Graham: GOOGL & AVGO Strongest Companies in Hyperscaler & AI Chip Gap

요약 보기

AI 버블 우려는 아직 초기이며 하이퍼스케일러와 반도체 간 성과 격차가 핵심. 구글·아마존·MS·메타는 저평가되어 투자 기회로 제시. 브로드컴은 커스텀 ASIC·네트워킹 실리콘 성장으로 매력적이라 평가, 사이버보안(CrowdStrike, Palo Alto, Tenable)도 언급. 토큰 가격 하락 전망과 S&P500 연말 목표 7840 유지. 소비심리 회복 기대, 최근 15% 조정은 재투자 기회.

다듬은 글

※ 자동 자막 원문(다듬기 전)

01. Now with me is Andrew Graham, Managing Partner of Jackson Square Capital . We are very happy to have you with us. When it comes to the macro context, I'm thinking about the big picture that we're looking at . What is the most talked about thing, are we going to see fatigue ? Are we going to see fatigue in AI ?

02. What are your thoughts ? No. It's been three and a half years since Chat GPT started , and we've been observing the bubble for almost two years . This is still in its early stages. Not many people talk about it , but the performance gap between hyperscalers that spend and semiconductors that receive money is troubling everyone.

03. That difference in performance is the root cause of this problem. If that difference narrows, I think we can put bubble fears aside for now . Let us discuss this in depth , because we have already discussed the differences between hyperscalars and semiconductors . Who are we looking at when we think of hyperscalers and what is your perspective on that group?

04. In my preference, I would mention Google, Amazon, Microsoft, and Meta. Okay. Yes. We want companies that own the infrastructure and software . Everyone has them , but these seem to have the best connectivity offerings. They also own the distribution and customer relations. That's how we created that ranking .

05. I think they have potential as token prices are falling . Token prices are stable. They have the potential to show improved returns on their invested capital in the second quarter. So I believe that that capex will ultimately yield good results . Yes. And they really stepped back . Looking at the PE ratios, we can say that they are now at a lower price compared to the beginning of the year.

06. They are definitely cheaper now compared to the last three years . So I think this is a good investment opportunity . Yes. You make a good point about three years . Yes. Tell me. You have a positive attitude about this , right? Yes. Yes. Now if we turn our attention to semiconductors, how is it different ?

07. What names are you considering there ? So my first preference in semiconductors is Broadcom. Broadcom may not be such an easy target, but I think everyone has overlooked it a bit . So, in my view, the era of custom ASIC silicon has begun, and that's because the major frontier labs in America are focused on making money at the end of the fourth quarter .

08. Up until then, everything was training , and then we moved into the inferencing phase. Inferencing requires two to three times more computing and three to four times more memory than training. So, the main purpose is to reduce the cost as much as possible, and custom ASICs do the same thing. These can operate 30 to 40 percent less in computing terms and nearly 50 percent less in power consumption.

09. We've heard a lot about the Broadcom and Apple partnership this week . What is your opinion on that? That is definitely a positive sign for market growth. I think we've overlooked the growth of networking silicon because we've focused so much on memory, and we're seeing similar growth in this as well, along with memory growth.

10. In this regard, it is encouraging that it is changing from 800G to 1.6 T. This is part of the scale up and scale out architectures happening in data centers. This is a great opportunity for Broadcom, which has a 70 percent share of the switching silicon business, which could achieve three- fold revenue growth this year compared to last year .

11. People are aware of Jackson Square Capital's top 10 holdings, which are Nvidia, Apple, Broadcom, Google, Eli Lilly, Amazon, Microsoft, KLAC, Caterpillar, and Coherent. I think so. When you look at what's happening in China, you said some of the best ideas are open-source Chinese models that are closing the gap in vulnerability discovery capabilities with American frontier models .

12. Explain it to me. Yes, they are definitely closing that gap . According to the Brookings Institution, they are only six to nine months behind the best models from America's Frontier Labs ; So, after the chaos surrounding the best model, like the Claude Mythos, I don't think American companies will compromise on spending on cybersecurity.

13. So in our order of preference is CrowdStrike, then Palo Alto Networks , and then we recently added Tenable One , which is in the vulnerability discovery and management category. I think when you talk about token costs going down or trending lower, but prices staying the same, is n't that a good mix?

14. Yes. That may change. Yes, that is, eventually. Because right now it's so good, it's kind of perfect . Yes, that's right. If we follow Moore's Law on token costs , we certainly can. Although acquiring compute capacity is expensive, I think token costs will continue to decline . Memory prices are high and there are other things , but they are getting more usage out of it .

15. Like the point I made about the custom ASICs that Broadcom is supplying. So I think token costs will fall faster than token prices . If they want to encourage enterprises to start building their AI applications , they will have to lower token prices . We haven't reached that stage yet. Maybe it will take another year, but it will happen eventually.

16. 7840 is your end-of-year goal, right? I still stick to it . And, you really liked the macro background. I mean, you said it's still supported by record levels of consumer spending and ongoing stimulus from that one big beautiful bill. Give some examples of how that fits into your thesis . Consumers are very strong and I think there is opportunity there , but I haven't seen it in the consumer discretionary spending segment yet.

17. You saw a two standard deviation low in consumer sentiment last month, right? I don't think it will fall off yet, and it won't get any worse . So, sentiment is only likely to increase, especially if gasoline prices fall. I think there is an opportunity to have some consumer discretionary stocks in apparel or other categories .

18. But I think where we need exposure is still in the tech space, networking equipment and semiconductors , especially in those that have been overlooked recently. So the 15% pullback we got is an opportunity to reinvest. Yes. And I also understand that you expect operating cash flow and capex to improve .

19. That is, for the most part it remains bullish. It's great to meet you. Thank you very much. Andrew Graham, Managing Partner, Jackson Square Capital. Thank you.