Earnings Call Transcript - MRVL Q1 2024

Operator, Operator

Good afternoon, and welcome to Marvell Technology Inc.’s First Quarter of Fiscal Year 2024 Earnings Conference Call. All participants will be in listen-only mode. After today’s presentation, there will be an opportunity to ask questions. Please note, this event is being recorded today. I would now like to turn the conference over to Mr. Ashish Saran, Senior Vice President of Investor Relations. Please, go ahead.

Ashish Saran, Senior VP of Investor Relations

Thank you, and good afternoon everyone. Welcome to Marvell's first fiscal quarter 2024 earnings call. Joining me today are Matt Murphy, Marvell's President and CEO; and Willem Meintjes, our CFO. Let me remind everyone that certain comments made today include forward-looking statements, which are subject to significant risks and uncertainties that could cause our actual results to differ materially from management's current expectations. Please review the cautionary statements and risk factors contained in our earnings press release, which we filed with the SEC today and posted on our website, as well as our most recent 10-K and 10-Q filings. We do not intend to update our forward-looking statements. During our call today, we will refer to certain non-GAAP financial measures. A reconciliation between our GAAP and non-GAAP financial measure is available in the Investor Relations section of our website. With that, I'll turn the call over to Matt for his comments on our performance. Matt?

Matt Murphy, President and CEO

Thanks Ashish, and good afternoon everyone. For the first quarter of fiscal 2024, Marvell's revenue was $1.322 billion, above the midpoint of guidance. Higher revenue drove our non-GAAP earnings per share to $0.31, $0.02 above the midpoint. We are guiding revenue for the second quarter to go to $1.33 billion at the midpoint and expect sequential revenue growth to accelerate in the second half of the fiscal year. Before we get to our results for each end market, let me start by discussing the tremendous opportunity that AI represents for Marvell. In the past, we considered AI to be one of many applications within the cloud, but its importance and therefore the opportunity has increased dramatically. Generative AI is rapidly driving new applications and changing the investment priorities for our cloud customers. Today's AI workloads require truly massive datasets. To efficiently process this data, the architecture for AI data centers is significantly different than standard cloud infrastructure. Rather than dual socket servers being the core element in the rack, the primary building block in AI is a system containing multiple accelerators such as GPUs. In large deployments, thousands of these systems are interconnected to form a data center-sized AI cluster, and what's required to interconnect these systems is orders of magnitude higher than in standard cloud infrastructure. To give you an idea, the latest dual CPU server in the cloud data center today can drive up to 200 gigabits per second of I/O and contains the network interfaces to support that bandwidth. In contrast, an advanced AI system containing 8 accelerators can drive close to 30 terabits of full duplex bandwidth, which is hundreds of times more bandwidth required to connect these systems together. Keep in mind that cloud data centers connect thousands of these systems in a single cluster to provide maximum scalability for their customers, with each of these systems capable of driving tens of terabits of network traffic. These clusters require a staggering amount of high bandwidth connectivity, all of which needs to be provided at ultra-low latency and high reliability. And within a reasonable power outlook. This connectivity is best provided by fully optically connected infrastructure, utilizing digital signal processing and low latency high capacity fabric switches. This is why we see AI as a strong growth driver for our PAM4 optical DSP platform. Furthermore, it is important to note that these DSPs are compatible with various network protocols such as Ethernet, InfiniBand, and other proprietary solutions for maximum breadth and flexibility. Going forward, we see this trend only accelerating. Today's most advanced AI systems are already using the highest bandwidth interfaces available, but the performance scale of current AI implementations is still constrained by network capacity. Take the AI system I discussed earlier, for example. Those systems typically contain four 800 gigabit per second optical interfaces for external connectivity, which provide an aggregate of 3.2 terabits per second of bandwidth, but that is in a system capable of driving close to 10x that bandwidth. To bridge this gap, we expect that the number of optical interfaces per AI system will continue to grow as these clusters scale, and the next generation of accelerators is expected to have even more compute capability and higher bandwidth requirements. Given the speed at which AI infrastructure is advancing, the technology refresh rate is happening at 18 to 24 months versus 4 plus years in standard infrastructure. Altogether, we expect a massive amount of connectivity in these AI clusters, increasing adoption of higher speed optical interfaces, and faster refresh rates to be major demand drivers for our PAM4 DSP platform. In fact, in our last earnings call, we indicated that the ramp in our industry-leading 800 gig DSP platform was driven almost entirely by AI applications. We have also announced the industry's first 1.6 terabit PAM4 DSP platform, doubling the throughput from the current generation, and we expect AI to drive the initial adoption of these products as well. Additionally, as inference is deployed at multiple regional data centers, they need to be connected by high bandwidth low latency optical links over tens of kilometers. This technology has noticed data center interconnect or DCI, and Marvell has been a pioneer in this market. We created the industry's first pluggable module for DCI, and we are now providing 400 gigabits per second in our latest DCI product line. We already see that AI cloud data centers are driving a significant increase in demand for 400ZR solutions. Another demand driver for our DCI products is that the next generation AI implementations plan to cluster accelerators across different sites. AI is also a key growth driver of demand for switching inside the data center. Marvell has a growing position in the market for low latency high capacity switches, and we are seeing strong demand for our products. We recently announced our next-generation 51.2T Teralynx 10 Ethernet switching platform. This platform is based on the low latency architecture we acquired from