Amazon CEO Andy Jassy & Jessica Lessin at Davos, Gemini’s Developer Boom | Jan 20, 2026

26:35For more coverage, I want to bring on Jessica to debrief her conversation and talk a bit more about what she is hearing on the ground in Davos. Jessica, welcome back to the show. It's great to have you here. Hi, Akash. Welcome to my tiny little studio. This is the unglamorous side of Davos, but great to be with you. Look, you're not wearing a big parka, so that's what I was half expecting. You know, next year we'll do it in the mountains. Great. Okay, well, so we just heard the conversation with Andy Jassy. What were the highlights for you from your end? Well, a couple things, Akash. I mean, I asked him a couple questions about OpenAI because really that partnership is a new one for Amazon.

27:18And as we've reported in the information, we think talks are ongoing. And obviously, we have the OpenAI ads news. And it was interesting to me that he said ads won't be easy. And while he welcomed more people in the space when it came to agentic commerce, I thought that was interesting. And, you know, Jassy's a nice guy. You don't get a lot of gloves off moments. But that one struck me as something where Amazon really is clearly paying attention to what these other AI chatbots want to do in shopping and advertising, but, you know, was pretty confident in his position. He also said, you know, we'll talk to these chatbots about integrating with them.

27:58But at the end of the day, people want free shipping. They want a broad selection. That's going to be uniquely Amazon. And it was kind of interesting to me. He was, of course, careful around talking about the partnerships with OpenAI, but he very much did leave the window open to saying, hey, we look forward to doing more with them in the future, it kind of raises the question, well, what do those deals actually look like? Absolutely. I mean, they have a little bit of a cloud deal. And obviously, AWS would like all the cloud deals, right? I also hadn't heard him speak before about his barbell of demand for AWS.

28:33This idea that the big model makers are really on one end and these sort of startup AI companies are on the other, but he's sort of interested in this middle coming online, which is more, you know, your typical business starting to use AI and wants to be on the same kind of infrastructure it's been using for cloud. So, you know, he's clearly optimistic about that. I thought he could have, you know, acknowledged a little more of the competition that Microsoft and Google have been providing on the cloud side. But again, you know, when I interviewed him two years ago, Akash, was the last time I sat down with Jassy for a lengthy interview.

29:13And I'm struck the difference. I think back then he was a little more tentative about talking about Amazon's advantages and AI was a little more on the defensive, was sort of checking off the things they were going to do in chips and models. And I thought, you know, he's a mild-mannered guy, so he doesn't really sell too hard, but he seemed more confident when he was talking about Tranium and competing against NVIDIA, for instance. I thought he was taking a slightly more confident tone than two years ago. And I think it's kind of interesting looking at the public company stock performance. We see the way these shares move when companies announce deals, when CEOs make gloves are off type of comments.

29:55I mean, maybe that's part of his own transformation in sort of waking up to the fact that, hey, we are in this moment where you have to talk somewhat aggressively about your products and your book. And maybe that was sort of a transformation he's going through as well. Yeah. And compared to others, I mean, our viewers watched it. I mean, he's a nice guy, a mild-mannered guy. He's not going to fight those battles in public. But behind closed doors in meetings, I'm sure with OpenAI, he's a really tough negotiator. So we'll see what kind of deals evolve both in agentic commerce, please let's rename that someone, and then the future of the cloud business.

30:34Well, so you talked about this sort of barbell analogy that he used, And it kind of reminded me a little bit about we just we saw comments from Satya Nadella coming out of the forum around, you know, for this to not be a bubble, we have to get out of just tech companies talking about AI. This has to be something that every company talks about. And it leads me to the broader discussion I want to have with you, which is your reflections on the ground at the forum and what people are talking about. What are some of the themes that you're hearing about, not just on AI, but politics, on trade, on business?

31:08Absolutely. Well, geopolitically, a lot of balls in the air. So I think there's a record number of heads of state here. President Trump is giving his address tomorrow. USA House is, you know, a big presence on the central promenade here. And so I think the business leaders and the finance leaders here are really trying to figure out what the shoe, what next shoe might drop on that front. On the AI front, it's bifurcated again. I talked to a lot of businesses and a lot of smart tech leaders who are like, it is early and slow in adoption. And I wrote about this last night in the Informations briefing.

