Navigation

Investiment.io
Financial analysis platform

Broadcom Inc.

AVGO • NasdaqGS

$333.40
▼13.61 (3.92%)• REGULAR

Broadcom Inc.

$333.40
▼3.9%
📊Summary📋SEC Filings👥Insider Trading
📄

AVGO Q1 2024 Earnings Call

March 7, 2024 at 12:00 AM

Company
AVGO
Quarter
Q1 2024
Date
March 7, 2024 at 12:00 AM
Speakers
2
Word Count
~5,787

Transcript Content

Operator
Our first question comes from Harsh Kumar with Piper Sandler.
Speaker
Yes, thank you, Hock. Once again, tremendous results and tremendous activity that you guys are benefiting from in AI. But my question was on software. I think if I heard you correctly, Hock, you mentioned that your software bookings will rise quite dramatically to $3 billion in 2Q. I was hoping that you could explain to us why it would rise almost 100% up, if my math is correct, in 2Q over 1Q. Is it something simple? Or is it something that you guys are doing from a strategy angle that's making this happen? Hock Tan (CEO): As I indicated, with the acquisition of VMware, we're very focused on selling, upselling, and helping customers not just buy but deploy this private cloud, what we call virtual private cloud solution or platform on their on-prem data centers. It has been very successful so far. And I agree it's early innings still at this point. We just closed on the deal late November and we are now in March, early March. So we had the benefit of at least three months, but we have been very prepared to launch and focus on this push initiative on private cloud, VCF. The results have been very much what we expect it to be, which is very, very successful.
Operator
Our next question comes from the line of Harlan Sur with JPMorgan. Your line is open.
Speaker
Good afternoon. Thanks for taking my question. Hock, on the AI outlook being revised from greater than $7.5 billion, I think, last quarter to $10 billion plus this quarter. As you mentioned, AI compute pulls your ASICs, but it also pulls your networking, optical, PCIe, connectivity solutions as well. So can you just help us understand like of that $2.5 billion increase in outlook? Is it stronger AI ASIC demand, stronger networking, stronger optical, etc.? But more importantly, are you also seeing a similar acceleration in your forward ASIC design win pipeline as well? Hock Tan (CEO): There's a lot of questions, a lot of information you want me to disgorge. Let's take them one at a time, shall we? Yes, the increase, as we have said before, it's roughly two-thirds, one-third or 70-30, which is AI accelerators, which are custom ASIC AI accelerators. We've a couple of hyperscalers compared to the other components, which are collectively considered networking components. And it's about a 70% to 30% mix. That increase of almost $3 billion that you mentioned, it's a similar combination.
Speaker
And then are you seeing a similar acceleration on the forward design win pipeline and customer engagements? Hock Tan (CEO): I have indicated that I only really only have to say, I don't count anybody. I do not go into production as a rail customer at this point.
Operator
Our next question comes from the line of Vivek Arya with Bank of America Securities. Your line is open.
Speaker
Thank you for taking my question. Hock, on, again, on the over $10 billion for AI, is this still a supply constrained number? Or do you think that this is kind of a very project-driven number, so it's not really supply that gets it? So if you were to get, let's say, increase supply, could there be upside? And then, kind of part B of that is, on the switching side, have you already started to see benefits from the 51 terabit per second switches? Is that something that comes along later? Like what is the contribution of 51T to the switching upside that you mentioned for this year? Hock Tan (CEO): Yes. No, our Tomahawk 5 is going great guns. Now it's not driven, unlike in the past, Tomahawk 3, Tomahawk 4 by traditional scale-out in hyperscalers on their cloud environment. This is all largely coming from a scaling out of AI data centers, the building of larger and larger clusters to enable generative AI computing functionality. And you're going for bigger and bigger pipes. Hence, the Tomahawk 5 51 terabit is a perfect solution, and we're seeing a lot of demand. In many cases, we are basically, they are surpassing the rate of adoption that we had previously thought. It is a very good solution in connecting GPUs. And with respect to AI accelerators, where I think you are focusing on, is that a constraint on supply chain? We do get enough lead time out of our hyperscale customers that we do not have a supply chain constraint.
Speaker
Thank you.
Operator
Our next question comes from the line of Stacy Rasgon with Bernstein Research. Your line is open.
Speaker
Hi, everyone. Thank you for taking my question. I wanted to ask about the core software business. You mentioned that VMware contributed $2.1 billion over the two months, which would suggest that the other software segments, including CA, Symantec, and Brocade, account for around $2.5 billion, indicating a sequential growth of about 25% and nearly 40% year-on-year. Is my calculation accurate? If it is, how is that possible? What is happening in the core business? How should we view the growth of the core business and VMware as we progress through the year? Is VMware still projected to reach $12 billion, or how should we consider that? Hock Tan (CEO): Yes, don't get too excited over that. I think it's certain products and contracts we obtained from strong contract renewals in the old Broadcom contracts, especially in mainframes, were very strong, as were some of our other distributor software platforms. So that has also accelerated, but that's not the star of this show. The star of this show is the accelerating bookings and backlog we are accumulating on VMware.
Speaker