Innovium and is built on Marvell's leading 5 nanometer technology platform. We are expecting strong interest in this product and anticipate AI to lead the industry's transition from today's 12.8T switches to 51.2T, enabling a quadrupling of network bandwidth. We expect 51.2T adoption to be a strong growth driver for the networking semiconductor market. In summary, Marvell has built a leading position in network connectivity for AI and we expect tremendous growth for our PAM4 optics DCI and Ethernet switching solutions fueled by the growing investment in AI. Perhaps even more exciting is Marvell's opportunity to address compute in AI through our cloud optimized silicon platform. As mentioned earlier, the architecture of an AI data center is fundamentally different than standard cloud infrastructure. In standard cloud infrastructure, the bulk of the compute is performed by CPUs, while accelerators are primarily used for off load tasks, such as networking and security. In contrast, in AI infrastructure, accelerators like GPUs serve as the primary compute engines, while CPUs are used for control purposes—a 180-degree reversal from standard cloud infrastructure. This fundamental difference results in a higher ratio of accelerators to CPUs in AI, with accelerators now dominating the compute opportunity. This combined with the sharp increase in AI investment is driving a higher proportion of the incremental compute in cloud data centers towards accelerators. Today, we see cloud customers enhancing their AI offerings by building custom accelerators of their own designed to address their specific needs. This is a core part of Marvell's cloud optimized silicon strategy, and we now see a much larger and faster growing opportunity for custom compute and AI infrastructure. When we previously discussed our cloud optimized silicon opportunities and revenue ramp expectations at our Investor Day in October 2021, we projected revenue from the first set of design wins to grow to $800 million annually once all the programs were in production. These designs included various accelerated computing applications including AI, security, storage, and video along with several networking applications. When we look today at the same set of design wins driven by AI, our total lifetime revenue expectations from these have increased significantly. At the same time, the relative proportion of projected lifetime revenue from AI has increased from approximately 20% in our prior forecast to well over half today. As a result, computing and AI applications have now grown to become the single largest revenue driver and opportunity for Marvell's cloud optimized silicon platform. With the continuing increase in demand from AI, we foresee annual revenue from those same cloud optimized design wins well exceeding the prior $800 million projection as these programs ramp over time. In fact, we have several custom silicon products tied to AI expected to ramp into volume production next year. For one of these programs, initial samples are already up and running at our customer's lab and qualification is proceeding well. In another case, we expect to take out this quarter and deliver first silicon before the end of the calendar year. As you can see, in addition to our strong position in network connectivity for AI, we believe we are well-positioned to address the compute opportunity. Over the last few months, we have taken time to carefully map our revenue from AI so we can track its progress over time. Given the tremendous progress we've made, including the recent demand increases from our customers, we are expecting our revenue driven by AI applications to grow at an accelerated pace. In fiscal 2023, we estimate that our AI revenue was approximately $200 million, up dramatically from the prior year. This revenue was primarily from our PAM4 optics and 400ZR DCI products. Since our last earnings call in early March, bookings for these products have increased very significantly. Consequently, we expect Marvell's overall AI revenue to at least double in fiscal 2024. Looking to fiscal 2025, we expect robust growth to continue from AI for network connectivity. Layering on top of this is the growth we expect from the ramp of the cloud optimized programs we discussed earlier in the call. In aggregate, we foresee our overall AI revenue at least doubling again next year. In other words, we are forecasting an AI revenue growth CAGR over 100% over the fiscal 2023 to 2025 timeframe. In the future, we expect generative AI implementations involving video and images to provide a tailwind to overall storage and exabyte growth, both in HDD and Flash. However, it is difficult to accurately allocate revenue from our storage business specifically to AI. So, the AI revenue forecast I just discussed does not include any storage contributions. In addition, we expect the increase in network traffic from AI will also provide a tailwind for our broader networking portfolio over time, which we have not yet captured in our AI revenue forecast. We've seen rapid shifts in our cloud customers' plans to spend on AI infrastructures, which is becoming a much bigger portion of their CapEx. We believe that Marvell is one of the scarce few semiconductor companies positioned to enable this trend and is uniquely able to participate in all three aspects of AI systems: networking, compute, and storage. We set out in 2016 to pivot Marvell to data infrastructure and have successfully executed that strategy, forming the right team, technology, and customer relationships to lead in this market. With AI becoming the ultimate data infrastructure application, Marvell is at the center of this incredible transformation. We look forward to sharing our continued progress in AI in the future. Let me move on now to reviewing our results and expectations by end market starting with data center. In our data center end market, revenue for the first quarter was $436 million, declining 32% year-over-year and 12% sequentially. Our overall data center revenue in the first quarter was higher than guidance driven by cloud where we saw stronger demand for our optical data center interconnect products from expanding AI deployments. As expected, storage was responsible for the majority of the overall sequential decline in our data center revenue in the first quarter, although we are forecasting sequential growth to start in the second quarter and our data center storage business to continue to grow in the second half. Looking ahead to the second quarter for our overall data center end market, we expect cloud revenue to grow over 10% sequentially. However, we are expecting the enterprise on-premise portion of our data center end market to decline and offset growth