저자: 미상 · 외국인
요약 보기
브로드컴은 AI 맞춤형 반도체(ASIC)와 VMware 인수 기반 인프라 소프트웨어를 결합한 풀스택 AI 인프라 기업으로, FY25 매출 약 640억 달러(+24%), AI 반도체 매출 65억 달러(+74%), 영업이익률 50.9%가 근거로 제시됨. 고객 집중도(상위 5개 고객 42%)와 밸류에이션 부담이 주요 리스크로 언급됨. 4월 29일 실적 발표와 5~6월 투자자의 날, AI 고객사 대형 계약 루머 등이 주요 촉매로 거론됨.
다듬은 글
Welcome back to Investing Talk podcast. Today we're looking at Broadcom. As usual, before we start, subscribe, like, and comment to this video to support the show. And if there's a specific company you want us to look into, leave it in the comments below. So, let's get into the analysis. Broadcom is um it's a fascinating story right now.
It's really transformed itself over the last few years, especially with this huge AI infrastructure buildout. It really has. I mean, our goal today is to kind of cut through the noise, right? We see these spectacular numbers, but the real question is, is it sustainable? And you know, what are the big risks that could derail this momentum?
Exactly. And when you look at Broadcom today, you're really looking at a company that has built this uh twin engine growth story. The scale they've achieved is it's just massive. The numbers are almost hard to believe. I mean, what are we looking at for the full year? Well, the latest forecast for their fiscal 2025 consolidated revenue is projected to hit 64 billion.
64 billion. Yeah. And that's up a whopping 24% year-over-year. For a company that size, that's that's practically hyperrowth. And it's not slowing down. The recent numbers Q4 2025, that was a record $18 billion. And the guidance for the next quarter is even higher at $19.1 billion. It feels like it's accelerating.
It is accelerating. And that's the key signal, right? The conditions driving this are getting stronger, not weaker. And uh if we look inside that twin engine, you can really see where the excitement is. It's the AI chips has to be. It's the AI semiconductor space. Yeah. Specifically, their custom silicone.
We should be clear though, they're not just selling, you know, off-the-shelf parts, right? Their AI semiconductor revenue surged 74% year-over-year in the last quarter to $6.5 billion. And that's coming from custom AS6 application specific integrated circuits. Exactly. These are, you know, unique proprietary chips they're building for the biggest players out there, the hyperscalers, your Googles, your Amazons, your Metas.
They need this custom silicon to run their massive AI models efficiently. The thing that jumped out at me though was the ambition. It's one thing to ride the wave, but they're projecting what over $120 billion in AI revenue by 2030. It's a staggering target. A six-fold increase from where they are in fiscal 2025.
So, how do they get there? Volume. It's just sheer volume. The data shows that in just one quarter Q1 of fiscal 25, they shipped something like 20 million AI related AS6. So, this isn't a small bespoke business anymore. It's mass production of the uh essential plumbing for the world's biggest AI factories.
Okay. So, that's the high growth maybe high-risisk hardware engine. Yeah. Let's talk about the other engine, the one that provides the the stability, the software business, infrastructure software segment. Yeah. And it's absolutely the anchor here. They've built it up through some huge acquisitions like VMware
and that brought in what about $6.9 billion last quarter. It did. And you have to understand this isn't consumer software. This is missionritical stuff. Mainframe security, enterprise cloud platforms, the things huge companies cannot live without, which means it's predictable revenue. How do we uh how do we quantify that for people listening?
The key metric is the net dollar retention rate. In Q1, it was 114%. Wow. Okay. So, break that down. It means that not only are they keeping all of their existing customers, but those same customers are on average spending 14% more than they did the year before. It's an annuity, a predictable cash stream that smooths out the ups and downs of the chip business.
And that cash flow lets you build a serious competitive moat, which which I think is the next logical step here. How are they so profitable? Well, it all comes down to vertical integration. They're not just selling one piece of the puzzle. They're basically the full stack AI infrastructure company. So, they control everything from the custom chip to the
to the high-speed Ethernet switches like their tomahawk line, the controllers, RF components, and now the whole software layer on top. It's the entire ecosystem. And that creates the lock in, a massive lock in. The switching costs are just enormous. I mean, think about it. If you're a hyperscaler and you design your entire multibillion dollar AI data center around Broadcom's chips and switches,
you can't just swap them out. You're looking at an estimated 18 to 24 months of a complete rearchitecture to change suppliers. One report had a great line for it. What was that? Broadcom's chips are the plumbing of AI data centers. You don't change plumbing midbuild. That's a perfect analogy. Yeah. And that lockin has to translate into some pretty spectacular margins.