Okay. So VMware is still running at like an $11 billion to $12 billion run rate benefit. So that sounds like that should accelerate. So the overall for VMware should be more than the $12 billion that you talked about. So the core business, the strength this quarter that was kind of a one-time we could model that kind of like falling off because you've shown at the overall software at 20? Hock Tan (CEO): Correct.
Operator
Our next question comes from the line of Aaron Rakers with Wells Fargo.
Speaker
Thank you for taking my question. Continuing the discussion on VMware, Hock, now that you have had the asset for some time, I'm interested in understanding how your go-to-market strategy for VMware compares to your previous software acquisitions. Specifically, how have you approached the segmentation of VMware's customer base? I know there has been some conversation around your engagement with the legacy VMware channel in the past, so I'm curious about how you are managing the go-to-market strategy. Hock Tan (CEO): I think we haven't had this for long, just about three months. However, it appears that things are progressing very well, as we anticipated. Our focus is not only on go-to-market strategies but also on enhancing our VCF stack, which we have developed and are actively selling. We're committed to supporting our customers in deploying it effectively in their data centers. Our attention is primarily on around 2,000 strategic customers who maintain significant on-prem distributed data centers. Many are exploring hybrid solutions. A lot of these customers have legacy but essential mainframes that remain critical despite being outdated. As they modernize their workloads, they have options, balancing applications in on-prem distributed data centers with deploying some workloads in the public cloud due to elastic demand. Currently, most customers find their on-prem data centers do not match the high availability, low latency, and resilience offered in the cloud, which is exactly what we're providing with VMware Cloud Foundation. It's effectively mirroring the public cloud experience, and customers appreciate it. In just three months, we've seen positive results reflected in our booking levels.
Speaker
Thank you.
Operator
Our next question comes from the line of Chris Danely with Citi. Your line is open.
Speaker
Thanks, for letting me ask a question. I have just a question on the AI upside in terms of a customer perspective. How much of the upside is coming from new versus existing customers? And then how do you see the customer base going forward? I think it's going to broaden. We know how you like to price, so if you do get a bunch of new customers for these products, could there be some better pricing and better margins as well? Hopefully, they're not listening to the call. Hock Tan (CEO): Thank you for your question, Chris. I appreciate it. Let me give you an overview of our view on the generative AI market. We see it as having two main segments. The first is hyperscalers, particularly the very large ones with extensive consumer subscriber bases. You can probably guess who these players are. They have a massive amount of data and their model focuses on retaining subscribers to enhance the user experience and create better advertising opportunities for their clients, resulting in a quick return on investment. This segment is characterized by a few large players who can scale up investment substantially. Here, custom silicon and AI accelerators are crucial, and that’s where our focus lies. These companies also purchase AI accelerators in large quantities, often in clusters, due to the demands of running foundational and large language models, alongside significant networking components. However, the value of networking, while growing, is still a minor part in comparison to the value of the AI accelerators we offer. The second segment, which is smaller, consists of the enterprise AI segment. This includes companies of various sizes that are starting AI initiatives driven by recent trends and discussions surrounding AI's potential to enhance productivity. Instead of utilizing public cloud solutions, many are attempting to implement these initiatives on-premises. In doing so, they tend to utilize standard silicon for AI accelerators, which falls into the merchant silicon market, where we do not have a competitive presence. On the networking side, however, these enterprises do purchase our networking products such as switches and routers, including devices from companies like Arista. It creates an interesting blend of opportunities in both segments, and we are observing growth in each.
Speaker
Thanks a lot, Hock.
Operator
Our next question comes from the line of Karl Ackerman with BNP Paribas. Your line is open.
Speaker
Yes, thank you. Hock, weakness in broadband, server, and storage customers is understandable given what your peers have said this earnings season. But perhaps you could speak to the backlog visibility you have with your customers in those markets that would indicate those markets could begin to order again and see sequential growth in the second half through the calendar year? Thank you. Hock Tan (CEO): You're correct. As I say, we are almost like near the trough this year. '24, first half, for sure, will be the trough. Second half '24, we don't know yet. But I tell you what, we have a 52-week lead time, as you know. We are very disciplined in sticking to it, and based on that, we are seeing bookings lately significantly up from bookings a year ago.
Operator
Our next question comes from the line of Christopher Rolland with Susquehanna. Your line is open.
Speaker
