Welcome everyone to The Information TI. My name is Akash Pasicha. It is Wednesday, January 14th. Today on the show, I am talking to Airbnb CEO Brian Chesky about the company's new chief technology officer that he announced today. I'm also talking to him about the current state of Airbnb's business, where it's going, and how he is integrating AI. We'll then get into some exclusive reporting from The Information around Microsoft's growing relationship with Anthropic. It is Wednesday, which means it is time to discuss our weekly finance newsletter. The topic for today, how one new tech bank with a name plucked from J.R.R. Tolkien's Middle-earth is trying to shake up finance. Excited for that conversation. And we will end with a segment about how a new proposed billionaire tax in the Golden State may or may not impact Californians' desire to leave the state. It is a fun show. So, let's get right on into things. Airbnb has a new CTO. The company said today that Ahmad Al-Dahle, who previously was a top AI executive at Meta, is taking the job. I sat down with Airbnb co-founder and CEO Brian Chesky to talk about how he plans to further integrate AI into the Airbnb platform and how he's thinking about the future of his business as the stock has basically been flat over the past few years. I also got some details about the current state of Airbnb's experiences, services, and hotels business lines, which he is betting on to re-accelerate topline growth at the company. Here is that conversation. Brian Chesky, welcome to TITV. It's great to have you here.
Well, thank you for having me.
So, look, you wrote a memo to employees introducing Ahmad Al-Dahle as your new CTO. And before we get going, the first thing that jumped out to me was that you wrote he grew up in Winnipeg, Canada, one of the coldest places in the world. And I just wanted to share I was also born in Winnipeg. So I can confirm it is cold.
Oh, it's true. I mean it is cold, but the coldest place I've ever been was Calgary. I got off the airplane in Calgary, my friend picked me up and gave me a jacket to put over my jacket. And I've heard Winnipeg's colder than Calgary. They have to plug in their cars at night. That's how cold it is. But it's a great hockey town and I know you're a hockey fan.
I love hockey and I grew up a place not quite as cold as Winnipeg in upstate New York, but it was kind of near Buffalo. So kind of familiar with snow and cold and hockey.
There you go. Great. Okay. Well, let's get into it. So, I want to understand how Ahmad fits into your vision for AI at Airbnb because you've said publicly that you want Airbnb to be the real-world connections app in an AI world. You've also talked about how models will become commoditized in the long run, will become something like electricity. And you're already using AI in many of your customer service tools on your platform. And so the question I want to start off with is how is Ahmad going to help you extend AI throughout the rest of the company?
I mean three things. Number one, the most fundamental thing is AI is this great workflow technology to make everyone more efficient doing their jobs and I think obviously software engineering is being revolutionized by AI but basically every single function of the company from legal to finance to design to marketing could be benefited from AI tools. So I want us to be one of those companies that started before generative AI where we're not an old school company. We're a new school company and we're using the latest and greatest tools. That's number one. That's just fundamental. Number two, a lot of people talk about AI and one of the common use cases is travel, a travel agent, right? We decided to start with the hardest part of AI with Airbnb, which is customer service. And so now 15% of our, you know, we were able to reduce our contacts in English by 15% by having the AI do customer service. That's really, really hard to do. It's one thing to say, where should I travel? It's another thing to say, I'm locked out. Can you help me? And so we want to take that customer service agent and bring it up-funnel. And ultimately, I don't think we're going to live in a world where you just use one agent, one chatbot to do everything. I don't think you're going to want a thousand agents that just like you don't want, you can't have a thousand friends. You don't have just one person in your life. And so I do think some specialization will benefit and not too much specialization. What we're going to specialize in is traveling and living. If you want to travel or live somewhere, you're going to go to Airbnb and we're going to be able to use the same models as the frontier model companies because we can license those models, but we're going to specialize. We're going to tune them and train them for our use case. We're going to design these really great interfaces that marry Airbnb's design with frontier technology to be able to help you travel and live anywhere. The other point I'll just say is a very small percent of Airbnb is the app that you see and you book. A large part of Airbnb is managing 5 million hosts. Having 5 million hosts in our community, handling over approximately a hundred billion dollars going through our app in nearly every country in the world trying to keep people safe every night, handling tens of millions.