It does. I mean, their gross margins are consistently in the mid70% range, but the operating leverage is what's truly incredible. The operating income margin recently hit a record 50.9%. Wait, so for every dollar of revenue, more than 50 cents is dropping down to operating income. 50 cents. It's just an incredibly efficient machine.
And all of that leads to their real superpower, cash generation. Free cash flow. Yep. Adjusted EBITDA was $43 billion in fiscal 2025. But free cash flow, the real cash they can use, was 26.9 billion. That's up 39% year-over-year. And some analysts think that number could triple by 2029. The street is looking at a potential FCF of around $94 billion by fiscal 2029.
And that, you know, that massive predictable cash flow is what lets them handle the debt from acquisitions and still invest billions in R&D. Okay, that's a really strong bull case. But you can't have this kind of runup without some serious risks. We have to look at the bare case. Absolutely. And there are two big ones that always come up.
The first is customer concentration. Right. This AI boom is really just a few massive companies building everything. Exactly. Their top five customers account for 42% of total revenue. So if even one or two of those hyperscalers decides to pull back on spending for a couple of quarters, Broadcom feels it immediately.
It's a huge risk. It is. But that's where the software cushion is so important. It's not just a talking point. It's a financial firewall. So, how does that work exactly? Well, the enterprise software business, it's non-yclical. It's annuity-like. And while it's only about 12% of total revenue, it generates an estimated 75% of their total IBIDA.
Wait, hold on. Say that again. 12% of the revenue brings in 75% of the recurring profit. That's it. It's an incredibly high margin, stable segment that provides a financial floor. It's the money that keeps the lights on and R&D funded even if the big AI customers pause for a bit. Okay, so that's a pretty strong defense.
What's the second big risk? It's valuation. Simple as that. The stock has had a huge run and it's trading way above its historical average multiples. If that AI growth slows down even a little, the stock is vulnerable to what we call multiple compression, meaning investors just aren't willing to pay that premium price anymore.
Exactly. It's a crowded trade. But management has been smart about the structural risks. You know, yes, they have a lot of debt, but their net debt to Ebida ratio is a manageable 1.8x. That's pretty strong. And they have a good track record of of absorbing these big acquisitions, right? A very strong track record.
They did it with CA Technologies with Simantech. They're ruthless about optimizing assets and improving margins. The market is betting they can do the same thing with VMware. And it looks like the insiders are betting on it, too. They are. The company returned $17.5 billion to shareholders last year.
But the real tell for me was the CFO's share purchase back in March. The $7.2 million buy, that's not a small gesture. It's their largest insider buy since 2020. That's a huge vote of confidence. And it's not just him. We're seeing billions in net inflows from big institutions this year. The smart money is buying in.
Okay, so let's look ahead. What are the big catalysts? The dates we need to circle on the calendar. The most immediate one is the Q1 earnings release that's scheduled for April 29th. And the market is going to be watching two things like a hawk which are first the AI revenue growth rate. Can they keep that incredible momentum going?
And second, the guidance for Q2. They need to show continued acceleration and looking a bit further out. The big one is the investor day probably in May or June. That's where we expect them to lay out their AI ASIC roadmap all the way through 2028, giving their customers visibility for years to come.
Exactly. And we're also expecting them to demo their 1.6 TB Ethernet switching chip, which would be an industry first, and just cement their leadership in high-speed networking. But what about uh M&A or even just rumors? Well, they just closed a big deal, the $6.1 billion acquisition of Vellis Technologies.
I feel like that one flew under the radar a bit. It did, but it's strategic. Vellis helps optimize the big AI models that run on Broadcom's hardware. So, it just makes their whole stack stickier. It bridges that hardware software gap and the rumor mill. Anything worth watching? There's persistent chatter about a big deal within the OpenAI ecosystem, which would be massive.
And another total unconfirmed rumor about a multibillion dollar deal with a big auto OEM for vehicle-tocloud AS6 for autonomous driving that would open up a whole new vertical for them. A huge one and it would really help with that customer concentration risk we talked about. It would give them a third powerful engine for growth.
So when you put it all together, it seems like Broadcom has successfully made the shift from being a, you know, a traditional chip company to this full stack AI infrastructure provider. That's the story. And they've got that solid software business as a safety net. The test for them now isn't about demand.
It's about execution. Can they integrate VMware successfully and keep delivering on these huge AI investments? If they can do that while keeping those 50% plus operating margins, well, that combination should be very rewarding for long-term investors. If you enjoy this analysis, make sure you hit that subscribe button to catch our next
Comments 0