Thanks for the question. So Hock, this one is for you on optical. So, our checks suggest that you're vertically integrating there. You're now putting in your own drivers, TIAs, you're starting to get traction in PAM4 DSP. I think you kind of had an early lead in 100-gig data center lasers as well. A lot of this should be on the back of AI networking that appears to be exploding here. I was wondering if you could help us size the market and then also talk about how fast this is growing for you. I think there may have been some clues in that one-third number the AI you gave us, but perhaps if you can kind of double click or square that for us, it would be great. Thanks. Hock Tan (CEO): Before we get too carried away, let’s remember that while components like PAM-4, DSPs, optical parts, and filterations exist, they are relatively small in comparison to Tomahawk switches and Jericho routers that utilize AI networks. As many of you know, traditional enterprise networking is currently experiencing a bit of a slowdown. It's important to understand that much of this demand is driven by AI. This can lead to a skewed perspective because the focus and content in networking related to compute differ significantly between an AI data center and a traditional CPU-based data center. I don’t want to mislead you. In the AI data center, there is indeed a considerable amount of content related to DSPs, PAM4s, optical components, and PCI Express switches, but they still represent a minor share of our overall sales when compared to switches and routers. In relation to AI accelerators, they are even smaller. As I mentioned, our AI revenue is projected to exceed $10 billion this year with 70% coming from AI accelerators and 30% from everything else. Out of that, around 20% consists of switches and routers, with the remainder being various retirement DSP components because, as I stated, we are not fully vertically integrated and do not produce the complete optical transceiver. These are typically manufactured by OEM contract manufacturers, like Eoptolink in China, who are more competitive. However, we do supply the key components we discuss, and looking at it this way helps clarify the weighting of the various values.
Speaker
Super helpful. Thank you, Hock.
Operator
Our next question comes from the line of Toshiya Hari with Goldman Sachs. Your line is open.
Speaker
Hi, thank you for taking the question. Hock, I think we all appreciate the capabilities you have in terms of custom compute. I asked this question last quarter on the group call back. But there is one competitor based in Asia who continues to be pretty vocal and adamant that one of the future designs at your largest customer, they may have some share, and we're picking up conflicting evidence, and we're getting a bunch of investor questions. I was hoping you could address that and your confidence level in sort of maintaining, if not extending your position there? Thank you. Hock Tan (CEO): You know, I can't stop somebody from trash talking, okay? It's the best way to describe it. Let the numbers speak for themselves, please, and leave it that way. Most of our products tend to always have, as we do here, a very deep strategic and multiyear relationship with our customer.
Operator
Our next question comes from the line of Vijay Rakesh with Mizuho. Your line is open.
Speaker
Yes, hi, Hock. Just on the custom silicon side, obviously, you guys dominate that space. But you mentioned two customers, only two major customers. I was just wondering what's really holding back other hyperscalers from ramping up their custom silicon side. And on the flip side, you're hearing some peers talk about custom silicon road maps as well, so if you could hit both? Thanks. Hock Tan (CEO): Well, first, we don't dominate this market. I only have two customers, so I can’t be dominating it fully. Second, developing custom silicon is a long-term process. It requires significant effort to create effective solutions for generative AI and other AI capabilities in our data centers, which involves much more than just hardware. You need to invest in software models that are compatible with your custom silicon and align with your business model. This leads to the creation of foundational models that need to be optimized for the custom silicon being developed. It’s an iterative process that constantly evolves, even for the same customers we work with. I mentioned this in the last call; it takes years to reach a point where you can confidently deliver production-ready solutions. This delay isn’t due to poor silicon but rather the incompatibility with the foundation models the customer has and the software and firmware layers that support it. Everything needs to function correctly, creating an entire ecosystem, which we do well in x86 CPUs, but this is still early-stage for AI accelerators like GPUs. It takes time. For the two customers we've been working with, one has been for eight years. Patience is essential; I only have these two for now.
Speaker
And on the peers getting into that market? Hock Tan (CEO): I have no comment on the custom silicon market. What I can say is that I am not interested in entering a market where we have a specific philosophy at Broadcom. Others may have different views, but my philosophy, which I have shared multiple times and is clear to my management team and everyone at Broadcom, is straightforward: focus on what you excel at and consistently invest in areas where you have a distinct advantage. By doing this, you stay ahead of the competition. Avoid pursuing opportunities where you believe you can compete, but where others are clearly outperforming you. That is my philosophy.
Operator
Our next question comes from the line of Matt Ramsay with TD Cowen. Your line is open.
Speaker