31:48This idea that AI is going to be a panacea and come into your organization and change everything is really giving way to a more targeted, measured, let's get this to work. I just 15 minutes ago got off stage with Brett Taylor, the CEO and founder of Sierra, talking about, you know, their very specific, vertically focused customer service agent. And he was saying, you know, I really hope we see more that don't try and boil the ocean here in business transformation, but are much more focused. So that's a big theme. IPOs, IPOs, IPOs. I was at a dinner with some bankers the first night, and it was all, well, I'd obviously rather own SpaceX than OpenAI and so on.

32:31So while, you know, both of those are - Is that obvious? Is that obvious? I don't know. I think it's quite obvious at this point, Akash. I mean, one, it probably makes a tough, well, I don't know actually how the businesses light up. But I think also one being SpaceX is really in motion. I'll be interviewing Sarah Fryer, the OpenAI CFO tomorrow, so we can get their IPO update then. But, you know, when you step out of the valley, which is good to do for me from time to time, there is still a lot of long-term skepticism about players like OpenAI. And, you know, I find myself pointing out what they've accomplished in building a consumer product at scale at, you know, faster speeds than anyone.

33:17But still a lot of skepticism. And whether that changes by the time they hit the public markets, you know, time will tell. And so looking ahead, all that skepticism, what sorts of questions does that raise for you in terms of reporting questions? things that you are hoping for answers for, not just from our newsroom, but from executives at the forum that you're talking with? Well, I think over time, you know, these businesses will either become incredible businesses or they won't, right? And so I think, and obviously we're biased at the information, we're really focused on the business and on the metrics and on all of those underlying pieces.

33:55I'll be interviewing Sarah along with many others on an infrastructure panel tomorrow. And I think obviously that's a key piece of the equation. And then the very big picture, you know, Brett Taylor was also saying he hedged his bubble language so eloquently. I don't have my notebook, but, you know, we're probably obviously or something like that in a bubble. And, you know, I pushed him on what would change that or, you know, because the capital has been flowing into so many startups for so long. He argues, you know, it's leading to too many startups in all these categories. And eventually there will be a winnowing out.

34:35And I said, what changed that? And he said, you know, probably something unforeseen in macro because the engine is chugging along. So lots of big ideas to follow out of here in the coming months and years. Great. Well, Jessica, I will let you get back to it because I know that you've got a couple, just a couple other meetings to get to. It all happens at night, Akash. So it is really a 6 a.m. to 1 a.m. kind of cycle here. So you can do it for a week. Maybe I'll give you a breather just to rest so you can be fresh for tonight. Thank you so much for joining us. That is the Information's Editor-in-Chief and founder, Jessica Lesson from Davos here on TI TV.

35:20Okay. MongoDB has a new CEO. CJ Desai took the job late last year after a lengthy career as a top exec at ServiceNow and then Cloudflare. MongoDB has outperformed the broader enterprise software sector over the past year, although SaaS companies broadly have had a tough go. It all means high stakes for CJ's debut, and I want to bring him on to talk all about what he's got in store for the company. CJ, welcome to the show. It's great to have you here. Akash, it's great to be here with you. And thank you for that kind introduction. Yeah, I'm excited to talk all about what you have in store. So look, you took the job on November 10th.

36:00And by my count, that means it is day 71 on the job, which means that, you know, you're making some progress. I'm sure you're setting some goals. I want to understand And what was your 100-day plan for MongoDB in terms of where you wanted to take the company and building blocks you wanted to put in place? And what progress have you been able to made against those targets that you set? Absolutely. And 71 days, I can tell you, has been a lot of fun. I have been traveling around the world, meeting many employees, of course, and most importantly, many customers. and also many investors. So my goal was at least 100 customers, 100 plus investors.

