The New Rules of AI: Inside Google’s AI Surge and Growth Strategy | Fast Company

Well, first, it's great to be here. Great to be with you and everyone here. I think what's most exciting about IO is that we firmly are saying and you're seeing that Gemini is in the agentic era. We've pushed what we're doing with Gemini, announced Gemini 3.5 Omni which is the ability to really ingest any form of content and we show today what you can then create with videos. You actually have to see it I think to believe it. But then really the opportunities that come across our platform. So for me one of the most exciting things is seeing what's happening with search and we think this is the biggest evolution of search ever. I think what is really exciting is we're now today we have or we are announcing that we've reached a billion monthly average users on AI mode and people are just using it differently, different ways, probably hopefully everyone here is engaging in different ways where you can go deeper and now adding capabilities. So it's a pretty long list but you see the Gemini applied to YouTube, to photos, to docs across the board and what that enables each of us to do to be more productive and more creative, kind of to push the frontier. And then I think the other part that's really exciting, it's a long answer because there's a lot and I know and we're only just going through it. It's been a lot of work and we're really excited to be here. It's mostly the momentum that we're seeing across the platform, but it's also the work to protect on the downside. Like we talk a lot as you know well that we want to be bold and responsible and so we're continuing to push what we're doing on things like SynthID which is the authentication something's been created with AI, want to make sure people are aware of that. We are also talking about something called Code Mender which is an extension of what we're doing in cyber security, it came out of Google DeepMind, the ability to identify cyber security vulnerabilities and patch. So a whole host of things on the upside and to protect.

It's gone up a bit.

It starts with a couple of basic kind of critical pillars which is one when you see platform shifts in particular something this profound we haven't seen anything this profound in our lifetime. You cannot miss platform shifts. You don't want to be behind the curve on this point number one. Point number two, when we look at the upside from AI, it is, we think, an incredible privilege to be living today when you're focused on what you can do with technology to advance science, drive economic growth, economic upside, improve delivery of critical social services, the advances in health care, in education, all that's necessary in cyber security and fortifying security. That upside potential is profound. And so to deliver it, we clearly need the compute capacity. If I go back maybe five years ago as we were starting to look at what the slope of these curves could be. We had conversations to your question about capital allocation which is to what end can you use this level of compute. And I think one of the very important lessons from the early days of Google and others is if you force too quickly to identify precisely where you'll get a return on an investment, you can get a quick tactical win, but that doesn't mean you're building for the long term and building for something that's actually going to matter. And so one of the key elements is how do you lift up and try and project out states of the world over a period of time? And what we're seeing even though it's still very early days on this journey is that we consistently talk about the biggest challenge is allocating the compute that we have. We are compute constrained given the opportunity and it's what we need internally. It's working with private sector, public sector. We see the upside to be quite extraordinary and we want to make sure that we're laying the groundwork and the foundation to be able to deliver on it. So we feel it's like I think to go into that level of growth you should have a certain level of anxiety and humility but we feel very firmly that missing this would really be the error that we cannot make.

I think there's a really important point that you should never rely so greatly on the external assessment when you're either down or viewed as at the top. Because if you believe in your core strategy and plan, sure it's great to have external signals, but to stay focused and resolute on execution and to have when you're at the top to have a healthy level of paranoia and stay focused and resolute on execution is really key. And I think there were sort of two tensions going on. On the one hand, it was back in 2016 when Sundar said we are moving from mobile first to AI first. And what that means is we are going to invest aggressively to lead in AI and we're going to have a full stack approach which at the time, what is that full stack approach? Well, it's models, it's chips, it's research and it's the application across all of our platforms. And he was very clear and amongst the senior leadership team it was very clear we will invest to continue to lead and be AI first. You're absolutely right that one of the areas that we care immensely about is delivering for everyone in a high quality way and models were hallucinating and as Sundar I think talked about in the piece one of the questions was if you wake up in the middle of the night and your child has a fever and you want to give your child Tylenol it better be right and Google stands for quality. And so that to us was very important. But I think things don't happen overnight. And the position that we're in and the momentum that we have that we're really proud of and remain, I think, paranoid in a healthy way about is because we've invested against this for many years. You know, chips, TPUs, we've been investing in for more than a decade. Google Brain we created in 2011. Google DeepMind we acquired in 2014. So there's been a consistency about investing that I think has served us really well. One of the other elements though to your question is we took a look at what we could do organizationally to accelerate the pace at which we were operating. And I think just like anything in life, the world's not static. It's super dynamic. So we should be organic about this. And there were things that we did that eliminated, for example, some seams between organizations. We combined Google Brain and DeepMind to create Google DeepMind. And I think that sort of organizational alignment helps. We also did something that I think is really valuable. We have a group called Google Labs which I think Sundar spoke about as well which spins up ideas pretty quickly and is very quick to kill them if they don't make sense or to launch them where they do make sense. That's where NotebookLM was from. Hopefully everyone here has used both of those and more and continuing to come out of there. So that kind of ethos has been really key and then just staying resolute about it.