from cloud. As a result, we expect revenue from our overall data center end market to be flat sequentially in the second quarter. Turning to our carrier infrastructure end market, revenue for the first quarter was $290 million, growing 15% year-over-year and 5% sequentially. We saw strong demand for our wireless products, as 5G adoption continued to expand in new geographic regions along with our customer-specific product ramps. This resulted in approximately 25% sequential growth in wireless revenue. Partially offsetting our wireless growth was the start of inventory digestion in our wired end market. Moving on to our outlook for the next quarter, we expect revenue from our overall carrier end market to decline in the mid-single-digits sequentially, due to continued inventory digestion in our wired end market. Moving on to our enterprise networking end market. Revenue for the first quarter was $365 million, growing 27% year-over-year and flat sequentially, which is better than our guidance. Our first quarter enterprise networking results reflected a strong ramp in custom ASICs offset by our planned reduction in channel and customer inventory of our merchant products. Looking ahead to the second quarter of fiscal 2024, we project our enterprise networking revenue to decline by more than 10% sequentially, due to inventory corrections in this end market. Turning to our automotive and industrial end market, revenue in the first quarter was $89 million, flat year-over-year and declining 10% sequentially. While our automotive business continues to deliver strong growth both year-over-year and sequentially, these gains were offset by a decline in our industrial business. Looking to the second quarter of fiscal 2024, we project revenue from our auto and industrial end market to grow sequentially in the low teens on a percentage basis. Moving on to our consumer end market, revenue for the first quarter was $142 million, below our guidance, declining 20% year-over-year and 21% sequentially. Looking ahead to the second quarter, we are forecasting revenue to grow sequentially in the mid-30% range, driven by strong seasonal growth in demand for our custom SSD controllers. Additionally, we expect growth from our ACD controllers as we start shipping closer to end market consumption. In summary, we are guiding sequential growth to resume starting with the second quarter, and we expect growth to accelerate in the second half of the fiscal year. A few markets remain choppy, and we expect revenue from our wired and enterprise end markets to continue to trend down due to macroeconomic uncertainty and inventory corrections. We anticipate revenue from our wireless end market to step up in the third quarter before taking a pause and declining in the fourth quarter. We see several positive signs in our data center end markets. We expect our data center storage revenue to resume growth in the second quarter and continue to improve as we start shipping closer to end market demand. Putting aside storage, we expect cloud to be the key growth driver for our data center revenue. We are guiding our cloud revenue to grow over 10% sequentially in the second quarter, and we are seeing a significant increase in demand from AI for the rest of the year. Additionally, we expect our cloud optimized silicon programs to ramp more meaningfully as the year progresses, in line with the revised expectations we outlined during last quarter's call. As a result, we are confident in our expectations for higher growth for the company in the second half of the year. We also anticipate significant margin improvement as we move forward. We expect that inventory corrections will be mostly behind us by the end of this year, and we are excited about the resumption of revenue growth driven by Marvell's specific product ramps. We have been laser-focused on a number of cost improvement actions to enhance gross margin. Consequently, we are confident in our forecast for our non-GAAP gross margin to return to at least the low end of our target range in the fourth quarter of this fiscal year. We've been in extensive discussions with our customers and have a clear view of their current roadmaps. While some programs have been pushed out, others have been accelerated. We are aggressively reprioritizing our investments, aligning to the highest ROI opportunities in front of us. Simultaneously, we are focused on driving additional efficiency to further reduce our operating expenses. We are making strategic roadmap adjustments and combining some of our businesses to reflect changes in the market. For example, resulting from our customers' growing needs for custom compute, we've combined our custom ASIC and processor product groups into a single organization. Other teams are also working on efficiency improvements. Consequently, we expect our OpEx to exit at a lower run rate than previously communicated. This discipline on expenses is expected to carry into fiscal 2025 and help us to deliver strong operating leverage going forward. Willem will provide more details in his commentary. Before closing, I'd like to express my appreciation to Rick Hill, who will be retiring from the Marvell Board at the end of his current term. Rick joined Marvell a little over 7 years ago as Chairman of the Board and acting CEO during a very difficult time. He moved quickly to stabilize the company, personally reassuring customers, employees, and shareholders. He helped establish a world-class board of directors and was instrumental in recruiting the first key group of executives. I joined Marvell in large part because of Rick. It is very rare to find an individual who is willing to contribute their time and energy and impart their wisdom to others for the satisfaction of seeing a company thrive and the next generation leaders succeed. At the most critical moments over the past 7 years, Rick always had the Marvell team's back, and his belief in the team enabled us to take bold steps to transform the company. Without his leadership and mentorship, Marvell would not be where it is today. I'm truly grateful to Rick for his exceptional contributions to Marvell, and it's an honor for me to succeed him as Board Chair. Additionally, I would like to express my appreciation to Dr. Ed Frank, who joined us from the Cavium Board in 2018. Ed has been an outstanding director and will be retiring from the Marvell Board when his term expires in June. I'm grateful to Ed for his years of dedicated service to Marvell. Finally, I want to thank our dedicated team of employees. Together, we are moving forward with confidence that Marvell's best days are yet to come. With that, I'll turn the call over to Willem for more detail on our recent results and outlook.

