¶1A year and a half ago, Google was on top of the world. They released the most powerful AI model on the planet, and it seemed like everything was going their way. But since then, they have fallen behind. The stumbles that Google has seen over the past year and a half have been significant, and we are potentially watching the biggest bag fumble of all time. Google just had a major shakeup in AI leadership.
¶2Jeff Dean who is one of the most famous computer scientists in the world responsible for building out much of Google's core architecture and infrastructure today their 30th employee left to start his own AI company. Then we have Deis Hassabis one of the greatest AI minds of all time co-founded and CEO of Google DeepMind just stepped down as CEO. Why are all these things happening right now? And how did we get from Google being on top of the world to many people questioning their future? Let me give you a little bit of background on Google and artificial intelligence because they have been on the leading edge for nearly two decades.
¶3That's what makes this story all the more surprising. So Google has had AI powering some of their products for a very long time. They've been on the cutting edge of AI. They even wrote the seinal paper attention is all you need in 2017. opened it up, opened the research and let everybody see it.
¶4And by the way, if you're not familiar with this paper, it is the foundation of all modern AI systems. Everything that OpenAI built, everything that Anthropic has built, all things like Chad GPT are based on this paper. That's how important this paper is. And they just did nothing with it. And at the same time, Google DeepMind had developed AI that beat the best players in the world at Go, beat the best Starcraft players.
¶5They really had the best AI systems, yet they just couldn't productize it. And the reason for that, I will tell you in a moment. It is something every company inevitably faces. And so that brings us to this tweet, which really highlights the reason Google fumbled so hard. I sometimes think about that Jeff Dean interview where he said they had an internal bot before Chad GPT but didn't think it was better than just googling.
¶6That tells a lot about what Google was thinking in that moment. But that's actually not the worst part. A reply by Tibo and Tibo is one of the leads on the codeex team at OpenAI. a very influential person in artificial intelligence who also just happened to be on Jeff Dean's team at Google DeepMind. I was part of that team basically chat GPT one year before it came out called LM Chat and then another code name.
¶7He's saying they had a Chat GPT product one year before Chat GPT came out. And if you don't remember what it was like when Chad GPT came out, it almost instantly changed the world. It was so big and so sudden. So they had that. Don't forget Google had it.
¶8And so he goes on, Google was too nervous to release it and DeepMind was blocked from shipping products that could disrupt Google. I think about this a lot. Okay, so much to unpack here. Google was too nervous to release it. So they were nervous probably for a few reasons.
¶9He details one of them. DeepMind was blocked from shipping products that could disrupt Google. Google is probably the best business ever created. You go, you search, it serves you ads and they generate so much revenue from that and the profit margins are so massive. It is this enormous cash cow that they have been milking for almost two decades now.
¶10And so anything that could threaten that [clears throat] cash cow is a threat. And even if they are developing it themselves internally, it's a threat and it got shut down. That is what Tibo's saying. And so I think another reason why they were nervous is simply because AI by nature is non-deterministic, meaning they can't fully control it. They can't prevent it from saying things that would reflect badly on Google.
¶11And they were very nervous about that. But the main thing to keep in mind here is all of this is called the innovators dilemma. And if you want to stay uptodate on the latest in AI, check out our newsletter forwardfuture.com link down below. So let me read kind of what this innovator's dilemma is. So well-managed companies like Google, those that listen astutely to customers, invest aggressively in new technologies, and allocate resources systematically can still lose market dominance.
¶12The paradox or innovator's dilemma is that the very decisions deemed logical and competent for near-term success are often the reasons leading to failure when confronted with disrupted technological change. So basically to kind of put it in the context of Google, they have this cash cow. It's going really really well and then they have this new technology large language models that is coming and it just seems like an alternative but it doesn't seem maybe as good and why would they disrupt their cash cow? Their customers want search results. Great.
¶13So give them more search results. But it turns out that very logical decision was actually going to hurt them in the long run. So we have Google's business here and then all of a sudden we have large language models and the large language models start to grow. Google fails to identify and adopt AI effectively and that's when we can start to see the downfall the collapse of the old business model which Google is definitely going through right now. their search results.
¶14Although yes, it's still strong, everybody I know is going directly to either Chai GPT or Claude and yes, even Gemini, but it's all AI. They're traditional 10 blue links and advertising around that is being disrupted right now. And so now let's bring it back to Demis and Jeff Dean. Let's talk about why these different factors may have caused them to want to either leave or step down and focus on other things. And if we read Deis' tweet, we can kind of get a glimpse of it.
¶15Remember, the innovator's dilemma is because of short-term thinking, not because of bad decision-making, but just because they're thinking about the wrong thing. Listen to this. I'm stepping into a new role as chair of Google DeepMind and chief scientist of Alphabet. Now, here's the important part. This will allow me to focus on long-term strategy, accelerating scientific breakthroughs, including leaning into my own work at Isomorphic to help cure disease.