So, I think one of the most exciting parts with AI is how we're going to reimagine what we're doing. It's the innovation that comes that's actually pushing the frontier. That's one core part of it. The operational effectiveness I view as sort of low-hanging fruit and it gets you on the journey and we should all be doing it. And so this kind of executive cockpit as we call it. My team found that they were spending a lot of time trying to ingest content from across Alphabet, ingest to make sure that they had the best of throughout. If we'd been working in a country, spent a lot of time around the globe with...

Exactly, the executives for the president CIO's office. We need to ingest a lot of content and it took a lot of time, it was too much sort of even at Google hunting back trying to find it and so the ability to ingest easily, to analyze easily, to follow up easily. It's a sort of basic, I would call, you said it correctly. It's administrative drudgery. So why do I get excited about it? That's just an efficiency tool. But that's given my team in particular ballpark 30-40% of their time back so they can focus where I need them to focus. And I think if time is our most precious asset, step one before you can get into really innovation and change is get rid of the stuff that's low-hanging fruit that's getting in the way, give everyone back the gift of time. That's why I'm excited about is kind of step one.

Well, if I go to life-changing leadership advice, probably the best sort of portfolio of advice came when I had the amazing privilege to work with then Secretary of the Treasury Hank Paulson during the financial crisis back in 2008. And the reason I go to...

I was at Morgan Stanley. I was running a business that covered all of the banks and insurance companies, asset managers. The economy was going, it was clear things were moving in the wrong direction. He wanted to have a team come down and advise, work closely with him. So I ended up leading a team of about 40 people. We would go around the clock trying to analyze what was going on during the housing crisis with Fannie Mae, Freddie Mac, and then over to AIG. So living through that period with them when I got to Google I was asked what were the lessons from the financial crisis which struck me as a totally bizarre question coming from a place where things had only been going sort of up and to the right and I realized that what I learned from him, what I learned going through that is relevant in good times and bad which is why I think most important to me what I anchor back on, the first thing was he said very clearly identify your greatest source of vulnerability and protect against it early because you cannot protect against it in that moment when you need it most. Now in the financial crisis that was how do you have the liquidity to run a bank but I think today it's relevant because AI provides such operating upside, revenue opportunity that if your competitors using it and you're not and the slope of the curve is pretty steep it becomes very difficult to catch up later and so getting on that runway as quickly as possible I think is key, one of the key things that I would take from that. The other that he said is you must have the will and the means. And too often by the time you basically have mobilized all of your team and have the will, you no longer have the financial means to catch up. And again, it was in a different context, but I still think it's relevant today because trying to mobilize, build that consensus is using that going back to the asset, time. Time is so precious and you may not be able to catch it later if you squander time. Those would probably be two of the most critical ones. And then the third sort of going back to your first part of the question, build a team that gives you horizontal vision. Now technology can be an asset in that regard, but the world is changing and so you really need to have the broadest sort of vision. And I keep saying give me horizontal vision, I can then connect the dots to the future and that's the other part of those lessons.