Willem Meintjes, CFO

Thanks Matt, and good afternoon everyone. Let me start with a summary of our financial results for the first quarter of fiscal 2024. Revenue in the first quarter was $1.322 billion, exceeding the midpoint of our guidance, declining 9% year-over-year and 7% sequentially. Data center was our largest end market driving 33% of total revenue. Enterprise networking was the next largest with 27% of total revenue, followed by carrier infrastructure at 22%, consumer at 11%, and auto industrial at 7%. GAAP gross margin was 42.2%. Non-GAAP gross margin was 60%, in line with guidance. Our gross margin results reflected the adverse revenue mix we had forecasted for this first quarter. GAAP operating expenses were $740 million, including share-based compensation, amortization of acquired intangible assets, restructuring costs related to headcount reductions, and acquisition-related costs. Non-GAAP operating expenses were $459 million, slightly below guidance. First quarter operating expenses reflected the step-up from the prior quarter, due to typical seasonality and payroll taxes and employee salary merit increases. GAAP operating margin was negative 13.8%. Non-GAAP operating margin was 25.3%. For the first quarter, GAAP loss per diluted share was $0.20. Non-GAAP income per diluted share was $0.31, $0.02 above the midpoint of guidance. Now, turning to our cash flow and balance sheet. During the quarter, cash flow from operations was $208 million. This included $40 million in payments for previously committed long-term capacity payments. Looking ahead, we are currently anticipating minimal additional payments for capacity for the rest of fiscal 2024 and beyond. As a result, we expect a much reduced headwind to operating cash flow as compared to fiscal 2023, where payments for long-term capacity totaled approximately $250 million. Inventory at the end of the first quarter was $1.03 billion, decreasing by $42 million sequentially. Our days sales outstanding was 69 days, a decrease of 7 days from the prior quarter. We returned $51 million to shareholders through cash dividends. As of the end of the fiscal first quarter, our cash and cash equivalents were $1.03 billion, almost all of which is held in systemically important banks. Our total debt was $4.68 billion, which included a drawdown of $200 million from our revolving line of credit. Our gross debt to EBITDA ratio was 2.1x and net debt-to-EBITDA ratio was 1.6x. Before I provide specific guidance for the second quarter, let me describe our plan for gross margin, operating expenses, and capital allocation for the rest of the year. As you are aware, significant inventory corrections in certain businesses caused an adverse revenue mix, which impacted our gross margin. We are optimistic that inventory corrections will be largely done by the fourth quarter and that we will be shipping closer to end demand at that time. We have also put in place multiple cost reduction efforts to improve gross margin. Internally, we are optimizing headcount and further streamlining operations. Externally, we continue to partner with our strategic suppliers to drive more efficiency in the supply chain. We are confident that as a result of mix improvement and our cost reduction efforts, our non-GAAP gross margin will start to improve. We expect modest improvement in the third quarter and by the fourth quarter, we can return to at least the low end of our target range of 64% to 66%. As Matt told you, we are making strategic roadmap adjustments and driving additional efficiencies, all while maintaining our long-term growth outlook. As a result, we expect to reduce our non-GAAP operating expense in the fourth quarter to be approximately $430 million. The fourth quarter of fiscal 2024 is a 14-week quarter. So, included in this forecast is approximately $10 million in expenses for the extra week. Therefore, on a normalized 13-week basis, we expect the run rate for our non-GAAP OpEx exiting this year to be $420 million. As we look ahead to fiscal 2025, we expect that typical seasonality and payroll taxes and employee salary merit increases in the first quarter will cause OpEx to step up sequentially. As a result, we are modeling first quarter fiscal 2025 OpEx to increase sequentially by mid-to-high single-digits on a percentage basis from the normalized 13-week exit rate of $420 million in the prior quarter. We are confident that this sets us up to drive tremendous operating leverage next year. In terms of capital allocation, we have $500 million of our total debt due in January 2023 that we are planning on paying off using our cash balance and free cash flow. This will improve our gross leverage significantly going forward. Following this payoff, we expect to resume our buyback program in the third quarter. Turning to our guidance for the second quarter of fiscal 2024. We are forecasting revenue to be in the range of $1.33 billion, plus or minus 5%. We expect our GAAP gross margin will be in the range of 44.3% to 46.8%. We project our non-GAAP gross margin will be in the range of 60% to 61%. We project our GAAP operating expenses to be approximately $694 million. We anticipate our non-GAAP operating expenses to be approximately $455 million. We expect other income and expense, including interest on our debt to be approximately $51 million. For the second quarter, we expect a non-GAAP tax rate of 7%. We expect our basic weighted average shares outstanding will be 861 million, and our diluted weighted average shares outstanding will be 865 million. As a result, we anticipate GAAP earnings per share in the range of a loss of $0.16 per share on the low-end to an income of $0.05 per diluted share on the high end. We expect non-GAAP income per diluted share in the range of $0.27 to $0.37. In summary, as we move forward through this fiscal year, we project our revenue to continue to grow sequentially, gross margin to improve significantly in the fourth quarter, and operating expenses to continue to step down. Our DSO and inventory have already shown signs of improvement. As a result, we are looking forward to driving tremendous operating leverage and significant improvement in cash flow generation over this year, while setting up a strong platform for growth next year.

