CEOInterviews.AI
Start App
Get an email when Matthew Murphy gives a new interview
One email per new interview. Unsubscribe in one click.
Matthew Murphy
Chief Executive Officer & Chairman, Marvell Technology

Marvell Technology Investor Day 2026 | AI Custom Silicon & Electro-Optics | Oct 06, 2026

📅 Oct 09, 2026 i101 214 MIN 26 VIEWS 313 SEGMENTS · 10 SPEAKERS
Marvell Technology Investor Day 2026 | AI Custom Silicon & Electro-Optics | Oct 06, 2026. Twitter - https://x.com/i101yt If you find our work useful, please support us by purchasing a Super Thanks— it truly helps us a lot. #earningscall #StockMarketNews #conferenceCall Earnings Call | Earnings Conference Call | Earnings concall | concall | quarterly results | Stock News | Full Year results | Fiscal Year results | investment news | stock latest news | Annual Meeting of Shareholders | Annual Meeting of Unitholders | Special and Annual Meeting of Shareholders | AGM | Annual General Meeting...

What Matthew Murphy said

Written from the verified transcript and checked against it. Every figure links to the moment it was said.

Matthew Murphy, CEO of Marvell Technology, presented a long-term growth plan at the 2026 Investor Day, emphasizing the company's transformation into a connectivity-first data infrastructure leader. He highlighted Marvell's $20 billion investment over the past decade, including $10 billion in M&A and $10 billion in organic R&D. Murphy raised the FY29 custom revenue outlook to more than $12 billion, up from the prior $10 billion target, driven by XPU attach opportunities. He projected total company revenue of approximately $20 billion in FY28, with data center revenue growing about 80%. For FY31, he outlined a path to $70-$90 billion in revenue, with data center growing at 55-70% annually. He also detailed a $400 billion total addressable market by 2030, up 4x from the last update, with interconnect growing at 65% annually. Murphy emphasized that the growth is driven by existing customer relationships and design wins, not new home runs.

Key takeaways

  1. Raised FY29 custom revenue outlook to more than $12 billion, up from prior $10 billion target.
  2. Expects FY28 total company revenue of approximately $20 billion, with data center growing about 80%.
  3. Sees path to $70-$90 billion revenue in FY31, with data center growing 55-70% annually.
  4. Total addressable market by 2030 is approximately $400 billion, up 4x from last update.
  5. Interconnect market growing at roughly 65% annually, reaching approximately $65 billion by 2030.

Numbers and commitments

FigureWhat it refers toTypeAt
$20 billion Total investment in Marvell platform over last 10 years commitment 8:06
$8.2 billion FY26 total company revenue metric 10:55
75% Share of FY26 revenue from data center metric 10:55
$12 billion FY29 custom revenue outlook guidance 17:23
$3 trillion Total data center capex by 2030 guidance 18:42
35% Compound annual growth rate of data center capex from CY25 to CY30 metric 18:42
$400 billion Total addressable market for Marvell by 2030 guidance 19:16
$20 billion FY28 total company revenue outlook guidance 22:25
80% Expected FY28 data center revenue growth guidance 22:25
$70 to $90 billion FY31 total company revenue range guidance 26:46

Chapters

  1. 0:00Opening and company history
  2. 6:51Investment and platform building
  3. 8:39AI workloads and connectivity
  4. 11:53Business segments overview
  5. 15:48Custom silicon and XPU attach
  6. 18:42Market opportunity and TAM
  7. 21:58Revenue outlook and growth targets
  8. 28:01Bottoms-up plan by business
  9. 43:59Engineering and talent scaling
  10. 48:24Supply chain and capacity

