Au-delà de la keynote avec Sundar Pichai

Welcome, welcome, welcome to >> Thank you. Google IO 2026. This is the dialogue stage. My name is Matt Berman. I'm the CEO of Ford Future. And today I am super excited to share a conversation with the man who has been leading Google for the last 10 years. Please help me welcome Sundar Pachai there. All right. Well, thanks for joining me. A real pleasure to be here. Congratulations on all the announcements, of course. Uh, I kind of want to dive right into it. I actually recently found out that you're the first PM of Google Chrome. >> Oh. Uh, yes. >> Yeah. uh you you have a unique insight and probably very strong opinions about the future of the internet. And so that's where I want to start. It seems like the internet is being transformed right before our eyes. Agents are being built. They're infiltrating the internet. And I'm wondering, do you see the future as agents being the entry point to the internet for most people? Uh it's first of all uh great to be here. Thanks for doing it Matt and love your show. Uh I love your content and I appreciate all of you joining as well. Uh look I I I do think you know agents are going to be a fundamental part of how we work because having used them you know if you're living you know maybe today the people who are on the frontier of how agents work are developers right you know and particularly with coding you know two years ago most developers started using these tools they they started giving them more and more autocomp completions and you were accepting them etc. But over the last few months, developers are actually doing agentic workflows, right? The the developers on the frontier and they are actually deploying agents, orchestrating agent. You saw the demo in anti-gravity for building an OS, you are effectively in an agentic workflow. So I think once you get the taste of using something like that and the superpower that comes with it I think so I think it's they're genuinely adding value in a way I think people will will use them. I think it's important we build it in a way that users feel a sense of control and agency and transparency when they use agents. I think that's important uh that's an important foundation but I think yes I do expect agents to be a core part of how we use the web but that doesn't mean it'll take away from people use the web for a lot of reasons right you know you're entertaining yourself you're trying to do something meaningful at times and you know it depends on if you're shopping what you're shopping for you know if it's your weekly groceries versus you're trying to buy your loved one a gift, right? And so I think it'll allow humans to use the internet in ways that gives them joy and purpose and not always be forced to deal with I have to fill these 18 form fields to renew a DMV license, right? So, you know, so that's how agents will separate it out, I think. >> Yeah. I mean, I'm I'm particularly excited to have agents actually do real world tasks. You mentioned the DMV, that use case is incredible. Um, but I also think about putting so much trust into agents to really uh be the arbiter of our information diet and and I'm wondering how how do we make sure that that we are putting the trust in the right agents and the agents are deciding correctly what information that we should have. >> We are already you know you are doing a version of that. I don't know like if you use Gmail and there's a spam filter working on your behalf >> filtering spam like in some ways you're like trusting an agent. It is an agent, right? Like you know not the way we think of agents today. So I think we've always you know I I look at it as working on Whimo. You have to make people trust sitting on the backseat of a Whimo and let the car go. You know that is an agent in some ways. But people are willing to trust. But that's because we've done the work over time to demonstrate to people both with data with how we have operated it that it's it's fundamentally safe and it's freeing you up uh to enjoy the ride. And so I think I think it's all depends on the value you deliver, right? And I think that's why I think the building the agency and trust with users is a shared journey and we have to get that part right. Part of the reason in Gemini Spark, Gemini Spark is actually very powerful under the hood but we are taking the deliberate careful step of making it work with your first party services like Gmail, calendar etc. before we expose third party with MCP and full computer use and browser use. You know it can do all that but we want to make sure users are in control and they feel comfortable we getting their feedback improving the product as we give those capabilities. >> You know Sundar I if I'm crossing the street you mentioned Whimo I actually trust walking in front of a Whimo more so than I trust walking in front of a human driver. I I don't know if other people agree with that, but there there is it was almost an immediate trust that I had for Whimo, so I'm I'm hoping agents are the same way. Um I'm old enough to remember the early days of the internet, which was an absolute wild west, but you got to raw information very quickly. And I think, you know, part of something I'm concerned about is that we're increasing the the buffer between us and the raw internet. We saw it, you know, a little bit with the browser, a little bit with apps, and now more so with agents. How do we account for that? >> I think, look, um, people in our experience with search and YouTube, YouTube