Don't expect Amazon to be a major player in AI model wars in the next 3-5 years, says Doug Clinton
We’re seeing very significant momentum and people trying to figure out how to run their generative AI on top of AWS. You know I mentioned we have a multi billion dollar revenue run rate that we see in AI already and it’s still relatively early days. That was Amazon CEO Andy Jassy yesterday on the earnings call, joining us now with reaction to Amazon’s report. Doug Clinton, Deepwater Asset Management managing Partner and Co founder of these numbers in 1/4 are are mind boggling. I think 143 billion in sales and a lot of that, I mean that’s a lot of books. They’re doing more than selling books if I got that right now, Doug, they’ve they’ve branched out a little bit, but a lot of it is a is a resurgence at at AWS it is and Joe they do do more than just books now. 25 billion of the 100 plus that you talked about came from advertising, 25 billion came from AWS. So they have a really diverse business now that is driving incredible profit, 10% operating margins, we haven’t seen those operating margins at Amazon for a while. So I think the quarter was really strong. You look at the guide, it was conservative. Amazon tends to be conservative. I’m not surprised that the stock is, let’s call it, close to flat now. Wouldn’t have even been surprised if the stock was down a little bit given the report, because I think the big headline to me was just like meta, just like Alphabet and just like Microsoft. Amazon talked about a big CapEx build coming up for 2024 as they invest into this AI opportunity that all the big hyperscalers see. Let’s see that the, the AI part of the business is eventually rivaling the businesses you just talked about, but it’s going to cost money to get there. And were they behind initially, I think they were a little bit behind relative to Google and Microsoft in particular. Obviously those are the two direct cloud competitors that they are up against. Microsoft has Open AI, Google has their own Gemini model and Amazon I think is they’ve sort of been figuring out how do we fit into this ecosystem without really having a meaningful, they do have their own large language models, but without having a real meaningful competitor to the Open AI and Google products. But I think they’ve done a good job with Bedrock, with Sage Maker. Those are two of their core products that they offer now in terms of AI services and managed services to companies that are building models on AWS and then hosting those AI applications on AWS. I think they see just as big an opportunity in AI going forward as Microsoft and Google do. And so they’re going to spend $60 billion likely given their commentary on the call on CapEx this year, even if it wasn’t in house AI development, they they sell to other AI companies. I mean, AWS is a place where you go for your AI capabilities. It is you can access a number of models from third parties that Amazon has partnered with. So Anthropic is probably the headline model that they talk about. They referenced it in their earnings note. They talked about Anthropic on the call and Amazon is an investor in Anthropic as well. But you can also access Meta’s Llama three, you can access Mistral, which is another open source model. So they have kind of a a wide array of a large menu of different models that you can access through AWS. So which begs a different question. So they’re working. I don’t think it’s been announced. I tried to ask Andy a little bit about it when I when I saw him during that interview that we did about a month ago now about Olympus, which is their their effort to to create a large language model that is akin to or competitive with ChatGPT or Claude or the others. Given the success that they’re having and given the way you just framed it, this is a place that people are going to come for AI. Do you think that they actually need to be investing and owning their own large language model and do all these models somehow ultimately become commoditized? I think what ends up happening in the model wars is we probably have three to five major models that survive and I would actually put Amazon on the outside of being one of those three to five. The issue that I think Amazon will ultimately have is if you think about what enables a company to create a great model, I think it’s three things, it’s data, it’s infrastructure and then it’s distribution. And so Amazon has certainly the infrastructure with AWS, they probably have some good distribution given how many customers come and use Amazon’s products. I just don’t think they have the same level of data that Google has, that Meta has or even that Open AI has to create a model that’s on par with those three. So if I’m Amazon, am I going to continue to invest in a model? Probably because I think everyone feels like they need to be experimenting here, even if just to understand the market, how these models function. But I wouldn’t expect them to be a major player in the model wars if I look out three to five years now as an investor.