36:51And if I can get to 80 to 90 % of employee base that I can go and meet them in person, that was definitely goal number one because then you really get a feel for, you know, what's going on, the strong momentum we have, how are the employees feeling, how are the customers feeling, and of course, how do investors perceive MongoDB? So that's been the goal. And so far, I would say against that goal, we are doing fairly well in meeting the customers and understanding what investors also expect out of MongoDB. Now, from a priority perspective, Akash, it's pretty straightforward. We want to be the strategic data platform for businesses that matter, which is Fortune 500 and Global 2000.

37:43So that's the businesses that matter because that's a large TAM. But also the businesses that matter or companies that matter is digital natives or AI natives that are from San Francisco area, Seattle or New York City or Tel Aviv. and we want to make sure that they are also building on MongoDB. So we reintroduced or relaunched, probably is the right way to say it, MongoDB last Thursday, which was January 15th, to San Francisco. It was super well attended by founders, engineers, and many. And that is very important to us. It is almost after four years we relaunched MongoDB. and I split my time, Akash, between your city, 50 % New York City, and 50 % Silicon Valley when I'm not traveling.

38:33And that allows me to have a solid interaction and base with what's happening in Silicon Valley right now. So I want to talk to you a little bit about the market that you're playing in, the data services, data storage market. On this show, we've talked a lot about the cost of compute, and that is very integral to the AI story. And we've talked about the journey to bring those compute costs down over time. Of course, the chip makers and chip designers have their newer chips that they're unveiling. I'm sort of less clear on what's happening with storage costs because, I mean, there's the cost of the chips to sort of run and train the models.

39:14Then there's obviously somewhere that the data needs to sit, and that is the business that you're in. Tell me a little bit about how storage costs have been changing over the past couple of years and how that story is changing now in the era of AI. I would say, you know, first, Akash, when you think about the database industry and you look at somebody like Oracle, that, you know, a company that was created at scale, now the data industry or data storage layer, databases, is people use different data services like you said that has been around for 50 plus years right depending on when you start counting the calendar on database software some could argue that it's been 60 plus years so this market is a durable market as in data storage or database no matter whether you're creating a brand new ai company where ai application is your killer use case or you are creating an agent at a Fortune 100 company, you still need that data layer.

40:20So that's number one. It's a very durable market, and you will always need a data layer, whether you are using AI, not using AI, modernizing your infrastructure, if you're an insurance company, or you are an AI-native company in Silicon Valley or New York City, you always need a data layer. Now on the storage cost, for example, MongoDB works across all three hyperscalers. So we are available in Google Cloud, Azure, AWS, but also we are available as an on-prem if you decide to run MongoDB in an air gap network because you have a critical application running inside your firewall. and storage cost is always, I would say, something that we pay attention to because you have both when you're running a database, when you're doing queries, when you're writing, there is some part that is compute, and then, of course, there is a part that is storage.

41:16And our goal is always to make sure that as the storage cost, from my perspective, they continue to go down. And if you think about loads and loads of data that run into MongoDB Atlas, whether it's whichever cloud it is over time customers that i have already spoken to have not raised that as a concern that when we run in a hyperscaler gcj whether it's google cloud storage or whether it's s3 buckets or whatever the case might be this is really important to us however akash what i would argue is that when you saw outages in hyperscalers in 2025 all three big hyperscalers had very prominent outwriters.

42:02And because of that, actually the cost that comes up in the conversation with customers is resiliency. So you go to a hyperscaler A, and there are great partners, that's where MongoDB runs for most of our customer base. They will say, oh, CJ, you know, my hyperscaler is saying we need to have intra-region resiliency that is expensive. But that is the part that's getting more expensive is basically the cost of protection, the cost of ensuring that this actually keeps running over time. That is 100 % true. That actually comes up more than the storage cost because if then you say, I don't want all my eggs in one basket as in one hyperscaler.