Questions asked in this interview

8
  1. 2:03:18So again this group introduced the concept of KV cash. So what is KB Cash?
  2. 3:07:00Like, you know, when you're talking to hyperscale, you know, CEOs, is this something that's like coming up?
  3. 3:08:35Is that more DSP share, more DSP volume, NPO, CPO, any kind of breakdown that you could give us?
  4. 3:10:53Would it be in custom attach, optics, and how are you thinking about your relationship?
  5. 3:13:23And what's different this cycle versus, you know, last kind of super cycles?
  6. 3:16:37Are you just basically giving us your guidance on the bottom? Are you being conservative?
  7. 3:19:09And I'm just curious, you threw in micropixel, is that a scale-up opportunity, and where are you with that?
  8. 3:25:53And second question, because of maybe the faster time to market and because more Marvell IP per XPU attach, is the gross margin profile relatively richer?
Matthew Murphy 0:00 ↗
Please welcome to the stage, Marvell's Chief Executive Officer, Matt Murphy.
All good morning, everybody. You good? Up here, you come on. All right, so welcome to Marvell 2026. Back, new stories, all here. I understand a few of you, and you very well. You got married here. Another at lunch said to me, you know, good call on the location at the Peabody. The Peabody is a great hotel.
And look, I agree. From April 2024 to now, it's definitely been a bull market for semiconductors, hasn't it? Excited to be back here today to show you what the next of Marvell looks like.
All right, let's go back in time. This is from Marvell's first ever investor day. There, I can recognize many of you from the backs of your heads. You find yourself. By the way, I'm not sure who this young guy is on stage, but hey, handsome, isn't he? This is what 10 years as a public company CEO will do to you.
I remember the first question I got that investor day. Chris Roll from, I think, here today, very seriously asked me first, what company is this? Is this Marvell? The reason he asked that question is because at that time Marvell had never communicated ever to Wall Street what the mission, vision was, where the company was going. The company, 2000 people, back then, consumer products, other devices.
Look at what we're talking about. Look at the explosion of data and bandwidth, compute moving to the end, security, the need for power efficiency. This was in March 2017, and we were talking about the data.
Marvell, 22 years, our products, one leading analog company, things, business exposure, every market, every major tech trend. One third thing I learned from that experience was that the market that you sell into is what matters. It's not just about having the best chip, the best part, the best widget. Market exposure and long-term advantage with that market, that's what matters.
When I came to Marvell, I had a strong point of view about where the semiconductor industry was going and where I wanted to take the company. And that point of view was based on data infrastructure in the cloud, even though at that time most of Marvell's revenue was tied to consumer. Back then, less than 10% of our revenue was actually in data center.
And that's what led us to develop Marvell's mission statement. Here it is. So we wrote this in January 2017, about six months after I became CEO, and nearly 10 years later, this mission has not changed. It's exactly the same words from January 2017. Now, this might sound like motherhood and apple pie today: make high-performance chips, move data, store data, process data, compute, networking, storage, all those types of things.
10 years ago, this was unique. There was no category back then. Data infrastructure, the whole semiconductor market was defined very differently back then. I remember recently I was in a meeting with one of our senior fellows, a long-time Marvell employee, was presenting to a big engineering group, and he said, I remember when you first presented this mission statement to us when we were sitting around at Marvell, and I'd already been at Marvell for 15 years, and sitting there watching him and thinking, I don't know what he's talking about.
Now, and then he stood up, he said, now 10 years later, I see it. We have actually all the pieces now: moving data, storing data, processing data, all the key elements. We didn't have those back then. We do now. That was the vision over the decade.
2018, years ago, we were talking about AI as a major driver of data center and how AI would both training and inference. Long time ago, GPT. AI, 2020, we were talking about the shift from application-specific workloads to data-centric workloads with functions like networking, security, and storage increasingly moving into dedicated infrastructure silicon.
2021, we brought into the era of cloud-optimized silicon. This was years ago, and we said that compute, storage, networking, security, and optics were all becoming increasingly purpose-built. And if you look at what we are describing,
2024, we knew we had front and back-end networks required AI, and how optical connectivity would be a major driver of the next wave of AI infrastructure. And then finally, last summer in June, we showed you the breadth of our custom capability.
June, we had a consistent point of view where the industry was going, and we invested aggressively ahead of these transitions. But you can have a vision, you know, you can see the shifts, but that is not enough. We had to build this company and the capability to capitalize on the opportunity.