What are some of the things that you could see AI automating on that vector of working with the supply?
Well, creating a listing is probably the most obvious one, right? You can in the future basically just add your address or even just say my current location is I'm home. We can then scrape an MLS. We could basically create your listing for you. If there aren't photos, you could take photos. AI could use computer visioning to then take the photos, write the description, it could verify the amenities exist at that property. So creating a listing would be significantly easier. That's just one example. There's like dozens of dozens of examples like that.
And so I wonder how you think about ROI in this AI area. Is it quantifying time savings either for your own workforce or for the people who are putting their homes up to be listed on the platform? Is that something you quantify? Do you think about increasing revenue if you're getting more bookings, being able to use a chat interface? I mean, how do you quantify ROI for AI?
Probably three things. First is just being able to move faster and deliver more features. You know, if for example, if every engineer is using AI tooling and you create like double the efficiency, then you can essentially ship features at twice the rate and that will of course lead to a lot more growth. So that's on the tooling. On customer service, you can reduce customer service contacts. That of course if an AI agent can essentially handle a ticket that saves a lot of money in the workforce, but also if it can respond faster, you can recover the customer and pass the savings on to the customer. The third thing is just essentially making it a better store for you. So making it easier to find the right booking increases conversion of supply. So as conversion increases of traffic, revenue goes up. So, shipping features faster, reducing customer service, increasing revenue by increasing conversion. Those would be kind of like three really big ROI metrics. Again, there's probably dozens. Those would be probably the three most important.
So, it leads me into sort of a broader discussion I wanted to have with you about the business at large. I mean, if you look at the growth trajectory for the business, you guys have no problem with profitability. It's very strong there. Growth is decelerating and the stock has very much been flat. And the solution to this you've said is you know experiences, services, and hotels in the long run could be multi-billion dollar businesses. Where are those businesses right now in terms of revenue?
Yeah. One thing I want to say is growth had been decelerating and now it's re-accelerating in our most recent Q3. We showed acceleration from Q2. We haven't reported our Q4 results yet. But I'm very optimistic.
Q4 was guided to seven to 10% around the same as the current quarter. Right.
Yeah. And I can't obviously comment on the results. Our earnings call is in I think four weeks, but we are optimistic that over the coming years we'll be able to accelerate growth in Airbnb. And how are we going to do that? A few things. Number one, creating new businesses. So you ask about experiences. Experiences we are really optimistic could be a multi-billion dollar business. Services, there is no Amazon for services. Whether you want a chef or you want a personal trainer or masseuse, we think that could be very popular for travelers and eventually locals. Hotels. People don't realize how big the hotel industry is. Google and Meta are essentially advertising companies. They're in the advertising industry. The advertising industry is smaller, I believe, than the hotel industry. So, we do not need a large market share of hotels to have a huge business. Hotels will absolutely be a multi-billion dollar business. One example, there were over 400 million searches for New York City on Airbnb last year. 400 million. Airbnb is essentially banned in New York. Adding hotels, we can make quite a bit of revenue in New York. So, there's a lot of these businesses and what we want to do is release more and more businesses every year. So, we think there's a lot of growth of our core of those new businesses. But in our core business, just one more comment, 70% of our business in our core business is in only five countries, US, Canada, Australia, UK, and France. There's a couple hundred other countries in the world that everyone can still grow in our core business. So, there's still a lot of growth and for that reason I think we can re-accelerate.
So, I'm just trying to get a picture of where those businesses are right now. I mean, we look at experiences. I mean, how much revenue is experiences generating for the company right now?
They're all very, very early. We don't report on a segment basis, so I can't share what those are, but I would say that experiences is a little ahead of services. Services a little ahead of hotels. And I think each of them will take a few years. Like we basically put a horizon of three to five years for these to be multi-billion dollar businesses.
But experiences is the one most likely to hit a billion dollars in your mind first in terms of revenue and contribution to the business.
That's interesting. Maybe not. It's the furthest along because it already existed and so there was already a higher revenue base. That being said, the TAM, the total addressable market, hotels is the largest obvious one. And then services is super binary. Services could be a medium business or a massive business. And the real question on services is whether or not locals book services in their own cities. If they don't, it's a good business. If they do, it's a hugely successful business.