¶16He wants to think long term. He doesn't want to build the Google DeepMind company anymore because you are basically beholdened to quarterly earnings and shareholders. He wants to be able to think long term because again he identified the innovator's dilemma. He identified that ultimately Google's cash cow is under siege right now and he doesn't want to try to manage that transition. He wants to go to the frontier.
¶17He wants to build the future. And so that's I think a big part of why he decided to step down. Then that brings us to Jeff Dean. Jeff Dean didn't just step down. He's actually leaving Google, which is a big big deal.
¶18And I kind of think how this played out was Jeff was just seeing other companies building the future, Anthropic, Open AI, and he just wasn't feeling like he could accomplish that within the walls of Google, which, you know, if you think about how important Jeff Dean is to Google, they probably gave him anything he wanted. They probably said, "You can have unlimited budget. You can have unlimited compute. You can hire or take anybody you want to go build this. We won't even tell you what to do.
¶19There will be no management." And even with all of that, he did not think it possible to build his vision to get excited and build the thing he wanted to build within the walls of Google. And that is extremely telling to the current culture at Google. And so he went off and now he's building his new company, Discovery Loop. Automating discovery to accelerate science and engineering for the world. And by the way, if I didn't make it clear how important Jeff Dean is and three other people from Google who went with him, look at the list of incredible accomplishments that they achieved while at Google.
¶20They built Google search, that little business, Google ads, content ads, Gmail, news, translate, Gemini, cloud TPUs, a piece of infrastructure called map reduce, which is one of the fundamental pieces of architecture inside Google that allows it to do distributed systems, meaning it can serve billions of people throughout the world and serve all of them in milliseconds. That's how it works is this map produced system. and everything else you're seeing here, including Alpha Star, Alpha Fold, which was the protein folding AI that they built. So, just an incredible team, incredible accomplishments, and they needed to leave Google to actually do it. But I actually thought it was pretty funny that this is from their pitch deck.
¶21And to think that Jeff Dean and team actually needed a pitch deck, which they definitely didn't, but they went through the exercise of creating it anyways, I thought it was hilarious. And uh it kind of just shows the humble nature of Jeff Dean. Now I think a lot of the turmoil within Google and the trouble that they're having now started from a few different things. One again it is very much the innovator's dilemma. They have this massive business that they don't want disrupted.
¶22Tibo said it like the Google DeepMind team was blocked from releasing their language model product simply because they thought it might disrupt Google search. I mean that is the literal definition of the innovator's dilemma. And also, I think a lot of it is middle management. A lot of people in there who just don't want their jobs threatened, don't want change. They like the way things are, and you know, they're they're not necessarily on the frontier of scientific research, and that's what's most excited to them.
¶23What's most exciting to them is bringing home the paycheck. And so, I think a lot of these middle managers may have caused some of these cultural issues at Google as well. Now, this is all speculation by me, just to be clear. But here's the thing. I'm actually still quite optimistic about Google.
¶24They lost a few of their best, but what has happened historically when people have left companies that I've either worked for or that I led myself is the team gets stronger. They get closer. You can always hire more great people. And obviously, Google has so many incredibly talented people, including friend of the show, Logan Kilpatrick. So, uh, I'm not really worried about Google.
¶25And there's a bunch of other reasons. I still think they're well positioned. So number one, they have all the data in the world. All of that training data will be used to train insanely good models and much of it is proprietary training data. No other company in the world has it.
¶26So they have it. So that's number one. They also have their own chips in the TPU, the tensor processing unit, and they're on like the eighth generation of it. So they're well experienced with building their own silicon. They're not necessarily battling every other company to get the latest Nvidia GPU, although they do serve with Nvidia, but they also have their own TPUs, and that is a strong competitive advantage.
¶27They also have incredible hardware with Android. They power much of the mobile phones in the world, and they also have a cash cow business. They have a ton of free cash, and all of that, they can just continue to make massive bets. So, I'm really not all that worried. Now, if I were Google, I would not continue pretending to be competitive at the absolute frontier with Fable and Soul.
¶28You just have to admit it, okay? They're not there, and that's okay. And if I were them, I would actually go so hard on open source. I would try to build the absolute best open-source model on the planet. Get people building on top of your architecture.
¶29get people building on top of your AI and then iterate from there. And I think this is a really strong strategy. We're already seeing it with AI companies out of China. And there are really only two US companies that can afford to have a very strong open-source strategy. Nvidia, which already has it.
¶30They're investing over $20 billion into open source and their Neatron family of models. And Google, they will benefit when open source proliferates. and they already have the Gemma family of models, although those are more optimized to be small models for your mobile phone. But if they went allin on open source and everybody started serving open source models and then they continue to sell their TPUs, then they win no matter what. Everybody will want to power the open source models with the hardware that powers them the best, the most efficiently, and they already have a strong competitive advantage with the TPU.
¶31So, if they just released all of these incredible models for free and then served them and sold the TPUs as well, I think that would be an incredibly strong strategy. So, I still am very hopeful for Google. I am certainly far from counting them out and I don't think you should either. Open source absolutely can be a winning strategy for Google. I actually detailed how important open source is in this video right