42:42Now, if you want to do resiliency across two hyperscalers, that is very expensive because data egress cost and other things that they'll charge you. So that comes up to say, CJ, how should we think about it? And MongoDB has an advantage because we can run our clusters across multiple cloud, and you can get resiliency between the two clouds or within a cloud two regions. So that's our architectural advantage. But I don't have the storage cost question that you asked me, which is a fair question come up. Yeah. Well, okay. So, but you said from your perspective, storage costs are coming down. And really, the cost that is coming up is the resiliency cost.

43:21I think that makes a ton of sense. What I want to understand a little bit is now in the era of AI, as data is becoming so much more important to training these models, and there are a lot of companies out there that are developing applications that they have to buy more data, they have to store it somewhere. It strikes me as a clear opportunity for your business, and the stock price certainly reflects that. What are the threats facing MongoDB's businesses right now? I would say when I, you know, and that's why I started with when you asked the question kindly, that customers is our North Star.

43:59And speaking to customers is where you understand really what the opportunities are and what potentially the threats are, right? I mean, that's your question. From my standpoint, in speaking to all of these customers, one of the positive surprises to me was that MongoDB actually is the foundational data platform where large banks, healthcare organization, public sector, or AI native or digital native companies are running mission critical application. I mean, that's pretty huge that they are running mission-critical applications. And our goal is to continue to serve them as they run this mission-critical application, whether they run in a public cloud or in the private cloud.

44:39Now, from my standpoint, serving them at the highest level, making sure the performance, because we are a unique database from my standpoint. as I said, the industry has lasted for 50 plus years, but this was the disruptive force that got created in 2007. And so when I'm speaking to customers like this one large bank in the UK, I was speaking to our sales team feels great because, and they should, because this is an eight digit ARR account for us, right? So the sales team is like, Hey, CJ, we have grown a lot. The bank runs mission critical applications on MongoDB. And when I spoke to the CTO and I said, okay, how many applications truly run MongoDB?

45:25And he said, CJ, currently you are only running less than 10 % of my workloads. So as we feel more comfortable with MongoDB, we will continue to expand for our new workloads or existing workloads on MongoDB. So from my standpoint, the opportunity is huge for us as a company, as a data platform. And really, how do we serve our customers at the highest level so we become that strategic platform, data platform for them, for real-time transaction processing, whether AI or otherwise? That is the opportunity we have. So I don't see it as a threat that we need to continue to earn the right to become that strategic data platform.

46:11Have you played around with Anthropic's latest release, the Claude co-work tool that everyone is talking about? Have you had a chance to play with it at all? I did. I have a huge respect for Anthropic. They're a great partner of MongoDB. And Mike and the team, Rahul, everybody there, they're just fantastic people to work with. So they did this release and everyone is going nuts over it. I mean, you can do so much more than people thought now. And I think this was the whole point of these tools is as it gets less and less technical in terms of what you need to build, you can build whatever you want.

46:50It also does re-up the discussion around, well, is AI a threat to the enterprise software sector? And, you know, where does that story shake out? I hear your point that data services, data storage, the market that you're in, I mean, there does seem to be a bit of a higher bar in terms of the infrastructure, architecture you need to build in order to offer that service. But taking a step outside of your business, where do you land on this debate? Will AI kill enterprise software? Will it replace it? Does it change structurally, the business? Where do you land on that? You know, I would say not only investors ask me that question, given just my personal experience in enterprise software for a long time, but also customers are asking me the exact same question that if LLM or frontier model providers continue to move up the stack, which is your question, they continue to move up the stack.

47:47I can create a software much easier from a velocity perspective, ease of use perspective, then is the enterprise software stack I'm running, is it truly sticky, right? I mean, that's the question. Can that be disrupted? And from my standpoint, speed matters, but also the focus on use case matters. So it is absolutely true that with AI, your software creation velocity continues to go up. And if your software velocity continues to grow up, which use cases do you focus on? And then for the customer who is looking at the enterprise software stack, do they say, this is good enough and I'm going to build my own CRM, for example, right?

48:34Which is the question, am I going to build my own CRM? We do see some companies, AI native companies here on the West Coast, United States West Coast, who say, I'm just going to build a bunch of applications myself and I don't need a package software. Versus some large companies say, is that the business I really want to be in? I just want to focus on my banking application and not worry about CRM. So it's more of a threat for the smaller businesses maybe who don't have, the switching cost is less. That's correct. And where you don't have a lot of stickiness, products are always replaceable, platforms are not.