And I wonder what you say to, I mean, we've had other travel entrepreneurs on the show and we've talked a little bit about experiences and services and the challenge that they have talked about is that it's hard to get people. Services is a bit separate from booking hotels or booking a trip. I mean it's almost a category on its own. And you've talked a little bit about how you segment customers in Paris and LA, people visiting, people living there, but I mean what's the biggest challenge right now getting people to offer services? I mean is it a supply issue? Is it a demand issue? What's the bottleneck right now?
The bottleneck is absolutely not supply. We're inundated with supply for people requesting services and experiences and I presume that that's only going to increase. I mean I can't predict what's going to happen with AI but with more and more jobs being automated I think more and more people are going to go to the service industry and I think the vast majority of hosting opportunities aren't going to be automated by AI. So I think that increasingly people are going to be looking to Airbnb to become hosts. I think the bottleneck is awareness that we have services, awareness that we have experiences and that you can trust that they're high quality. So, it's really just a matter of people being exposed, using them and then word of mouth increasing. Experiences are going really, really well. In Paris for example, we decided to focus on Paris for experiences that is growing very, very quickly. So, if you type in Paris and experience, you'll see that's what I think a future marketplace can look like. One of the things that's surprising is the Parisians are booking experiences in Paris. That I wasn't expecting. They're working really, really well. On services, we really focused on Los Angeles. You type LA, you'll see a really robust store for services. It's working in LA, so we're now bringing it to cities all over the world. We just launched grocery stocking and airport pickup. Those are working out really, really well.
So, you get an Airbnb, we have an integration with Instacart where you can have groceries stocked near your fridge. That's very convenient. And then in Europe, Uber is not always allowed in an airport. It's really hard. So we work with some local companies in Europe to be able to do welcome pickups where we do this really simple integration. The driver has flight information they can pick up the airport and take it to Airbnb. As you think about getting the revenue growth rate up to that 20% mark that you've said that you hope to get the business to. I mean advertising is one easy way if you sell ads that's one way you could sort of juice the revenue growth. Why not just start there?
I think advertising is another multi-billion dollar business. It's at least a billion dollars. And so the question is why not start there? Well, the reason not to start there is number one, it's not the most perishable opportunity. I love this Jeff Bezos framework that you can do anything but just start with the things that are most perishable. And I think the most perishable things are going into new service categories. I think there's a real opportunity to do that. The second thing is there's the old model of ads within search results pre-I like the Google model of adding ads but we are now going to be living in a world of AI search and so we want to design ads for an AI native search world and no one's really figured out how to do that and we might be one of those people to try to figure that out. So we don't want to create ads for the old world we want to create an ad unit for the new world. So we're waiting to get AI search right before we think about doing ads.
How do you think that plays out? You come from a design background. I have so many questions for you about what designing a good customer experience looks like with advertising, which is the way that a lot of these platforms are going to have to generate and monetize more revenue. You know, what does a good customer experience look like from a design perspective with ads? Talk a little bit about that.
Let me just back up and say something about the whole ecosystem of consumer then I'll talk about ads. I don't think generally speaking people have figured out how to make money on AI consumer. I think AI enterprise is going really, really well. On the board of Y Combinator the vast majority of startups are enterprise. Very few companies are applying for consumer. And I think what we've seen is it makes sense models, the queries are fairly expensive but you can pay for them with enterprise. I think with consumer there have to be one of three models to make money consumer either advertising, paid subscriptions or essentially e-commerce transactions. I think Airbnb is one really good model because we have a very high ticket dollar. So you can make a lot of money on every transaction. So we can afford to invest quite a lot in AI from a transaction increase standpoint. If it increases conversion that's great. I think advertising though is probably inevitable for a lot of chat bots. And then the question is two questions. One, does the company optimize for engagement or not? Two, do people distrust the results if ads are delivered? So, in Google search results, I never distrusted Google search results because the ads were on top. I said, 'Okay, well, those are ads. They're clearly ads. The things below them are what Google suggests.' So, I never distrusted it. With AI, it's really important that the questions people no longer distrust the veracity of information. And I think that's a design challenge. And how I would solve that, I haven't totally, we're in the middle of figuring that out. But I think that's the biggest problem. The second which is not our problem. We don't need to optimize for engagement. But I think it's really, really important that a company's north star is user experience not engagement because when engagement is the north star then I think suddenly the company becomes the customer. So I think it's really important for a chatbot that people trust the chatbot and so therefore you just have to make sure that it's not compromised at all.