Operator, Operator

Thank you. Today's first question comes from Ross Seymore with Deutsche Bank. Please go ahead.

Ross Seymore, Analyst

Hi guys. Thanks for letting me ask a question. It sounds like things have changed significantly since your last call, Matt. You went through the AI side in great detail and then the cyclical side sounds like it's bottoming. I just wanted to have you dive into a little bit about the customer behavior. Is it just the cycle side that you find the bottom on finally? Is it the AI side that's really accelerating? Just talk about how that behavior has changed over the course of your quarter, if you could, please?

Matt Murphy, President and CEO

Yes. Hey, Ross. Thanks for the question. Yes, I think it's encouraging to see revenue growth in the second quarter. So overall, with the total company revenue bottoming and then encouraging to see cloud pick back up, growing sequentially in Q2 by about 10%, driven by AI. So that's certainly been a positive. I think I’ll just talk about the rest of the business. There are a number of moving pieces. A lot of it pretty much came in line in terms of things like carrier, automotive, etcetera. So, we haven't seen a ton of behavior change there. The biggest change, I would say is that I see two things related to AI. The first is just the level of bookings, backlog, and forecast increase, particularly for our optical products, which has increased significantly. That's a positive. The second is that we took the time between last quarter and this quarter to quantify our AI revenue stream in a very detailed manner, where it was coming from, and what's driving it. We now have it embedded in our system as a separate end market that we can filter by. So, we have a solid view of the history and the future. What you're seeing is a really positive story, which is basically around a $200 million level of revenue for AI last year, and then we see it at least doubling this year and then at least doubling again next year. That's probably been the biggest change, the uptick in AI, which was a little behind some of the announcements maybe going on last quarter, but it's definitely here for us. The broader cloud infrastructure side, non-AI, is also picking up and will also grow in the second quarter and throughout the year, and actually has a great setup for next year as well. Those are some of the things that have gone on. And just one final note on the AI stuff—I said in my prepared remarks, but just for everybody on the call, we deliberately don't include storage at this time in AI. We found it difficult to map that particular set of products directly to AI systems or not. Clearly, there's going to be a tailwind for that business over time given the amount of data that needs to be stored. The bottom line is you’re seeing a very clean and exciting AI number.

Ross Seymore, Analyst

Thank you.

Operator, Operator

The next question comes from Timothy Arcuri with UBS. Please go ahead.

Timothy Arcuri, Analyst

Hi, thanks. I'm kind of wondering if you can frame, Matt, how much you think revenue is going to start to grow in the back half of the year. There are a lot of moving parts with the wired and enterprise business still digesting inventory. Data center is expected to be up in the back half of the year. It's driven by cloud, and there is some storage coming back as well. However, many of the big networking OEMs are still sitting on record inventories: Cisco, Arista, Juniper. So, can you just sort of frame how much you think revenue can grow as you get into fiscal Q3 and fiscal Q4? Is it reasonable to expect mid-single digits or even better for each of those quarters? Thanks.

Matt Murphy, President and CEO

Yeah. Hey, Tim. So, I’d just give you some data on how to think about it. First, we start with the second quarter. We're seeing sequential growth. When you look at the data center piece, cloud growth is being offset by continued weakness in on-premise enterprise. As you've noted, that's been a clear issue for the overall server market for the last five quarters, if you take the read-through to enterprise networking—the same thing. There's a lot of inventory on the customer balance sheet. We've signaled for some time that we expect that to continue to correct throughout the year. The company has inflected back up in terms of growth in Q2. Looking beyond the second half, if you look into next year, we have AI doubling again in our fiscal 2025, with strong continued growth in our cloud infrastructure. Non-AI cloud will also drive meaningful growth. This is in optics, storage coming back, plus some cloud optimized programs. Automotive continues to grow next year. Storage is coming off the bottom in our guide for Q2, but should be back next year. Many very strong growth drivers are in play. We need to draw a line between the first half now and next year. The way you’re thinking about it is right; some markets will be choppier like enterprise. Additionally, we said that in carrier. Our 5G wireless business has performed incredibly well over the past few years. We expect that to continue in Q2 and Q3 but probably will drop off in Q4 after a long run. Our own product cycles are underway, and they’ll be offset by all the growth in data centers, providing a great set-up for the second half—cloud AI-driven growth will more than offset comprehensive weakness in some of the other end markets. Hopefully, that's helpful.

Timothy Arcuri, Analyst

Thanks, Matt. So, I guess you don't want to put a number on it. So, like mid-single-digits, you don't…

Matt Murphy, President and CEO

We don't, yeah. We're not – like most companies, we're not going to guide the second half specifically at this point, but clearly, we expect growth to accelerate off of where we are today.

