New Deals Show AI Compute Spend Keeps Growing

Bloomberg

Watch New Deals Show AI Compute Spend Keeps Growing - Bloomberg on Dailymotion

Transcript

00:00Your top holding is NVIDIA and all Parsley back, but Microsoft's a big one, Broadcom, Alphabet,

00:05really all of the big winners. Is this playing out as you anticipated?

00:10Yeah, well, I think that, you know, AI is a huge productivity driver, in my opinion,

00:15and I think that these companies are really well positioned to essentially capture that

00:20and be the platforms for, I think, growth in the economy. And, you know, what I'm looking for

00:25going forward is, like, you know, the new use cases really come through, like agentic AI,

00:31for example. I think it's, to fill this capacity, it can't just be you and me using Chacham ET.

00:36It's got to be, you know, thousands of agents for each company, like, making decisions and executing

00:42and unlocking, I think, labor, essentially, is what I'm looking for, and that's what's exciting

00:47going forward. It's exciting when you see NVIDIA's name referenced in about three of the compute

00:53deals just on this Monday alone. But take us to perhaps the one that you don't own. Interestingly,

00:58you don't have exposure to Amazon and AWS, but it signs its deal with OpenAI. Has there been

01:02methodology around that, about its own cloud prowess, or perhaps worries about the lack of

01:07growth that we'd seen in previous quarters, bar from the previous numbers they just reported Friday?

01:13Hey, well, I think that they had a nice earnings report, and I do think that there seems to be a lot

01:19more momentum now. You know, you talk about the deal that they just did today, and just

01:24essentially, like, rolling out a lot more compute and accelerating AWS, I think, is really good for

01:30the business. I think they're well-positioned to essentially leverage AI in their e-commerce

01:36business, as well as be an infrastructure provider. So, I don't think you can count them

01:40out at all, and, you know, it's good to see them, you know, outperforming.

01:44What's interesting is someone who got beaten up is in the portfolio, which is Meta. And I'm

01:49interested in on your take of how we're deciding who we reward of capital expenditure and who we

01:54don't. Yeah, well, I think that, you know, AI is a long-term investment for Meta. And I think that

02:02the market is probably digesting the, you know, heightened level of CapEx and the depreciation

02:09that's coming through the P&L. But I think that's a little bit too short-term to view it. And you

02:16look past that, you know, the returns are there, and these are going to be great investments.

02:21And, you know, Meta definitely has the digital formats to leverage, you know, AI and a tremendous

02:28scale across customers. And so, I think it's exciting what they're doing in the SMB segment,

02:34as well, if it unlocks a lot of ability for, you know, companies to essentially, like, reach

02:39customers with just a credit card and asking for the ROI from ad campaigns.

02:45I just want to go to what the president of Microsoft was telling our network a little bit earlier

02:49today about the risks of under-investing, really, here. We spoke with Brad Smith. Just take a listen,

02:55Tony.

02:56Our biggest challenge is not a risk of getting ahead of demand. It's actually keeping pace with

03:02demand. And what we are finding is that our customers, you know, whether they be enterprises

03:08across the economy or governments or non-profits, they are looking to us to build out this capacity.

03:15They want to use this compute. So, for us, if we build it, it will be put to use. As it's put to

03:23use, there will be a return to our shareholders. You, of course, have exposure to Microsoft, Tony.

03:29Brad really talking about how they're expanding more geographically as well and investing over

03:34in the Middle East, not just taking money from the Middle East. And I'm interested in what you

03:38think about Microsoft's need to depend on neoclouds. We saw them strike a deal with an Australian

03:42neocloud provider for Texas data centers. Is that something you like to see, that use of their money

03:49and capital expenditure on neocloud as well as their own?

03:52Yeah, I think it allows them to be flexible and nimble. I think that standing up these data

03:57centers is no small feat. And a lot of times it requires various players. It requires the data

04:03center capacity, the power. And so I think it's encouraging to see Microsoft collaborate with the

04:09ecosystem and plug in partners. And so to me, I think it makes sense. I think it's also a testament

04:16to the demand signals they're seeing and kind of the workloads really take off. And, you know,

04:22I think AI is really hard to do. Not every company that is trying to do it is successful. But I think

04:27the ones that are finding success are, you know, seeing tremendous promise. And so I think that's

04:31what essentially gives people confidence to invest here. Confidence to keep adding when we've seen

04:38runaway performance, not just of NVIDIA at $5 trillion, but another one of your holdings,

04:43Micron, for example. I mean, significant outperformance. When do you decide that

04:47you want to cash in, profit take? Yeah, we're all constantly managing position sizes in the portfolio.

04:54But, you know, in my opinion, I think that you want to be invested in the areas where there is

05:01essentially a bottleneck. There's scarcity and that performance depends on it. So I think that

05:07HBM, memory, you know, HDDs and now, you know, NAND, there's becoming more of a shortage. And so

05:15to me, the industry is getting better and it's a critical component. So, you know, from my perspective,

05:23there is opportunity for these companies to compound value over time. You are, of course,

05:28equity focused. But when we have Alphabet selling debt in Europe and the U.S., any of that financing

05:35of this AI spend give you pause? Yeah, well, I think it's novel. But, you know, you look at the

05:44cash generation of a lot of these companies, they're really blue chip. And, you know, to me,

05:49I think that this investment is smart. And to do some appropriate level of debt financing, I think,

05:55can make sense. In addition, I think that the infrastructure, you know, there's a lot of debate

06:01of, like, the useful life of this. I know that NVIDIA comes out with a new generation every year. But

06:06I think that these are longer-lived assets than people appreciate, potentially. And they're not

06:12just, like, three years obsolescence, but probably more like five or six. And so, you know, in terms

06:18of the appetite to lend against it, I think that in three or four years, it's going to be more

06:23commonly accepted. And I think we're probably on the path there. Because these are really, like,

06:27I think, you know, revenue-generating assets. And there's a lot of productivity coming off of it.

06:33And, Tony, that productivity, as you started the conversation, is about how we use, deploy

06:37generative AI. And Palantir is one that really has. But how do they fight also the macro headwinds,

06:43the government shutdown? Do you have any concerns at the valuations that some of these companies trade

06:47at? Hey, absolutely. I mean, these are, you know, very elevated multiples. And so, they need to keep

06:55delivering the growth here. You know, Palantir is definitely, I think, at the tip of the spear in terms

07:00of implementing AI. I mean, they started doing that in the government. And now, the commercial

07:04business is growing rapidly. And so, you know, I think the government's shut down. Like, I think

07:10that's probably a temporary thing. They could get a pass for that just because the government's not

07:15going to be shut down forever. And, you know, you look forward. I think what really matters is that

07:21they're driving significant ROI for their customers. And they're able to do it in a way that connects

07:26their data. And I think that's, like, the change in IT services is, I think, that you need to come

07:33with a solution and the engineers to help your customers instead of implementing, like, a SaaS-based

07:39off-the-shelf solution. So, I think that's what's exciting. And, yeah, we'll see later on today.

Category

Transcript

Be the first to comment

Recommended