Would you ever integrate more with OpenAI?
Absolutely. I say that knowing that you're friends and knowing that you've said that the models aren't there or maybe the interface is not there yet, but they're not yet. They're not there yet, but I think they'll get there. And when they get there, we'd love to integrate. Listen, like other travel companies decided to integrate with ChatGPT. And I told Sam, we don't need to be the first. We just need to be the best. I don't need a race to it. The SDK like three things. Number one, it was a little bit hard to discover Airbnb in the chatbot. Now it's becoming more easy. You didn't have to install it. Number two, there was a lot of latency. So it took many, many seconds to load. Three, 90% of people when they book and they send a message, they have to verify their identity. So, there were just some interface things we needed to be able to do the integration, but we are absolutely interested in showing up in Chat and Gemini and others. I don't think they're an existential threat at all. I think two things can coexist. I think what can coexist is we can win from specialization, but also some people are going to start their travel inspiration at chatbots and they can continue it on Airbnb. So, I think both can coexist and as long as your experience is great, we will do it. But we're not one of those companies that wants to be the first to integrate. We want to be the best integration.
Right. I want to ask you quickly before you go, Brian. There are a lot of budding entrepreneurs in our audience that look up to you. If you weren't building Airbnb right now, what business would you be building?
I would do something with consumer AI. And I don't know. I mean, I could give you a bunch of categories that I think are really interesting, but right now, again, I'll just say this. A lot of people are afraid to create consumer AI companies. When I joined Y Combinator, almost every company was a consumer company. The app store had just launched and everyone was making apps for regular people. Now almost every single person is creating an enterprise AI company. It makes sense. But the vast majority of the economy is not going to be enterprise. It's going to be consumer. The biggest companies in the world are going to be the companies that touch the most lives. And the companies that touch the most lives daily, daily living are going to be consumers. So my advice to entrepreneurs is don't be afraid to start a consumer company. It's possibly harder. It's a little more of a hits business. It's more marketing than sales. A little harder to control. But don't be afraid of it because AI has not changed daily life yet. And it won't change daily life until proliferation of consumer apps come. Now, what would they be? I think health is interesting. I think education's interesting. I think e-commerce is really interesting. E-commerce is a great business model, right? We talked about there's not a lot of business models for consumer AI. E-commerce is a business model. So these would be really obvious spaces.
Right. Well, Brian, I want to thank you for coming on the show. We really appreciate it. It was a great discussion and hope to have you back again soon.
I hope to. I hope this is the first of many interviews.
Great. Well, that is Brian Chesky, the CEO of Airbnb here on TITV.
Okay. Microsoft is growing its relationship with Anthropic. New exclusive reporting from The Information today reveals just how much Microsoft is spending on Anthropic. I want to bring on Aaron Holmes, one of the reporters on that story to talk about what we've learned. Aaron, welcome back to the show. It's great to have you here.
So, what did we learn about Microsoft's work with Anthropic?
Yeah. So, I mean, over the past year, we've seen Microsoft and Anthropic get a lot closer, even though OpenAI remains Microsoft's most important AI partner and provider. And specifically, that's looked like, you know, Microsoft becoming one of Anthropic's biggest customers, spending, you know, on track to spend more than $500 million annually on Anthropic's models. And we've also learned that Microsoft is now incentivizing its salespeople to sell Anthropic's models on Azure just as much as it incentivizes them to sell OpenAI's models. And you know, beyond that, I think Anthropic is also starting to even get more involved in helping Microsoft build products that use its models. So, across the board, these companies are getting a lot closer.
Now, you reported that Anthropic is on pace to spend $500 million annually with Microsoft and that would make it one of Anthropic's biggest customers. Did I have that right?
Yeah, Anthropic is spending the money on Microsoft. No, Microsoft is spending the money on Anthropic. Yes, Microsoft is spending $500 million annually on Anthropic's models. At the same time, Anthropic did commit earlier or late last year to spend billions of dollars on Azure cloud services. So, the money is going both ways. But yeah, I mean the money that Microsoft is spending on Anthropic is pretty significant for the startup. Anthropic projected that it was, you know, going to make in the realm of 7 to 9 billion in revenue last year. So this is pretty significant that it's getting this much from Microsoft.