So what did we do? The mission in 2017 became the guiding principle and allocated capital, architected Marvell over the last decade. In some areas, we invested in strong technology and expertise that we already had inside Marvell, a very talented team when I joined in engineering. In other areas, though, we pursued very strategic and, quite frankly, very ambitious M&A to add critical capabilities and domain expertise in a number of new important areas where we didn't have that capability.
At the same time, though, we didn't just buy things. We narrowed our focus through divestitures, which created significant shareholder value. We received premium multiples for these businesses that we sold, and it allowed us to hone our focus on the data center.
I got to tell you, the real value of the acquisitions we made and those capabilities that we then gained came after those teams were inside Marvell and then part of our platform. We brought these teams together around one strategic direction and continued to invest aggressively in those businesses, and that allowed those teams to achieve far more than they ever could have on their own.
If you take a step back, we invested roughly $10 billion bringing those capabilities into Marvell, decades and decades of expertise. We also invested another roughly $10 billion organically behind the platform. This was rocket fuel for the roadmaps and helped establish the leadership positions we have today and that you're going to hear about. So that's roughly $20 billion of investment in the Marvell platform over the last 10 years. This has truly been a decade in the making.
The platform has actually arrived at the right moment. AI workloads are fundamentally changing the way that data moves through the network, moving beyond traditional training and reasoning models, moving beyond training into reasoning models, AI and inference now at scale. Those workloads are driving more and more data movement and diverse traffic patterns across AI infrastructure. That's creating greater demand on connectivity. Efficient data movement is becoming a fundamental constraint on AI infrastructure.
And this isn't just us saying it. This is what we hear from our customers. The world's largest hyperscalers are rethinking their network architectures. AI infrastructure scales across the industry. These changes are driving significantly more connectivity across the system. So take a moment and read it and look what they're saying. It's all about connectivity.
Now, if you look at the AI landscape over semis, there are three very large companies whose businesses are primarily driven by competing in compute. But compute is what they do really well, and we've seen growth and value creation in the market from the compute wave that's gone through. Then you have three large memory companies whose businesses are overwhelmingly concentrated at memory. That's what they do really well, and again, we've seen tremendous growth and value creation there as AI has driven.
Connectivity unlocks system-level performance, and Marvell is a connectivity-first company. That gives us a very different position in the ecosystem. We work with the leading compute companies, the leading memory companies, and our customers, all the world's largest cloud providers. In many ways, we're the enabler to scale. We have exposure across the ecosystem regardless of which compute or memory architecture is deployed.
Put some numbers behind that. FY26 last year, Marvell generated $8.2 billion in revenue. 75% of that was in data center, and we break that down, we're going to talk about today, interconnect, switch, and storage, and custom. Now, most of our custom revenue last year was in compute, and some of our other revenue is also in compute. Landing in compute, 70% connectivity. As we go forward, XPU attach ramps, more and more of the custom business will actually be part of our connectivity as well.
Coming back over, we'll walk you through these categories today and show you how the technology is evolving, how the market is expanding, and what that means for Marvell. And then later, we'll come back to these businesses and look at our growth expectations and the long-term opportunity across each one.
Right, so with interconnect, as these network architectures evolve, connectivity has to scale every distance in the system, from links between data centers to connections all the way inside the package. And these are variations of the same problem. Each distance represents a fundamentally different engineering challenge, requiring different technologies, different skill sets, and unique expertise. At longer distances, like scale across and scale out, the industry has already transitioned into optics. Marvell is a leader in those markets, and that's driving our interconnect.
Larger wall, that wall, to increase the distance that can reach. And that means connections inside the rack are starting to begin to transition to optics. That's entirely new for Marvell. This is CPO, co-packaged optics, becoming increasingly important, and that translates into the optical capabilities we've been developing for years, and you'll hear much more about this today.
