Aziz Hamzaogullari on Oracle, OpenAI, and AI Infrastructure

JOHN KAVOLIUS: Thinking about a company that impacts a lot of the portfolio holdings in your strategy, OpenAI is a company that has enormous market expectations, right? 

There's talk about them potentially IPOing this year, but they're a large buyer of some of your biggest positions. How do you and the team kind of model out their ability to make good on capital commitments? And I know the company that's been in the news most about is Oracle, right? And so how it's impacted there. 

AZIZ HAMZAOGULLARI: So first of all, Oracle we have held this name for almost 20 years now. And during the time, there were many times that the demise of the company was called by some analysts and all that-- there was a time that people thought that ERP software is going to displace them, cloud computing were going to displace them. 

So when you look at Oracle, their core business is database and application software, and they basically sell these products. And the decision makers make a 10 to 15 year decision, because once you install these databases or other software, it's very high switching cost. 

So meaning, if you were to try to switch from Oracle to another vendor, it would be very, very painful for your company. Everyone on this call, most of their company is using Oracle Database and all your client data is residing there. All mission critical data is residing there. 

And in addition to that, they have a very strong sales force. They have 30,000 plus sales and marketing people who have what I call account control. They know the needs of the account before the need appears, because they're in constantly talking to the company and understanding what their IT needs are. 

They spent over $75 billion over the last decade for R&D, and they have a very great-- good business, profitable business model-- high margin, high return, mid 30s operating margins. And like 50% of our companies, it's led by their founder, Larry Ellison, who's a very visionary leader, who manages the company with long term focus. 

And he has a stake-- 40% stake in the company. So he's very much aligned with us in terms of the company. Now, Larry Ellison is a visionary, and he did a couple of big, very long-term decisions in the past 30, 40 years. First, was the architecture of the database that he came up with. 

People caught up to it only 10 plus years later, like IBM and Microsoft. Second, his recognition that he can increase the stack by adding application software in addition to database. That was another major decision. And now this is, I would say, the third big decision he's making, which is building Oracle infrastructure for OpenAI, for Anthropic, and all other AI vendors. 

And there's a big, big worry right now, will there be returns on this and on-- so first of all, in terms of valuation, the company is lower than before OpenAI announcements, so just-- that's very important. Second of all, we believe you're buying all the AI optionality for free, because if you look at the core business, which is the database and applications which I just talked about, with 95 plus percent retention rate-- Client retention rate-- this predictable good cash flow, you're buying that business without any benefit of-- at a discount, without the benefit of the AI. Within AI, I think both Microsoft and Oracle have high exposure. Microsoft has 20 plus percent ownership of OpenAI. 

In addition to that, if you look at the backlog of Oracle, roughly 50% of that backlog is directly linked to OpenAI. And for Microsoft, it is 30 plus percent, and for Google, around 15%. Let's take the worst case. Let's say OpenAI doesn't live up to the expectations that everyone is looking to. 

This capacity that Oracle is building is, first of all, fungible, so meaning it can be used by other vendors. Second of all, the way you build these data centers is you first build the walls and bricks and mortar, and the last stage is where you spend the most money. 65% of the spending is chips, which is NVIDIA and others. 

If you see a slowdown in OpenAI needs and you cannot-- you find out you cannot sell this capacity to anyone else, you will simply not open the additional data centers. Each data center is around $50 billion. So our view is that the market currently definitely is not giving any credit for OpenAI or any other AI projects. 

And we also believe that market is obsessed with the near term cash flow implications. We have gone through this with Microsoft, with Amazon, with Google, with Meta. What happens is these companies go through investment cycles, and every time people get worried, the key question is, will you get return on these investments or not? 

And you determine that by looking at what these projects are used for. And we believe that OpenAI is a real company. They have significant growth. But even if that doesn't happen, as I mentioned at the outset, it's not just large language models. With AI, you have other drivers, like humanoids, robotaxis, government spending. 

So I think the demand will be significant. It's not going to be straight up. So we believe that Oracle is very well positioned and we believe it's very mispriced, but this is part of investing. Again, 2022, people were questioning Meta when we were buying more of it, and it was the worst performing stock in our portfolio. 

Google, just last year, people were thinking that Anthropic will come and take Google over, and now they're realizing, not that fast. It's a great company. Visa was the worst performing company in 2010 for us because of all the legislative risks and all that. So you have to take a step back and understand the structural drivers and also quantify the risk. 

And if you're getting paid for that risk, then it's an opportunity.