is a great example, right? People have this sense of connection with the creators they like and follow, right? And, you know, so that's a big part of what they're looking for. So I think people are in different mindsets. I mean there are times they want to discover content on the web like shopping is delightful for a lot of people in a lot of moments right and so they're not trying to fully outsource that right like you know that's why I try to distinguish between what feels like something which may be a chore at times and what feels like something which is delightful. Similarly, I think you know uh it could be in news, it could be people have their trusted sources at at least at Google through search and YouTube and through agents. We think you know there will always be an incredible value from the ecosystem which users want to connect to and the agents should be in the job of doing that. >> Yeah. Um but you are right. There are times the agents are playing a role in the middle. Sometimes it's very good because you know it helps improve user satisfaction because users are getting to what they want in a better way but but there's a layer of abstraction too and that's what you're talking about. Yeah, >> but I think it's always been true with technology a bit. Uh uh but I don't think ultimately you know it's also make at the same time it's also in the tools with which you can create content are also exploding. So people will also be creating more content. So I think there'll be a new balance which we will find. But yeah, it is an interesting moment of evolution. >> Yeah, I I like that. But I think I think you're saying there's definitely going to be a strong place for agents to do that curation on our behalf, especially in the era of complete uh slop domination. >> Uh but also, you know, that feeling of exploration can still be there. >> Yes. >> Um Okay. >> Because it's a fundamental human need that doesn't go away. >> Yeah. Exactly. Um Okay. So speaking of the wild west and I I know you're probably feeling this pretty strongly on the uh at Google there have been increasing cyber security cyber attacks the models are getting better at cyber um obviously Google has been thinking about cyber security for decades are you seeing cyber attacks especially AI enhanced cyber attacks ramp up uh at Google >> look we've been seeing being to be very clear like we've deeply cared about cyber uh you know because we've been working on frontier technologies for a while like Google pioneered many of uh important security frontiers like zero trust and so on right like you know we we have worked hard to keep the company at the frontier and also we operate many products and platforms around the world that that touch billions of people I we've been pretty aggressive in deploying agent Agentic workflows. Our internal security teams use agentic workflows to help detect vulnerabilities and then how do you work to patch them? And we we have steadily seen over the last two years as the model capabilities have progressed you know we are able to detect more vulnerabilities and we've been working very hard to patch them. I think mythos was a point of inflection of capturing that moment in time. Um you know that they put out a model which was really you know well built for that particular task and it was frontier there right but what we are excited about it is part of the reason we are sharing maybe one of the uh undermentioned announcements today at iOS code mener. So, code mener is a product which we use internally and which you're building to share externally. It not only helps you identify the vulnerabilities, generate patches, test and verify that they work and deploy them >> and it's running 24/7. >> That's right. And like you know and and real time we completed our recent acquisition of viz you know so viz is state-of-the-art in being able to do this realtime monitoring uh of of vulnerabilities etc. So I think combination of what we have with viz and code mender I think we are using it internally uh to stay at the frontier and I think it's an important moment for the industry. I have to say I'm heartened by the cross industry collaboration going on in this moment. uh I think one of the examples I would call out today be it synth ID or watermarking companies coming together coming companies coming together around cyber that is so important for this industry with this technology right so I'm encouraged by uh those trends as well >> so you mentioned mythos so I want to talk about that for a moment obviously anthropic decided not to release mythos publicly just a handful of companies we have openai releasing releasing uh GBT 5.5 cyber which which approach do you think is more appropriate? Which is right for Google? Is it uh you know is there is there some model that is just too good and and you're going to hold it back or is the more iterative deployment strategy that OpenAI takes more aligned with what Google believes? Look, I I think it depends on where you feel it depends on the model capability. If it's if it is not fundamentally changing what's out there already in terms of the state-of-the-art, I think it's definitely okay to put it out. But in the security world, there's a wellestablished practice. Google has done project zero for a long time, right? And like you know we have teams of people who find vulnerabilities then we notify the vendor give them 90 days to patch it before we acknowledge the vulnerability in the in the wild. And so there are wellestablished practices