In the longer term, the same physics continue to apply. As bandwidth keeps increasing, the transition to optics moves closer and closer to compute. Opportunity, Marvell's scale in, and is about that today.
On to switching. We established a strong position in scale-out switching with multiple customer wins and a business that will be well north of $1 billion next year. Now we're moving to scale-up. Accelerators require extremely high bandwidth, low latency connectivity. We're unique in supporting all the major scale-up protocols, including NVLink, UALink, and Ethernet, giving customers flexibility across their different architectures. As bandwidth increases, scale-up is becoming increasingly optical, an opportunity for Marvell, and you'll hear much more about that today as well.
Storage franchise, decades of compounding leadership, hard drivers over time. That technology base across a much broader range of storage and memory technologies today: HDD and NAND controllers, memory controllers, and CXL memory expansion, and best-in-class dense SRAM used in our custom products. What's important is that we have proven technology and decades of expertise across every layer in the storage and memory technology stack, and increasingly those same technologies are becoming the foundation for custom solutions that we design for hyperscale AI infrastructure. So customers can start with proven Marvell IP, work with us to optimize and customize it for their specific architecture.
Began as a leadership position, has become a platform. Unique architectures.
Started with the XPU, but the opportunity today is actually much larger. As XPUs continue to grow, they pull through a whole new set of functions around them: networking, storage, memory, security, and inference acceleration. All of these functions are attached to the XPU, and increasingly we're seeing them all move to custom solutions. That's what we call XPU attach.
With the XPU, you're typically talking about a traditional full custom ASIC model. XPU attach is a little different. These opportunities are more IP-led and semi-custom in nature, and that difference between full custom and semi-custom is actually important. I'm going to take a minute just to talk about it. In many cases, Marvell works with the customer on the architecture and takes the lead on the front-end design with the proven Marvell IP set at its core.
And that IP comes from technologies that we've already developed, shipped at scale, and proven in high-volume manufacturing. So it's proven. That model is gaining significant traction with customers.
We've announced multiple strategic agreements around shipping custom silicon to all of our hyperscale customers. We've grown this business organically from scratch over the last few years, and last year it was about $4.5 billion.
April 2024, we expected custom revenue to more than double in FY28, putting us on a path to more than $10 billion in FY29. Today, that is on track against that plan, and based on the opportunity we're seeing, we're raising our FY29 outlook for custom. We now expect custom revenue in FY29 to more than triple from FY28, reaching more than $12 billion, and putting us well above our prior $10 billion target. Now, this update to our expectation is primarily driven by the rapidly expanding XPU attach opportunity we see in front of us, including the strategic agreement that we announced in August.
AT&T, more is coming, and leading to multiple areas I just mentioned. That's what gives us confidence to raise the outlook from more than doubling to more than tripling. You digest that one for a minute.
All right, let's take a step back and look at the market opportunity. By 2030, we expect total data center capex to reach roughly $3 trillion, growing at about a 35% compound annual rate, starting from calendar 25 to calendar 30. And importantly, that assumes that the current rate of capex moderates over that period, moving closer to historical levels in the outer years. As you guys all know, we're growing faster right now than 35%. So in our model, we expect some normal deceleration to a more normal base case in the outer years.
Now, within that, accelerated compute represents roughly $1.6 trillion of that. We see approximately $385 billion data center TAM addressable by Marvell. And when you add in the opportunity in the rest of our business, communications and other, that's another approximately $15 billion. Put all together, we see approximately $400 billion total addressable market for Marvell by 2030, at a 45% compound rate from 2025 to 2030, faster than GDP.
And to put that in perspective, just so everybody knows, there's a lot of big numbers out there these days, but for us, the last time we laid out the TAM for the company, it was less than $100 billion. So this is like a 4x increase to our opportunity that's occurred. That's how fast the market is moving for us. And the growth, though, one thing is not concentrated. Let's go deeper and show you what's underneath the $400 billion.
Right, there's a lot of numbers up here, so let me just give you a minute to absorb them, but I'm going to start at the top and work my way down on the slide. So in interconnect, we see it growing at roughly 65% annually, reaching approximately $65 billion by 2030. Scale-out continues to grow, and new opportunities emerge in scale-up and scale across.