Thank you very much for squeezing me in, guys. Just kind of a 2-part thing on the custom silicon stuff. I guess, Hock, if some of the merchant leaders in AI who are interested in some custom networking stuff from you either in switching routing, would you consider it? And the second question is for Kirsten. The business model around custom silicon for most folks is taking our e-payments upfront and selling the end product at a lower gross margin, but a higher operating margin, and you guys have wrapped this massive custom business with no real impact on gross margin. So maybe you could just unpack the philosophy and the accounting about the way that you guys approach the custom silicon opportunities just from a margin perspective? Thanks, guys. Hock Tan (CEO): I'll address your question about the business model rather than getting into specific numbers. There's no clear definition of what an AI accelerator is. Currently, an AI accelerator, whether for merchants or customers, requires more than just numerous floating point multipliers for matrix multiplication and analysis. It also needs substantial memory, essentially high-speed cache memory. The chip isn't just a basic multiplier; it includes memory and has a three-dimensional architecture. Designing or building memory isn't something we excel at in AI accelerators, so we source high-bandwidth memory from specialized suppliers, which is common in the industry. When these components come together, that defines an AI accelerator. While I might achieve a favorable gross margin on the logic chip, it's impossible to apply the same margin to the high-bandwidth memory, which significantly contributes to the overall chip cost. This fundamental math indicates that the gross margin for the entire AI accelerator is lower compared to traditional silicon products. The addition of memory leads to this natural margin reduction, and no one can rationalize applying higher margins to memory. Consequently, the overall margin is inherently lower. However, regarding the logic aspect, with the high-density floating point multipliers we develop on advanced silicon, we can maintain margins comparable to our overall gross margin.
Operator
Our next question comes from the line of Edward Snyder with Charter Equity Research. Your line is open.
Speaker
Thank you very much. First, related to a housekeeping item, Hock, you referred to the second customer. You also mentioned that achieving results takes years of iterative work, which anyone familiar with TPU history would understand. You've previously stated that ramping up takes time. Could you provide some insights into the impressive growth of your custom silicon products? Is a significant portion of that growth coming from your second customer, and considering the low revenue figures, is the growth rate generally comparable? I also have a question about VMware. Hock Tan (CEO): You should reach out to VMware customers. I can't discuss my customers individually, sorry.
Speaker
Okay. Okay, never mind. That's a waste of time. So closing VMware held kind of a significant shift in your software strategy from focus on the largest 1,000 or so customers to hundreds of thousands now. Why should we expect once you get through, I don't say, the low-hanging fruit of selling into the, like you mentioned, the first 1,000 customers with the VCF product, that your OpEx is a share of sales, firstly, in sales and marketing would start to increase because that's the big leverage Broadcom has had over almost all your acquisitions in software, and that seems to be changing now? Hock Tan (CEO): We have a shift here. You're right, now we've got. We are spending more on go-to-market and support because we have a lot of customers with VMware. They are 300,000, but we stratify. We have the strategic guys. We sell, upsell the VC private cloud very good. The long tail of what we call smaller commercial customers, we continue to support and sell improved versions of just vSphere compute virtualization to improve productivity on their service. We don't attempt to say, go build up your whole VCM. They don't have the skills or the scale to do it. Only it adds up you're right, the costs of my spend, OpEx spend, be it support services, go-to-market will increase. The difference between that and, say, CA, under acquisition we did is we're growing this business very fast. You don't have to increase your spend growing this business. So we have operating leverage through revenue growth over the next three years.
Speaker
Great. If I could squeeze one more in. You'd mentioned several times actually in the last quarter that there were two divisions you're going to divest including Carbon Black, and that's changed. What has changed? Is the market outlook kind of softened, and you just wait and see? Or did you change your strategy and how you're going to integrate? I'm just curious why last quarter you said you probably get rid of it in months and now you're keeping it. Hock Tan (CEO): We believe we can create more value for our shareholders by integrating Carbon Black, which is relatively small, into Symantec to establish the enterprise security group. The effect on revenue and profitability will not be substantial.
Operator
Ladies and gentlemen, just in the interest of time, I would now like to turn the call back over to Ji Yoo for closing remarks. Ji Yoo (Head of Investor Relations): Thank you, operator. In closing, we would like to highlight our Broadcom enabling AI and Infrastructure Investor Meeting on Wednesday, March 20, 2024, at 9:00 a.m. Pacific / 12:00 p.m. Eastern Time. Charlie Kawwas, President of Broadcom's Semiconductor Solutions Group and several General Managers will present on Broadcom's merchant silicon portfolio. The live webcast and replay of the investor meeting will be available at investors.broadcom.com. Broadcom currently plans to report its earnings for the second quarter of fiscal '24, after close of market on Wednesday, June 12, 2024. A public webcast that Broadcom's earnings conference call will follow at 2:00 p.m. Pacific Time. That will conclude our earnings call today. Thank you all for joining. Operator, you may end the call.
Operator
Thank you. Ladies and gentlemen, this concludes today's conference call. Thank you for your participation. You may now disconnect.