00:00Anthony, of course, I mentioned the NVIDIA deal, but we also have Intel's first public share offering since their public
00:06listing, what, in the 1970s or so.
00:09How does this really come through into the Asia session?
00:14Yeah, let's go through the numbers on Intel really quickly.
00:17So it can be upsized to around $20 billion.
00:20That's about two times the normal ADV of the stock, which has been much higher in recent days.
00:25Whereas the pricing is around 6.5% discount, which compares kind of favorably with a 4% One Sigma
00:33move in the stock.
00:34So the deal is tight and it is very well oversubscribed, but it is a big deal.
00:39$20 billion to absorb at this stage in the AI cycle is a lot of money, and it will be
00:45interesting to see how the stock trades aftermarket.
00:48It was weaker into the announcement of the deal.
00:50Now, going back to NVIDIA, what is interesting here, and that ties in all the concerns that you mentioned, is
00:57concerns about capital discipline.
00:59NVIDIA becoming the banker of choice to the AI ecosystem is not what shareholders signed up for with that stock.
01:06And it's very similar to the Korean situation, where shareholder return is becoming a much bigger facet of go-forward
01:13expectations from retail.
01:14And as a result, you're seeing this weakness early in Nextrade, in Samsung and Hynix, as it appears that the
01:21AI ecosystem remains extremely committed to high levels of capex versus shareholder return.
01:27So that debate is really strong on the hardware side, and it's coming at a time where the software side
01:34of the AI trade is doing really well.
01:35So, for example, again, last night, you saw the software space in the U.S. trade really well.
01:40You're seeing HS Tech in China trade really well, Tencent, Alibaba, and other kind of China hyperscalers.
01:46So the focus has moved from hardware to software because of the capex concerns and the spending concerns around the
01:54hardware side.
01:55And that concern will not be alleviated by the capital raising you're seeing today.
01:59At a time when broader investors are questioning just a sheer amount of spending that's gone into AI, we really
02:06see these big players doubling down.
02:11We really do.
02:12And, you know, honestly, I was hoping for a little more detail than we got after the initial report on
02:19this $500 billion investment plan.
02:22Even after the NVIDIA announcement this afternoon, we still don't know what the timing of these investments might be, what
02:31the structure will be for the financings, and how much of the plan goes beyond the string of AI deals
02:37that NVIDIA and the Wall Street firms have already made.
02:41I think that, as you hinted at, this is going to continue to drive concerns about the so-called circular
02:49AI deals that NVIDIA has been linked to over the past several months.
02:53You got to wonder, Avril, is Compute the new asset class?
02:58Yeah, making it an investable product, right?
03:01And worth noting as well that this is going to be third party, so it's not going to be sitting
03:06on NVIDIA's balance sheet.
03:08Aside from that, just look at the names that are in the mix, right?
03:11All the big private market firms, BlackRock, Blackstone, KKR Polo, even Goldman Sachs in the mix.
03:19And as we've heard from Jensen Huang, you know, he pretty much went to them with the idea and nobody
03:23said no.
03:24So very interesting in terms of what this means for the AI infrastructure build-out.
03:29Of course, as we highlighted earlier, this is coming as Intel is looking at a share sale as well, could
03:34be to the tune of $15 billion.
03:36But it's also looking at something towards $20 billion, sources say.
03:40And this would be the first time, this first share sale since the early 1970s.
03:45So quite something as well.
03:46And of course, it's not just NVIDIA, not just Intel.
03:49We're also watching the likes of Anthropik, a $9 billion deal with the likes of Macquarie as well as GIC.
03:56So just this idea of how much financing is required for the AI infrastructure build-out, Yvonne.
04:04Yeah, and it continues to really add on, right?
04:06When you take a look at what TSMC numbers suggest as well, we're talking about 45% rise in terms
04:11of monthly sales.
04:12Does in some way signal this AI demand is sustainable in some ways?
04:17You square that with, of course, this financing and all the headlines we've been seeing across some of these deals.
04:22It certainly leads to a lot more focus back on the attention of where to next in this whole AI
04:27trade.
04:28Let's bring in Felix Wang now, Managing Director of Global Technology at Hedgeye Risk Management.
04:33Felix, it's great to have you back on the program.
04:35Let's start with the NVIDIA news here.
04:37I mean, $500 billion certainly is eye-popping.
04:41I'm just wondering what went through your mind when you saw that headline.
04:45Yeah, it is a huge number.
04:47Thanks for having me back.
04:49You know, there's a reason why Jensen wears a shiny, expensive leather jacket.
04:54He's able to see far ahead and exert control over the AI game and competition.
05:00And right now, a new part of the game is about financing, lowering the cost of capital, lowering the financing
05:07constraints for their customers.
05:08That could be another structural advantage for NVIDIA.
05:12So, the way I'm thinking about it is this.
05:14If NVIDIA and AMD both have similar compute and TCO economics, but NVIDIA's customer wants, say, for example, $10 billion
05:23of compute, and they only have $1 billion of EBITDA,
05:26so that means they have to borrow.
05:28They have to borrow billions of dollars of financing at an attractive rate, and NVIDIA can facilitate that lending.
05:35So, to me, that's an advantage.
05:37In effect, they made NVIDIA's product cheaper without really cutting GPU prices.
05:43Again, it's about controlling of the game, extending their mode.
05:47But I also see this financing layer, if you will, as a double-edged sword for NVIDIA because, yes, it
05:55can absolutely extend the hyperscaler investment runway by allowing them to borrow at attractive rates.
06:02But it also makes future demand kind of more sensitive to credit conditions, credit volatility, and raises a lot of
06:11questions on what we consider to be real demand.
06:13So, this is why, you know, NVIDIA, if you look at their five-year CDS, it jumped today, the most
06:20in two weeks.
06:21Financing can prove forward demand and raise the risk of overbuilding.
06:25So, it is a double-edged sword here.
06:27I just want to pick up on some of those AI themes and that story we mentioned a moment ago
06:31about NVIDIA tapping U.S. investment giants.
06:34I mean, Jensen Huang said he approached six of the investment giants.
06:37They all said yes, $500 billion in financing for AI infrastructure with compute power acting as the collateral.
06:45Now, I know you say AI is not necessarily a bubble, but it sounds like you've got a few caveats.
06:53Well, yeah.
06:54I mean, AI isn't necessarily a bubble, but the market needs an earnings reality check.
06:59And I won't, you know, so I won't say that we're in a bubble for that, but because AI investment
07:06is just enormous, I mean, we can all agree on that.
07:09The economic benefits are beginning to show up.
07:13Productivity gains could be substantial.
07:16We expect that they will be, but investors have already priced in a lot of that future success.
07:21So if the market continues higher from here, which we do expect it to, I will say, it would be
07:26because of an earnings-led melt-up versus a valuation-led melt-up.
07:30So if earnings deliver, we're very positive on technology.
07:35In fact, we're not bearish on technology at all.
07:37We're very bullish on the earnings power of AI.
07:41Where we exercise some caution is the price that investors pay for that growth.
07:47So the technology may be right, but the stock can still be wrong.
07:50So with valuations and expectations really high, companies need to continue to deliver really strong earnings growth to justify those
08:00valuations.
08:01So you need to look at technology and all the different ways that you can play AI in order to
08:07make sure that you're not overpaying and you don't get hit, but you also can be productive with your investment.
Comments