in the security industry around how to do it. So I think it makes sense to me if you suddenly have something which dramatically changes the frontier. First of all, I think it's important to work closely with the government on that >> and and you approach it in a responsible way. So, so I I think that is consistent with how the security industry works and but I do think it's important to also def make sure enough people get access to it so they can patch their systems and so on and and they go hand in hand. So I think there's validity to that approach. So is there some threshold by which you would say past that point uh we can't release it or or is it more let's look at what the landscape of model uh capabilities are let's look at how cyber is right now and let's make the determination on a model permodel basis. >> Yeah that's what I would say like is the next one you're introducing does it dramatically change the frontier? Is it a 1 to 2% improvement over the current state-of-the-art or are you taking these 20% jumps? That's where the judgment comes in and I think that's how I would change my approach. So on the topic of model strategy, something near and dear to my heart is open source. Uh basically Google and Nvidia are the only companies with a real model open source model strategy nowadays. Um, you know, Google's model is smaller, meant to run on edge devices. Why not release a large open-source frontier model? Look, we um, first of all, Google, we've been big fans of open source. Google was built on a lot of open source systems. We have worked on many big things which are open source. I mean I personally worked on chromium and Android and so on, Kubernetes and I can name many projects which Google has contributed pretty strongly to the world in open source in AI. We've been building Gemma models and we've been updating them year after year, right? And they're awesome and and uh you know and I think the recent release of Gemma 4 was a great release and you know so we are pushing it. I think all of us are trying to make you know make sure the frontier takes a lot of investment to get the frontier done right you've you've seen our capex dollars right and like you know so you're working you're putting a lot of R&D dollars to generate those incremental frontier models so I think and and you you're you're discovering new techniques as part of doing those models so we all have to be mindful of that but I think we are also committed to making sure there's a open source ecosystem which is able to develop and and and so we we take a balanced approach there uh approach there and I think we'll continue to take that balanced approach that's how I see it >> yeah and and by the way I do love the Gemma models I run them locally at home they are fantastic so definitely thank you um you you know obviously a company of the size and the resources of Google if you're making that decision of okay large closed source large open source we can't do both um probably most startups in the US also will struggle with that decision. Like what what is your sense of the business model for open-source in America? I is it is it viable right now? Look, first of all, we not only do open source, part of the reason we invest so much in flashlight and flash models is so that we are giving a range of options, right? And and those models are workh horses too to support. But to your question on open source, it's not that we don't, you know, we are we've been moving the open source frontier too. There's a lot of very very good open source models like particularly from China which startups are adopting too. >> Yeah, we're going to talk about that. So you know I think I think it depends right you go through moments in technology where the frontier moves so fast maybe sometimes the open source may not be able to fully keep up with it but then there are moments where open source will take leaps if the technology curve slows down or you know and you know takes a break right so uh it's tough to predict predict predict it fully in the future uh but I expect there to be a demand for a strong open-source ecosystem and I think we will definitely play a part in it and I hope others do do too. There there have been a number of open source ecosystems that have been really successful over many decades in technology but I think just the upfront cost of of baking a model makes it extremely difficult especially if you're putting out the model and then all of a sudden your competitors are serving inference at a higher margin than you. Um, but I I am hopeful. Uh, I I do love open source and we're going to talk about the workhorse models in a moment, but I want to talk about China and and they have been putting out incredible open-source models. um you know if you can put yourself uh in the shoes of another uh enterprise CEO and you're looking at the landscape of which AI model to choose for your business and you're seeing you know deepseek at a fraction of the cost but still near the frontier. Uh why wouldn't America adopt Chinese open- source AI? What's the argument to go with American AI? So you're saying why wouldn't uh uh you just use the best open source models available? Yeah. >> Look, I think you know look at the end of the day what are companies trying to do right? They're trying to solve problems, right? These are and and they're trying to solve a problem with the solution. So the question is what are the solutions available? Remember they are designing let's say you're doing something in customer