Operator, Operator

The next question is from Vivek Arya with Bank of America Securities. Please go ahead.

Vivek Arya, Analyst

Thanks for the question. Matt, I just wanted to understand what specifically is included in AI this year. I think you're suggesting about $400 million or so in terms of a double from last year. How much of that is your electro optics, are you including the Innovium switches and that, what about your cloud optimized silicon, where I think the number was $200 million for this year? So, what exactly is included in AI for this year? And when you spoke about the potential for it to double again next year, of those three pieces, are all three going to double, or is one going to grow more? If you could give us a little more clarity around what exactly is included in AI for this year and amongst those, what will grow next year?

Matt Murphy, President and CEO

Great question, Vivek. So, think of this year’s AI at least doubling, so $400 million plus, and that is almost completely driven by PAM4 optics, PAM4 DSP, as well as some DCI driving a lot of AI traffic and regional data centers. It’s mostly driven by optics. Some networking, as you pointed out, and probably a little bit of cloud optimized, but primarily from the strong 800 gig ramp that we're seeing in PAM4. That continues into next year, and we layer in the cloud optimized ramps as well. Both will contribute to the incremental growth and the doubling next year. As we get through the year Vivek, we'll be able to quantify this a little more. It's incredibly dynamic and exciting. There are new positives almost daily. As we go forward, we'll share more, but for this year, think of the PAM4 train running very strong into next year, and then we’ll start layering in the cloud optimized on top. We have confirmed we're tracking to the initial estimate from last time, which was the $200 million for cloud optimized. This might be higher, but that's a safe number. And on a high level, we won't guide a specific number for next year. What's important to note is that the peak revenue will exceed the previous $800 million as the total lifetime volume of the same design wins has significantly increased, and the AI proportion has increased significantly. So, all of this is tracking in the right direction for this year and next year.

Vivek Arya, Analyst

Understood. Thank you, Matt.

Operator, Operator

The next question is from Ambrish Srivastava with BMO. Please go ahead.

Ambrish Srivastava, Analyst

Hi. Thank you very much. That's a lot of very helpful details there, Matt. I actually wanted to turn on to Willem on the gross margin side. What are the assumptions for the fourth quarter? Because you guys sound very confident about hitting that? Just kind of help us understand the assumptions; how much of it has to be based on the top line recovery?

Willem Meintjes, CFO

Yes, Ambrish, thanks for the question. Yes, we're confident about getting back to the bottom end of our range in Q4. Main drivers include strong wireless carrier, which we expect to continue through the third quarter before taking a step down in Q4. We've seen the storage recovery starting, and we expect that to continue through the fourth quarter. Additionally, we’re seeing strength in our optical products, driven by AI. All of that will positively impact our gross margin. Moreover, all the OpEx reduction initiatives that we're driving internally, along with efforts in our supply chain, should also meaningfully impact our P&L in Q4. So that gives us confidence.

Ambrish Srivastava, Analyst

Yes. So just a clarification, because you said the inventory corrections would be largely over. On the cloud side, does that mean we are behind? Do you see networking being the vestige of that inventory correction?

Matt Murphy, President and CEO

Yes, I think I would—let me clarify one thing. We're not completely out of the woods yet on cloud. Think of the growth as moving back quite dramatically, as you can imagine. The way to think about it is the AI portion of it is moving up strongly, but the traditional infrastructure side is also going up; it’s still got a little bit of a way to clear. There's a little bit more tailwind there regarding inventory digestion on the cloud side; additional improvement should kick in. Storage is still below shipping demand, which will also be a positive for exiting the year. There are still some pockets where inventory overhang exists, as expected. However, we have positive signals that we should be in pretty good shape shipping back to what normalized demand would be. Willem’s point, just some product lines are not experiencing that problem, and wireless 5G shipped quite hard on that one. But when you add everything up, we expect normalized shipping soon.

Ambrish Srivastava, Analyst

Got it. Thank you. Super helpful.

Operator, Operator

The next question is from Blayne Curtis with Barclays. Please go ahead.

Blayne Curtis, Analyst

Thanks for taking the question. Matt, I just wanted to go to some of your AI comments; I thought it was super interesting. You talked about the mix of lifetime revenue moving from 20% to 50%, and you also gave us that extra year. I'm just trying to understand if you've seen an influx of new projects that you've started that would then come out. Has the mix shifted so much. It would suggest a pretty high percentage of your out-year there.

Matt Murphy, President and CEO

Yes. I think, Blayne, there are a couple of pieces to that. First, we had an initial set of design wins that we sized and articulated at the October 2021 Investor Day. That was projected as $400 million going to $800 million. At that time, about 20% of that lifetime revenue was allocated for AI, 80% for traditional cloud infrastructure. The AI piece has taken off significantly since then. To clarify, that’s just going apples to apples with what we communicated two years ago. Those projects are currently coming to fruition as expected. Simultaneously, we've layered in a significant number of incremental new design wins. Many of those are underway down the line next year, so our design win funnel, which I was reviewing at the company level, has expanded dramatically, with the biggest portion of that coming from our cloud design win funnel. The AI portion has become very significant compared to where it was two or three years ago. That means we have a number of aspects—wins we claimed two years ago going into production, an incremental activity since then. So, this indicates long-term potential even higher than we initially predicted in AI. Overall, we're entering very early innings here across the board.