Now, one of the questions I wanted to understand here is what types of models is this? Because Microsoft obviously has an expansive suite of products. Are these coding models? Are these large language models? What are they buying from Anthropic?
Yeah. So, you know, as of the $500 million annualized figure, most of that was coming from coding models, which Microsoft was using to power AI agent coding features in its GitHub Copilot software. And since then, the amount that Microsoft is spending has actually probably grown a lot because they've started to also add Anthropic models to other AI products like Office 365 Copilot and Security Copilot. So at this point it's really, you know, across the board that Microsoft is using different Anthropic models for different types of AI products.
Is Anthropic reciprocating? Are they spending with Microsoft?
They are. So they, you know, Microsoft...
Also invested 5 billion in Anthropic in September, and around that time Anthropic said they would commit to spend several billion dollars on renting cloud servers from Microsoft. They also said they might contract several more gigawatts of cloud server space in the coming years. Although it's hard to know exactly what that means, but yeah, basically both of the companies have now become large customers of one another.
And who do you think has the leverage in this relationship?
Yeah, I mean, with all of the ways that money is changing hands back and forth, it's a little hard to say who's benefiting more. But I think my take is probably that Anthropic is getting the most out of this right now because they really want to grow their enterprise business and getting access to customers of Microsoft's Azure cloud service is a way to massively expand their potential universe of customers. At the same time, Microsoft does already have access to reuse OpenAI's models for free. So it maybe doesn't need Anthropic quite as much as Anthropic needs Microsoft. But regardless, it is obvious that both companies are benefiting a lot from this partnership.
And I want to end off with a chart that I would encourage people to look at in the story. You basically mapped all the ways in which the AI model companies and the cloud providers and companies like Nvidia and Meta, the ways in which they are doing deals with each other. And it's a great matrix. My takeaway from that is everyone's working with everybody. My question for you is, is that inevitable? Do you see it as a risk? What did you make of that?
Yeah, I mean, I think in some ways this was inevitable, especially when you look at cloud providers like Microsoft, Google, and AWS who want to have customers on their platforms using all models and they don't want customers to have to use a different cloud provider to seek out different models. And we've also seen Microsoft CEO Satya Nadella talking about how he thinks models will become commoditized, essentially that in the future all different model providers will essentially be able to offer the same type of intelligence and it won't matter as much which model is underlying a specific tool. That hasn't happened yet, but I think that is sort of the strategy that we're seeing from the cloud providers in a macro way.
Great. Well, Aaron, I want to thank you for coming on. That is Aaron Holmes, our Microsoft reporter here at The Information. Okay. The AI boom has given rise to a new class of companies called inference providers. As AI applications rely on models and on compute, inference providers are becoming more and more important to the story. One big fish in that arena is Base 10, a company that was last valued at 2.15 billion last year. I want to bring on co-founder and CEO of Base 10, Tuhin Srivastava, to talk about the future of his business. Tuhin, welcome to the show. It's great to have you here.
Hi. My gosh, how are you? Thanks for having me. I'm doing great. Thanks for joining us. I'm excited for the conversation here.
So look, I just want to make this clear. So you are not a cloud provider and you're not building the models. You connect the two. Is that right?
That's right. We sit on top of it. So the way that we think about the world is that there's going to be a lot of models in the world. They're all going to need to run somewhere, and they're going to need some specialized software to be able to run in production. So, you know, we're kind of that middleware that sits between all these different clouds that we work with. Today we sit on around 10 different clouds and the folks building models. And today we power some of the fastest growing companies in the world. Think about like Bridge, OpenEvidence, Clay, Cursor. They would all run production workloads with Base 10. And yeah, that's exactly where we stand.
And so help me understand this. I mean, if I am building an AI application and I want access to models, one way to do it is I could just go straight to the companies like OpenAI, Anthropic, and I could say, hey, I'm just going to use your API and get access to the models that way. But the other way to do it is you go through you guys. The third way is I could just go to the cloud providers and they give me access to all these models as well, right?