49:14So if you don't have a lot of stickiness, which is typically in the SMB segment, then if you have exposure to SMB segment as an enterprise software company, that is real. But if you are mainly serving Fortune 500 Global 2000, then the threat is there. But if you continue to innovate and innovate fast, then you can continue to stay sticky. Right. Last question for you. Talent is something we talk a lot about on this show. And I wonder what your pitch is to the top AI researchers, the top AI engineers. They have this decision here. They can go join a fast-growing AI startup, maybe even a late-stage startup that may go public in the next 18 to 24 months.

50:01Or they're looking at all these enterprise software companies, of which MongoDB is sort of part of that generation, right? I mean, I'm just thinking in my head, I mean, it must be a tough pitch in some cases to convince them to come to the, quote-unquote, the older enterprise software world that you are operating in. What is your pitch to them? How do you convince them? My first pitch I start with that we are only 18 years old. So we are not old, old and old and age is relative. So age is relative. So we are not super old. And data, data is the fuel for AI. Data is the foundation for AI. So can you come here and continue to innovate?

50:42And I do have this conversation. I actually had this conversation with somebody just last week. And they understand that data layer is here to stay. And if they can leverage that data layer from long-term memory perspective or persistent layer perspective, MongoDB can be a great place to innovate on because the customer base, we have 62 ,500 customers, Akash. So it's a massive impact you can have in a classic B2B software company that's not that old. Great. Well, CJ, I want to thank you for coming on the show. It is great to have you here. And I look forward to seeing all of the initiatives that you implement at the company.

51:19That is CJ Desai, the president and CEO of MongoDB here on TITV. Thank you. Okay. Google's Gemini is continuing to get some great traction as developers rush to use the tool with their own programs. Our Google reporter, Erin Wu, wrote a deep dive on that trend. Her story had a ton of exclusive reporting and inside anecdotes, and I want to bring her on to talk all about it. Erin, welcome back to the show. It's great to have you here. Thanks so much for having me. So what did you find in your reporting about Google's Gemini traction with developers? Essentially, over the past year, as the quality of Google's models has improved, the number of people using it has gone way up.

52:02And so looking at the growth curve for the number of people using the Gemini API, it's really just exponential growth, increasing growth. You can see that in the curve. And so the numbers we quoted in the story, it went from 35 billion API calls. That's essentially when someone makes a request to do something with the Gemini API to$35 billion in March to$85 billion in August. And so almost a 3x increase in just a few months after they released Gemini 2.5. And that was the first model where people were like, oh, like Google released a good model. Like Google's catching up. Okay. And tell me, in terms of the business model here, not having worked with APIs myself, which I know is actually probably the most basic thing that everyone should do in their life, from what I understand.

52:47I haven't done it. I'll get there, okay, to our listeners. But the business model here, in terms of what it means, they make money every time a developer accesses the API. How does that translate into revenue? Yeah, so Google charges for API calls, so there's like a small cost associated with that. But what I learned from talking with sources is that it actually goes beyond just the money that's going for API sales. And so they've calculated internally that when someone spends money on an API call, that leads to a lot more dollars being spent elsewhere on Google Cloud because now they have to pay for storage, they have to pay for databases.

53:22There's all of these other products that now just becomes easy to buy because they're already doing this AI stuff via Google. And so there's actually an outsized impact from these API calls, even though the API calls themselves aren't necessarily making as much money. So this is the whole get them in the door so that they spend more money elsewhere type of strategy. And so what does your reporting show the net impact this could have on Google's cloud revenue overall? Right. So this could be good for Google's cloud revenue overall again, because when people are coming in to spend on Gemini, the Gemini API, they're spending more on other things.