service. You want predictability, you want reliability like you want consistency and like you know so you want safety, security. So companies are optimizing for a lot of factors. So I think that gives a place for both open source models and there'll be providers who will take the open-source models and like build that ecosystem around it which makes a lot of sense. there'll be closed source models and and you know it'll be open marketplace and people will have a lot of choice. I think I am more okay if it is open source with the right licenses. it should matter less where it came from, right? Like you know I think over time there are good ways to inspect open source. Not not saying that's exactly true particularly with the how the AI models are developed but with open source comes a community which is responsible for it cares about it. So if something is wrong is happening in that software it's not like it's going to go unnoticed. So I think that creates a level of trust for people to adopt that technology. I think so I worry less about are we adopting open source models from China or more that are we doing enough in the US to make sure we are staying at the frontier. >> Yeah, >> that's how I think about it. >> Now I know like Google is is big on code design full stack and and kind of just continuing on on where the open source model is coming from. I've heard that argument that is if it's open source it doesn't necessarily matter. we're going to fine-tune it or customize it for our needs. But ultimately, if we continue to build on top of, you know, China's open source, there's also an argument that they're going to optimize their models for their own chips and then all of a sudden we're kind of built on uh another country's technology. Is is that an incorrect argument? Look, I I think the fundamentals of AI, the way people should be building use cases on top is because the models are changing so fast anyway, you need to build it in a way in which you're able to evolve the models underneath, right? I think that has got to be the way, you know, you're working through this moment, right? And so I so I think you have to be dynamic enough that you have to be able to adapt because the model frontier the model ecosystems are changing pretty sharply and so that's the way I would think about it now and you know it's too early to predict if this is a real concern or not. >> Yeah. Um okay talking about model strategy one of my favorite things is watching the frontier labs and seeing how their model strategies play out. uh anthropic and open AI seem almost exclusively focused on the absolute frontier and Google has that but you guys also put a lot of emphasis on on what I'll call kind of the the workhorse class of models the flash class of models talk a little bit about why why is that such a big part of Google's strategy >> look I I mean in our mission statement we have this thing to to make technology universally accessible and useful >> we've always deeply cared for what is the most important technology in our lifetimes that it diffuses as broadly as possible and we get really excited at driving efficiency and making sure the best models can work in the fastest possible way cheaper cuz we need to do it for search right because we have to give it to billions of people >> we want to put it in Gemini and so we want to give it to developers so that they can do powerful things with And we've had a lot of success with the strategy and I think 3.5 flash particularly I made this point during the keynote but I've heard anecdotally from a lot of CIOS who are so concerned about how much their companies are blowing through budgets. >> Yeah. Right. You can feel it talking to them and I think the problem is going to get worse as we go through the year. Right. And I think that's where I think the flash model will really shine because particularly in a agentic workflow where you need these things to be repeatedly used and used a lot of times I think it's so important to have a model which is very capable but is fast and efficient right and and even accounting for token effic token use flash is remarkably costefficient right and and and so I'm really exced We are finding it internally. We are using it as a blend of pro and flash, right? Internally and and I think most companies should learn to use it that way. To be very clear, we're super committed to being at the frontier on every category. I'm excited for pro which we're working on. Uh but I think flash has a unique role to play in this constrained compute constraint time. >> Yeah. No, I I I agree completely. especially most companies are are not, you know, solving math olympiad problems. They're not at the absolute cutting edge of science. They need real work done and that that is truly where the flash model shines and not everybody is token maxing and trying to or unc unconcerned with the budget. So I I definitely appreciate that. Um, I like if you think about the future of AI, is it just truly a race to self-improving AI? Is is it like maybe just to play the the devil's advocate for a second? The Flash family of models are great now, but ultimately whoever reaches self-improving AI first wins and then nothing else matter. Do you think about it like that? Look, I think first of all I mean there's a responsibility that comes with this technology and you know I think we all need to be careful to avoid this race condition at all costs >> uh and and and and and you know I think we owe it to humanity to make sure we deploy this