In switching, the market is growing roughly 40% annually, reaching approximately $85 billion. This is driven by growth in scale-out switching and the new emergence of scale-up switching. In custom, the market is expanding dramatically. We see it growing at roughly 55% annually, reaching approximately $235 billion by 2030, driven by the growth in XPUs and then the layering in emergence of XPU attach. Communications and other is approximately $15 billion. Put all together, you get $400 billion. I showed you a 45% annual rate to 2030. So these are all very healthy growth rates, but I want to point out interconnect stands out at roughly 65% compounded growth over the period. So as I said earlier, when you look at where the growth is coming from in the industry, it's connectivity.
This breadth gives the overall growth. All out.
In FY26, as I said, Marvell generated $8.2 billion. In August, we laid out approximately $15 billion, that's 50% year-over-year growth, showing that even as the company gets larger, our growth rate is actually accelerating.
Now today, we're raising that outlook again. We expect approximately $20 billion in total company revenue in FY28, representing about 67% year-over-year growth. And then within that, we expect data center revenue to grow approximately 80%.
Now, when we raised our FY28 outlook in August, that increase was driven by interconnect, scale-up optics, and switching. The additional billion of growth we're seeing for next year is being driven by the same trends. Scale-out optics continues to accelerate, scale-up optics continues to grow, and our switching portfolio is firing on all cylinders. So once again, it's connectivity leading the race.
Now I'm going to put this $20 billion in context for everybody. So you take a step back, when we were here in April 2024 on stage at the PE, the prior year we had just printed $2.2 billion data center revenue, which is our FY24. At the time, we told you we believe we could grow our data center business to $15 billion by FY29. That was going to be 7x growth in five years, and I think people had a hard time wrapping their head around it. And I know it was ambitious. I got it, it was ambitious.
A year ago in June, we had our custom AI event, and we basically said, look, the opportunity is developing faster than we thought. At that time, we raised our FY29 data center revenue from $15 billion to approximately $18 billion. In other words, the target we had said a year earlier was already moving higher.
Now I'm going to show you where we are today. $20 billion total company revenue we outlined for FY28, approximately $18 billion coming from data center. That means we now expect to reach the $18 billion data center target we set for FY29 almost a full year early. And that would represent roughly a 70% compounded annual growth for our data center business over that period from FY24 to FY28.
Okay, so stepping back for a second. So we put some ambitious targets in front of you over the last few years, and the Marvell team has executed. We've gone from $2.2 billion data center revenue in FY24, and we've got our plan to go to $18 billion next year. We're hitting our goals, and we do so earlier than planned.
With that perspective in mind, now let me show you how we're thinking about the future here.
For the purposes of our investor day today, we're going to set our growth plan starting with FY26 last year and outline the opportunities for FY31. This matches the TAM framework I showed you earlier from CY25 to CY30.
So FY26, $8.2 billion, $6.1 billion was data center, and I showed you just now the plan to get to $18 billion data center revenue in fiscal 28 with $20 billion total company revenue. So during that period, FY26 to 28, that's about 70% growth in our data center business compounded. And if you remember, that's about the same growth that we were looking at from FY24 to FY28. So basically, we've been growing pretty consistently since calendar 23 in our data center business.
Now for FY31, we have a bottoms-up plan that I'm going to walk you through shortly. This is not just top-down. I'm going to walk you through the bottoms-up in just a moment, but let's start at a high level. So looking forward from FY28, you assume the $20 billion total revenue plan, and you roll up by customer and product and design win, and you layer it up. We see a path to continue growing our data center revenue at roughly 70% annually through FY31, despite the moderating assumptions. But look, it's a dynamic market. So today, when we talk about our revenue goals, we're going to give you ranges. And so at the top end, 70%, low end, 55%.
55 to 70% growth range, that implies data center revenue of approximately $67 billion to $87 billion in FY31. Now, if you add in $2 billion for comms and other, you get a range of $70 to $90 billion in revenue for Marvell in FY31. You digest that for a minute.
I understand the absolute numbers are large, but the growth rates are below what we're already delivering today.
So with those numbers in mind, let's take a step back and look at the company as a whole. $8.2 billion revenue last year, plan to deliver 70% in five years, 80% in six years. That would represent 10x growth over a five-year period and 55 to 60% compounded annual growth over the same period. That's the growth framework behind the long-term outlook that we're setting today. A very strong growth plan, but what gives us conviction is not the top-down view, it's the bottoms-up.