54:03And this is even though it's been only recently that the Gemini API has started to have even positive profit margins. Like the first couple models, Gemini 1.0 and 1.5, had largely negative profit margins because Google was discounting so much to get people in the door. And even Gemini 2.0 only sometimes had positive profit margins. But they've started to turn that around and charge more with 2.5 and later models just because the models are better. and so now they can compete on quality instead of just the price okay so more api calls more revenue more profit from what i'm hearing sounds like all great news yes the next question i have is are developers liking what they're getting in terms of the tool that they're purchasing yeah and so like i read about two things in the story essentially google's trying to sell the gemini api but they're also trying to make a business selling software on top of that because as we're talking about margins, that's also where you can really start juicing the margins.

55:07If you're selling more complicated software, you can basically just charge more for that. So the flagship enterprise product that they're selling now is called Gemini Enterprise, as you would expect it to be. It's essentially a mixture of the chatbot, and you can search across your company's information, and you can also build agents using this. This is something that Google is expected to highlight during earnings. The product in some metrics is doing like really well like it's already at 8 million paid enterprise subscribers and then like another like more than 100 million people have signed up online at the same time like i talked to a lot of people using this there have been some like mixed reviews like there's some people who are really happy about this like they're using it for a lot of things and then there are some people who are like it's kind of like this low key like does not work like there's why doesn't it what are the issues i mean sometimes the agent builder doesn't work like sometimes things don't connect.

56:02Like I talked to one consultant who was showing me like, look, like here's a screenshot of like trying to make like the most basic like summarize my emails when it like was not working. And I talked to another person whose like job is like working with companies to install Google Cloud. And he was saying like, yeah, like I think it's a good product. Like the slim majority of my customers are in favor of it, but it's close to like 50 50 of people who like it and don't like that said like people are signing up for this right now another analyst i was talking to was like well like you know like people are still like i'll give this a shot like they're not they're not disappointed enough to like completely stop using it but it's it's a nice him product like there are kinks to be worked out like the thing doesn't totally work but like people are excited are those are those kinks that you talk about, are those unique to Google's Gemini?

56:55I mean, the issues, are they unique to Gemini or, I mean, do they have the same complaints about OpenAI's products? Yeah. So, I mean, I think like the better comp here is like looking at something like Microsoft, looking at something like Salesforce. And we've written a lot about how both Microsoft and Salesforce are really struggling to try to sell these products to enterprise customers. And so, So definitely Google is not alone in this. I mean, Google, if anything, would argue, look, we're doing so much better than Copilot. We want to see this as the Copilot killer. This is what can give us a foothold against Microsoft.

57:28But everyone's kind of struggling with this. It's very early days still for enterprise adoption of this technology. Great. And last question for you, Erin. In terms of questions that you have continuing your reporting on this topic, what are you wondering? I mean, I really want to know when this stuff is going to start working. I want to know when we're going to start seeing broad scale enterprise adoption of these AI tools and the impact that that's going to have. There's obviously a lot of questions coming up, especially around earnings, how this is going to boost Google's overall cloud revenue, what that means for Alphabet's growth.

58:06Investors will obviously want to know what's the return on investment that Google's getting from like the$90 billion in topics it's spent next year. So there's this is a story that like in some levels, like seems kind of small, but also like ties into a lot of questions about like Google's performance and like what's happening in the enterprise. So it's definitely something I'm going to keep watching. Well, and I think that the most revealing part of your reporting, I think for me, is this idea that if the purpose from a strategic perspective right now with these APIs and with Gemini is to get people in the door so that they spend more on the traditional cloud business that has obviously carried the Google Cloud segment overall.

58:50I mean, that certainly tells you something about the relative sizes of these operations right now and just what the overall strategy is. So, Aaron, I want to thank you for coming on. We, of course, have Google earnings coming up in a couple weeks, and so I am excited to speak with you then, And if not before about what we learn, that is Erin Wu, our Google reporter here at The Information. Okay, that does it for today's show. A reminder, we are on this stream Monday through Friday at 10 a.m. Pacific, 1 p.m. Eastern. I want to thank you all for tuning in. We really do appreciate your viewership.

59:23I'm already excited for our next show tomorrow. Have a great rest of your Tuesday. Bye-bye for now.