technology responsibly. You are right in your question that there is this current moment where people feel like the curve is so steep and like you know uh you know where you are in the curve matters but just like in a few months ago when we launched 3.0 I know people are like oh we are so in the frontier no one will ever be able to catch up and you know I think at the frontier labs uh it's very dynamic the competition is fierce we all have our strengths and weaknesses we all also have different cadences of our pre-train pre-training release cycles so the peaks don't exactly match so all this creates that perception gap which shifts widely in like four to six weeks But I think a few few labs are really at the frontier and then there's a big gap. Um and I think I think there are scenarios in which things like recursive self-improvement come into play. But I think if they come into play with that no difference from the cyber moment, we all have to handle those moments far more responsibly than today. Right? And so that goes hand in hand. And so I I think the more AI becomes advanced, the more it's a societal conversation versus a single company conversation. Yeah. Yeah. Well said. So all of uh we know we're talking about a lot about models, but all of it is downstream from compute. Uh I I'm always both kind of uh impressed and in awe of Google's ability to serve, you know, you're serving your own models in inference on the API. You're also powering your suite of products with Gemini to literally billions of users. You're also allowing your competitors to use your inference. You're also selling TPUs. Um, I've I've heard the reason for this is because I mean Thomas told me he said we planned really well, which you know makes a lot of sense. You've been at this for 10 plus years. I think I've also heard that Google's revenue is literally constrained by compute. So I wanted to give you the opportunity what is the state of compute at Google right now? Look, I all of us are I think we have made a set of you know bold right decisions over the last few years to invest in compute and scale it up aggressively. But having said that I don't think any of us sit in the chair and say we wish we we had you know you look back and say I wish I had done a little bit more. So we're living in one of those moments in time and you know there are costs going up too. So like for a given budget you may be getting less compute than you had previously planned for you memory prices what have you you know so the costs are going up. I think we plan you know we are able to plan long-term for cloud separately from our you know own internal needs and we do long range plans and we plan for it. I think it's good and some of it is like when you're on Google Cloud and you're supporting customers your your customers may look at something and say well I want access to that like they may look at the demo of 3.5 flash on anti-gravity and say how are you exactly running it at 800 tokens per second and like could we get access to that you know so you're also supporting customers through those journeys and hence you're meeting them in the in in what they're asking for and but it is not an easy balancing act and you We are we are constantly thinking as far ahead as possible and we are making tradeoffs uh like every other company right now and >> so do you have maybe an obvious question more demand than you have compute to serve it >> absolutely >> like what is the scale of that >> and and and hence the emphasis on something like 3.5 flash even more >> right and you know could we have done an even better omni model yes but like you know how can we do an omni model which we can give to as many people as possible So there's constantly you're making trade-offs of trade-offs of like that including on you know do you build a very large model which you know like an ultraiz model that will increase the capability frontier but then who all can you give it to so you're constantly all of us are making these trade-offs and sometimes you make a trade-off it looks like well maybe you done that trade-off a bit differently >> yeah so we only have a few seconds left one last question for you what is the main bottleneck for compute for Google right now. Is it land? Is it political will? What is it? >> I think I think the way it works is by definition bottlenecks work this way. If you think something is a bottleneck and you solve it, something else becomes a bottleneck, right? Like you know that's the whole like definition of bottlenecks. I think at various times I think you're having a few areas your ability to physically cons permit and construct data centers the power they need in and then very quickly you get into the core components for these systems. They are all the bottleneck. >> Yeah. and and it's a it's it's like you have you need all of that to work together to get a square set of a chip you need right and like you know and so uh I feel like there are a few parallel bottlenecks going through and it almost doesn't matter there are like a few of them which are bottlenecks >> and at various times you may you may conclude it is memory but then if everyone concludes it's memory tomorrow you turn around and say okay no no no it's it's actually this and so you know but but I think there are systemic bottlenecks across all layers of the stack now >> yeah whatever the bottleneck is in the moment that's the most compute you can have in that moment okay everybody please help me thank Darpachai thank you so much thank