Well, you've hit the area that I'm most excited about for a whole host of reasons. First of all, I think that when you think about the opportunities in healthcare, yes, you're right. You're seeing that free me from administrative drudgery part of it. But there's for every industry in every area and most profoundly in healthcare, three things that people should think about. One is innovation. How do you truly take advantage of this opportunity? Two is risk management. And three is administrative operational leverage, free me from administrative drudgery. And that mental model I think applies to every industry. And regardless of the industry I'm addressing, I come back to health because it's so personal to every one of us. And I like to actually bring it to life talking about it in the context of cancer because 50% of the people here, 50% of the people in America will be diagnosed with cancer in their lifetime. I was diagnosed, I've had cancer twice. So this is deeply personal which is why I'm grateful you asked this question and if anyone here is going through this my heart is with you deeply and I'm also deeply encouraged by what we can see is the art of the possible as a result of AI applied to healthcare. The tragic thing, one and two, diagnosed, the fastest growing cohort is people who are 50 and younger. Within that disproportionately women are being hit with cancer. And when you look at the disparity across the United States for those of us who get the privilege of being treated in hospitals here in New York, there are many cancers that actually are treatable today. But the survival rate for us is about 50% higher than for many parts of the country. And I think that disparity is completely unconscionable. And that can also be addressed with AI. To me, the most exciting answer to your question is that we firmly believe that cancer will be conquered in our lifetime because some cancers will be eradicated and others will be made more manageable. And for any of us who've gone through cancer, manageable is a gift. A healthy, long, full life. I'll take that. And it's not just Google. I'm on the board of Memorial Sloan Kettering and I asked the president can I say it this way, do you believe this and he said the word conquered is correct, some eradicated, some more manageable. And I've spoken to people in industry. So how do we deliver that outcome? Number one, drug discovery. AI applied to drug discovery is profound and one of the best examples is our colleagues Demis Hassabis and John Jumper pioneered what is known as AlphaFold, they won the Nobel Prize for it. Simply stated, it's about 3D prediction of proteins. It's protein folding. It has been described to me as the single greatest contribution to drug discovery in our lifetime. And we open sourced it. It's being used by three and a half million scientists around the world, 190 countries. It's not just addressing cancer, it's addressing a lot of diseases, but accelerating drug discovery. Number one. Number two, I said risk management. So what is risk management? It is finding a needle in a haystack. That is early diagnosis of disease. We pioneered early detection of metastatic cancer. That can be a game changer. And when I asked my oncologist how important is this, he said the only way to democratize healthcare is with AI. It starts to get to that 50% differential. And then you brought up the okay, it's getting rid of administrative drudgery. Why is that important? His answer on why that's important is because it restores humanity in the patient-doctor relationship, the patient-nurse relationship. It's also preventing burnout. I think that's a really important element of it. So those three are here today and that model is relevant in any industry. And I would say when I was first diagnosed with cancer, that's what I can talk about forever because I think it's so important for each one of us. My kids were little. They were five, seven, and nine. And it was terrifying not knowing whether I would see them graduate from elementary school let alone anything else. And to sit here today and say that it is conquerable in our lifetime. My view strongly is when a technology can address issues like this we must. If we can we must. And I say it this way and it's been something that I've been spending a lot of time thinking about because clearly there's a lot of anxiety about AI as well. And what's the downside? Well, we need to invest to protect on the downside because that upside, if we're not chasing that upside and addressing that upside, shame on all of us. It's unconscionable. So, that's why I think there's a whole lot more that is truly in flight that is here that is going to make a difference to each of us, not just in health, but in education, in crisis alerts. It can be used in other areas. It's supporting small businesses. That's what we need to capture. And it requires both amplify that upside and protect on the downside.

Well, I'd love to do a rebrand. I mean, even if it had been called augmented intelligence in the day, I think it would have been very helpful. And a lot of it is augmenting intelligence and advancing what we can do. Look, I think the narrative is part of it. Look forward to working with you on that. I think people need to see the art of the possible. What is the upside opportunity? And a lot of the anxiety is if you don't see the upside, of course, you're going to focus on the downside. I think it's human nature. I view it as two building blocks that will help turn that narrative. One is there's concern about energy pricing, energy affordability. And what is really not clear is that electricity prices, according to Lawrence Berkeley National Labs, some research they did, it's gone up faster in states without data centers than in states with data centers because the large workloads help amortize the cost. But beyond that, that's sort of like a nice starting point. In order to be able to continue to build, we and others have said, look, when we show up, we're going to add energy capacity. We're going to pay for not just what we use, but also the infrastructure to support the growth. We're going to invest in the community. We're going to immunize effectively from the impact of our growth. So this is not as well understood and probably shame on all of us it needs to be but I think you actually said something else that's super important. People across the country in the heartland of America in the industrial have to see this affecting their life positively and whether that's through healthcare, education, I think small businesses are super exciting. About half of small businesses in America which are the engine for job growth are using AI, those who do are seeing an uplift in revenues and an uplift in profitability and we saw that and said we're giving free Gemini training to every small business because every town has small business leaders who are super respected and they can see what the upside is. So, I think it's kind of hitting the two and then absolutely making it really clear why it matters. And final point on this, what struck me is intriguing, I was at Stanford University not too long ago at a symposium. It was called technology and national security. And a lot of the conversation was about US versus China. And the reality is that we're not going to be able to deliver the upside from AI if we don't have the energy infrastructure to deliver it. And with anxiety about building out energy infrastructure, it creates sand in the gears. The conclusion from the symposium was potentially the greatest threat to national security is actually anxiety in America about AI more than anything else. And that really what's imperative is to work together to make it clear how the upside benefits people across the country and how the downside risks are really protected because it's so important for our children, grandchildren, and all of us. And so we're trying to get upside, downside.