Blayne Curtis, Analyst

Thanks for that. I want to follow up on the carrier side. You previously mentioned India might be a source of growth, but now it seems like it may drop a bit. Just trying to understand if anything changed in your expectations for the India rollout.

Matt Murphy, President and CEO

Yes, Blayne, the way to think about it is generally consistent with our story for 5G that unfolded over the last three years. It has grown virtually every quarter. Last year provided a solid foundation; we hit about $600 million. We had to ramp up to $290 million in Q1, then again in Q2, and we actually expect strong growth in Q2 and Q3 before a step down in Q4. The base station market is very lumpy; deployment is full of concern about CapEx. We’ve seen this run in our 5G wireless business over the last few years. I'm excited about the current situation, but we haven’t yet seen much indication that things will get more challenging next year compared to before. Inventory reduction at 5G won’t slow our overall growth, and we see that demand ramping back as we've had a hot run thus far. We've been able to power through these challenges by being in new products, given that we don’t have a lot of 4G overhang.

Operator, Operator

The next question is from Tore Svanberg with Stifel. Please go ahead.

Tore Svanberg, Analyst

Yes, thank you and congrats on turning around the ship here. I had a question about storage, Matt. I know it's hard for you to identify how much was related to AI, and you don’t want to give a number going forward either. However, as we start to think about CXL and the potential growth in this area, I would think that could be an opportunity for you next year. Any color you can provide on the size of the storage business in AI over the next few years?

Matt Murphy, President and CEO

Got it. Yes, Tore, you bring up a very good point. When I referred to storage earlier, I meant the existing product lines we have, primarily hard disk drive controllers, preamplifiers, and flash controllers. We've spent a lot of time trying to map those out, but we ultimately determined they weren’t straightforward enough to specify at this point. You mentioned CXL, and that is indeed a tremendous demand we are seeing on the customer side. We have significant pressure to roll that product out due to the request. It will definitely drive another segment for us. We also have products like AECs, which are gaining traction. That's another factor we want to consider. We have several pieces in play, so we're leveraging our core DSP and analog mixed signal capability, trying to provide solutions that encompass CXL, ACs, PAM4 DSP, optics, DCI. The goal will be to establish ourselves as the primary supplier for these solutions.

Operator, Operator

The next question comes from C.J. Muse with Evercore. Please go ahead.

C.J. Muse, Analyst

Yes. Good afternoon and thank you for taking the question. A follow-up on storage and trying to think through the ramp in the second half, both for data center and consumer. For data center, I think that business was probably down $150 million peak to trough just for the controller side. How should we think about that step-up into October and January? The same thing on the consumer side, that business could be up about $50 million sequentially in July. Is that sustainable into the second half of the year? Thanks so much.

Matt Murphy, President and CEO

Sure. Let me take it in two parts. On the data center side, yes, it did bottom out in Q1. We expect it to be up a little in Q2, grow a bit over time, but it’s probably more accurate to say it will recover mostly by Q4. We expect Q1 and Q2 to be very low, but growth should return in Q3 and Q4. We're looking at around 75% to 80% recovery by that point for our models. So, while it may not fully rebound, it will normalize. The positive takeaway is that it has come off the bottom and is expected to grow. Now, for the consumer side, this market is not a super strategic space for us. There are moving pieces at play here. We performed well in Q1; performance has been a little funky in Q2, with around 30% growth. However, looking at the average, it seems to be where it’s meant to be. Our model suggests it will trend down over time, but if anything beneficial happens, that would be a bonus we wouldn’t bank on consistently.

C.J. Muse, Analyst

Great. Thank you.

Operator, Operator

The next question comes from Karl Ackerman with BNP Paribas. Please go ahead.

Karl Ackerman, Analyst

Thank you. Gentlemen, I was wondering how much of the uptick in your data center business outlook is coming from the expanded TAM opportunity from a competitor exiting investment of their networking ASIC division where your Innovium business competes quite well?

Matt Murphy, President and CEO

Oh, I see. Yes, you're talking about the switching portfolio, Karl? Yes. We are aware of that repositioning by one of the companies; we have not really seen that solution very much in the market. Perhaps that's one reason why they’ve reprioritized, but we see a lot of traction for our product line. The way to think of this is, if you've got all this CapEx shifting to AI, what happens to the traditional cloud infrastructure side? Will it affect Marvell? Do things like switches, networking get less spend? Or do optics? My answer: we see strong growth for our portfolio in traditional infrastructure, including our switching platforms. Some of that will go back into AI as well. As you look at the spending, the dollars are going to be needed to upgrade the entire front-end network and connections required to move massive data inside and out of data centers. We believe our traditional cloud infrastructure portfolio will continue to thrive in the second half of the year and next year, which includes switching. We have doubled down on this area, combining the Innovium product line with our broader switching organization and appointing a new general manager to run all our higher layer networking technology. We are leveraging this together with our optics, and we're expressing significant investment into co-packaged optics moves forward.

