博通是一个被低估的投资机会，得益于强劲的AI需求和预期的营收增长，尽管存在客户集中风险。

The first name I will put forward is Broadcom, and the stock symbol is AVGO. Broadcom announced its earnings earlier this week. I think it was announced on Wednesday and the numbers were fantastical. Revenue reached $29.6 billion for the quarter, an increase of 86 % year-over-year. With AI semiconductor revenues now accounting for 56% of total revenues, this sector alone has grown by 221%. As for next year's projections, they raised them to $34.8 billion . This represents a 93% year- on-year increase in total revenue, and the AI chips segment is expected to rise by another 236% year-on-year.

So, the request already exists. They work with lots and lots of great players. For those who are unaware, Broadcom helps other big players design their own AI chips. They have large customers like Google with Tensor Processing Units (TPUs).

Currently, the TPU V7 is being shipped in large quantities. Shipments of the V8 product began in the third quarter, with expectations of reaching significant volume in the next quarter. Meta is expected to begin shipping its products in the next quarter. The " Halabino" chip, a promising artificial intelligence chip, has also reached the final design stage, the step that precedes mass production.

So, they have big customers, and Anthropic buys a lot of TPUs from Broadcom and Google, and there is a lot of enthusiasm for this trend. Now, the stock has experienced some decline. I mentioned that the stock was unfortunately negatively affected. The last time we recorded, and as we record now, the stock price is in the middle of three hundred, that is, approximately $350 or $340.

The market is wary of a few things here. One of these things is that Lip-Bo Tan mentioned that a large part of their demand comes from major artificial intelligence companies. You are talking about "Open AI" and "Anthropic". And I think now, Rachel, that the market is not very happy to have this type of customer as a major part of the order book.

Isn't that so? What would happen if OpenAI failed? What if " Anthropic" fails? I believe that if either of them fails, another big player will step up to meet that demand for computing power. I simply believe that computing will pass into other hands, and to whoever offers the best artificial intelligence technologies. So, I'm not worried about that, but I understand why the market is worried.

The second thing, which I found interesting before passing the conversation on to you, Rachel, is that Broadcom mentioned that one of the factors limiting its growth is its supply chain. And part of that supply chain is the laser. The demand for lasers far exceeds the supply . The other issues are land, energy, and infrastructure. When building multi- gigawatt data centers, it is essential to ensure the availability of land and power to operate graphics processing units or AI computing technologies. Because the last thing you want is what I consider "broken computing"—computing that actually exists but is not working. Therefore, they stated that one of the major issues for some future growth in the coming years is : Can these chips actually be operated? The concern is that if it cannot be activated, will this lead to an oversupply of inventory for many of these players, and will this slow down or reinforce the "AI bubble" hypothesis that some are talking about ?

But regardless, I think Broadcom represents a great opportunity right now . They have many excellent steps. But the risks always remain. Competition in the artificial intelligence market and supply constraints exist, but demand is present and clearly seen in the triple- digit year-on-year growth rates.

One thing that really caught my attention on the earnings call, and I think it caught the attention of many investors, is that Broadcom now has a clear vision, as they described it, of generating $115 billion in revenue from AI semiconductors in fiscal year 2027, and then $230 billion in fiscal year 2028. That is, it's effectively doubling and then doubling again.

Another thing, as you know, CEO Hock Tan said that Entropic is on track to overtake Google in 2027 to become Broadcom's largest customer for custom AI chips. Google will therefore move to second place there.

As I mentioned earlier, unlike Nvidia's general-purpose graphics processing units, Broadcom designs custom AI accelerators specifically for particular AI labs. It is specifically designed for major cloud computing service providers. So, when you look at Broadcom, it's more about investing in who their preferred design partners are in the long run, rather than just focusing on who makes the best chip. I believe this is an important distinction. But yes, these order targets, these custom chip orders, are indeed coming from a handful of cloud service providers and AI labs.

I believe that the concentration of customers you referred to clearly poses a real risk. But these major clients are also among the most financially secure companies on Earth . And I think you can see too , yes, you have a small number of customers, but these are the leaders in the AI revolution we are witnessing. These are the companies that are building the artificial intelligence infrastructure as we know it. Broadcom is really a direct beneficiary of that.

Note on valuation: Broadcom shares are currently trading about 30% below their 52- week high. I believe this represents an investment opportunity. I'm not necessarily saying it's trading at a low valuation, but I believe it represents an undervalued opportunity compared to its peers.

I think it has a lot of room for growth, and when you look at a company that expects to double its AI revenue in the next fiscal year and then double it again the following year. Its shares are trading far below their record levels. I think it's an interesting combination of risk and reward. I think this is a really well-run company . As you know, they are looking at incredibly large and growing markets for their AI semiconductor revenues. So, yes, as we enter the 21st century, will they be able to maintain this rate of growth ? I think that's still a matter of time , but they do n't have to maintain that rate for 5 or 10 years to become an exceptional company. Therefore, I think there is a lot to admire about this stock if you are looking to take advantage of these enabling factors that we are seeing.

Yes, of course. Just to add something quick before we move on to the second arrow, the $ 115 billion in revenue from artificial intelligence next year, they said they were conservative in their estimates, weren't they ? Actual demand is much higher, but they have to take into account many of the supply problems we talked about earlier.

Also, the other thing that I may be underestimating . The other thing is, as I mentioned, right? Regarding Anthropic and OpenAI, I would say it depends on who you talk to. This could be a positive or negative indicator. I can understand why the market might penalize the company; I didn't mind having Google as your first customer because I know Google is capable of paying. But the fact that Anthropic is the first client is perhaps the aspect that makes me feel somewhat hesitant about potential revenue growth.

Much like Broadcom, it announced its profits and the market punished it, and rightly so , wasn't it? I think that for Broadcom, for example, the reason for penalizing it is logical. I wouldn't be too pessimistic about the stock, but I can understand why the market is punishing it.

Unfortunately, as we saw with Lam Research, and as we saw with Broadcom, only a limited number of customers need and purchase the products offered by Credo and all these semiconductor companies.

You have competitors like Broadcom, Marvell, and Astera Labs, all of whom touch on a part or section of this business.

Many, but not all, of them are profitable and well-protected companies that are driving this demand, and they are the ones buying this capacity from companies like Credo, Broadcom and Lam. This creates a dynamic where there is a concentration of customers, but these are also the big players in the field. So, this is something I think you should be comfortable with if you're putting money into these types of businesses. I mean , you know, Broadcom benefits from the demand for more custom chips.