and inference-specific chips down the road.

26:09And so for now, it seems to me like investors are still willing to make a bet on there being a pathway for the inference-specific chips. Great. Well, Chris, I want to thank you for coming on. It was a great discussion as always. That is Chris Miller, the author of Chip War, here on TI TV. Amazon is continuing to build its relationship with OpenAI. A new exclusive story from The Information Today reveals that there could be another deal in the works between the two companies. This one, a commercial agreement. I want to bring on the two reporters who worked on that story. Anissa Gardizi is our cloud and compute reporter.

26:45Catherine Perloff covers Amazon. Welcome to you both. It's great to have you here. Thanks, Akash. Hi. Anissa, I want to start with you. Tell us a little bit about how Amazon is getting closer to OpenAI. Right. So the two companies have been getting closer over the past couple months. Amazon talking about making a very large equity investment into OpenAI, and the companies also have an existing cloud agreement. But the new thing here is that the companies are now talking about a large commercial agreement in which Amazon would use OpenAI models to power Amazon's AI products. And so Amazon would become a customer of OpenAI, just like OpenAI is a customer of Amazon.