Karl Ackerman, Analyst

That's very helpful. Perhaps if I could ask another follow-up. Could you discuss the debate on the competitive dynamics of AI? There have been some ASIC design houses in China winning AI designs from hyperscalers, but given the extended revenue trajectory you spoke about today of your AI programs, it doesn't seem like the competitive landscape has altered your view on the long-term opportunity in compute offload in different areas of AI. Can you put a finer point on that?

Matt Murphy, President and CEO

Yes. As you heard from our comments, even these initial wins we’ve discussed a few years ago are coming onto production. The opportunity as it stands already consumes a significant amount of our focus, and I would say our design win funnel has grown dramatically while much of that is related to compute offload acceleration and custom computing. On the competitive landscape, there are overseas challenges, and considerations relevant to national security may intensify as AI technologies evolve. So, companies with strong U.S. bases are likely to win, and those that can invest in bleeding-edge technologies like 5-nanometer and 3-nanometer will be well-poised to do so.

Karl Ackerman, Analyst

Helpful. Thank you.

Operator, Operator

The next question comes from Harlan Sur with JPMorgan. Please go ahead.

Harlan Sur, Analyst

Hi, good afternoon. Thanks for taking my question. Maybe just a quick follow-up to the last question on switching. Did the team get first silicon back on your next-generation 51.2T Teralynx 10 switching chipset? What feedback do you have on its performance, and have you shipped samples to customers? Your design win funnel is clearly expanding at a significant pace, as you mentioned. I think it was last October when the team announced their 3-nanometer IP libraries.

Matt Murphy, President and CEO

Yes, great questions. The Teralynx 10 product we call TL10 is in hand and looking good. There’s a lot of demand to ramp this product up as fast as we can; it’s all going well. In terms of 3-nanometer, there has been a lot of activity here. I’d say we're extremely well positioned today and the design wins are across nearly all our major end markets. Our customers are going to become increasingly selective about costs, power, performance, timing, and the like for these leading-edge technologies. Each will find its place within the applications they serve.

Harlan Sur, Analyst

Perfect. Thank you for the update, Matt.

Matt Murphy, President and CEO

You're welcome.

Operator, Operator

The next question comes from Matt Ramsay with TD Cowen. Please go ahead.

Matt Ramsay, Analyst

Hey guys. Thank you for taking my questions. I have two different questions for you. The first one is short-term guidance where it seems like consumer SSD is driving growth. Any additional color on your confidence about the sustainability of that growth considering the pace of recovery in different markets, in China and in return of enterprise spending in the notebook market?

Matt Murphy, President and CEO

I wouldn't read too much into this consumer storage business evolution. The way to think about it is, it's a limited number of products, really customized designs. I wouldn't assume this keeps going. Average those out, and there's some lumpiness. We were down in Q1 and coming back in Q2. However, average this, and it should trend down the second half of the year, more in line with our Investor Day model, so don't expect much growth.

Matt Ramsay, Analyst

Thanks. My second question, encouraging to hear the momentum in the design win pipeline, particularly with storage. There's a near-term correction and storage face. However, can you help me square that with OpEx cuts you’ve made? Is this the time to invest or reduce? Any clarity would be helpful.

Matt Murphy, President and CEO

Absolutely. We do our strategic planning annually, usually in August or September. In doing so this time around, we recognized significant market changes in conditions. Consequently, we made a number of decisions on programs where the sponsorship was lacking or where previous projections were incorrect. With the rise in AI types of products, we’ve aggressively repositioned our spending footprint, and we’re exiting the year with a lower OpEx profile. It's appropriate for conditions now, without sacrificing growth opportunities. We believe that our revenue will show strong growth and a great drop through for profit as a result.

Ashish Saran, Senior VP of Investor Relations

Can you get the last question please?

Matt Murphy, President and CEO

That's the last question, yes.

Operator, Operator

The last question today will come from Srini Pajjuri with Raymond James. Please go ahead.

Srini Pajjuri, Analyst

Yes, thank you. Thanks for squeezing me in. Matt, your enterprise networking business and carrier have grown strongly in the last couple of years. Yet you're guiding for a softer second half. I'm curious as we go into next year; it looks like 5G has reached saturation while macroeconomic conditions remain uncertain. What are your expectations for growth in those segments as we enter H1 next year?

Matt Murphy, President and CEO

Sure. Yes, Srini. Your assumptions are correct. Our enterprise business has notably outperformed expectations. But given the macro conditions and the previous phenomenal growth, we do expect softness next year. When examining the enterprise market, expect it to be flattish year-on-year; it’s doing well, but won't grow drastically. The carrier side has seen robust 5G deployment globally; while that continues, there are reasonable expectations for normalized demand as we see inventory digestion. However, content gains will help offset some overall market trends. Great. I think we can wrap now. A lot of questions, but we want to answer as many as possible. We appreciate everyone’s participation today. I look forward to talking more in subsequent calls. Thank you.

Operator, Operator

This concludes our question-and-answer session. Thank you for attending today's presentation. You may now disconnect.