All right, let me now walk you through what underpins the bottoms-up. Here we go.
So I'm going to walk you through one by one each business and what gives us conviction in our outlook. So with interconnect, scale-out has been incredible. In the last few years, that has been roughly doubling each year. In this model, we actually assume that that growth rate moderates over time. At the same time, though, I want to say, keep in mind, we're just beginning our transition to 1.6T in volume in the market, and we have 3.2T on top of that. Scale-out, which has been growing, very, very strong future.
And in scale across, we already have a very strong position with new use cases for that market that we're going to hear about. And scale-up optics, which is part of this, that's all in front of us. We have virtually no revenue in this area today, with strong growth beginning next year, and we already have significant traction. So on interconnect, main, scale-out, high-growth market, and then execute on the scale opportunity in front of us. That's what we've got to do.
If we execute that plan, that's $7 billion at the midpoint in FY31.
Now let's look at switching and storage. Storage is already a billion-plus business, and we assume that continues. In switching, as we're going to be well north of a billion next year with our 51.2T product just ramping in first year of production, and our 100T product still in front of us.
Based on the customers and programs we already have, we see scale-out switching becoming a multi-billion business over time. On top of that is an entirely new opportunity in scale-up switching. We already have deep, multiple, multiple deep customer engagements in this area today, and as they materialize, those are going to provide another meaningful layer of growth in our switching business. When you put those pieces together, $10 billion.
Now, custom actually might be the simplest of all. I just showed you we expect this business will reach more than $12 billion in FY29. Our existing programs are on track, and we're assuming that those programs ramp into that plan and continue to grow at a normal rate from there.
On top of that, though, is the rapidly expanding attach opportunity supported by the design wins and our strategic customer agreements. And just to be clear, we've taken a very pragmatic view in judging the total opportunity covered by the large agreement we recently disclosed.
So if you start with more than $12 billion in FY29, the growth we're assuming from FY29 to FY31 is roughly in line with the market. We think that's a very reasonable assumption given the programs and the customer positions we have today, and that gets us to approximately $30 billion in FY31.
By FY31, we expect custom at that time to be much more balanced between XPU and XPU attach sockets, with no single socket concentration.
Finally comes communications. We're assuming we maintain our position with very modest growth over the period, approximately $15 billion.
Put all that together, that's the bottoms-up case to $90 billion in FY31. These businesses, I think, very reasonable. Have very exciting. You think about it, let's take a look at the ones.
By 2030, we have approximately $400 billion of addressable markets for Marvell, up 4x from the last time we updated you. FY28 next year, we expect approximately $20 billion in revenue, up significantly, driven by connectivity.
In custom, we expect revenue to more than triple, reaching more than $12 billion, putting us well above our prior target. Looking out to FY31, we see a path to $70 to $90 billion in annual revenue, with $80 billion at the midpoint.
We have deep, trusted relationships together. They see Marvell as their primary connectivity provider and deeply embedded across their entire architecture. So although the overall market by hyperscale, Marvell's portfolio of products being delivered across this customer base, and no socket controlling our destiny.
So when you see that kind of revenue growth, you're probably already doing the math. In fact, I see a lot of heads down in the audience on your PCs. Look up, stay with me. Hold on, we're almost there. What you're wondering is what the earnings power is for such a company at such scale.
I'm not going to give you that right now. I'm not going to do it. We have a new CFO at Marvell. He's outstanding, and his name is Dan Durn. He's been with us a long time, but him for a very long time. This is his first investor day with Marvell, and I don't want to steal his thunder. He'll come back later and show you growth, canvas over financials, gross margin, operating margin, EPS, the whole nine yards. So for now, I've given everybody the revenue as a starting point.
Before I wrap up, Marvell, that makes all this happen. What you have, and commitment all around. You've been here for a little over years, and never been more excited about the technology, deep, trusted relationships, important in the industry and foundations. For many years, we have an opportunity looking in front of us that is larger than anything we've ever seen before. I truly believe the best is still in front of us. So thank you so much, everybody, for joining us today, and I'll talk to you soon.
It's great to be here with all of you today talking about this. This summer, I also celebrated my anniversary and had a blast, but I'm here to tell you that our future is exciting in front of us. We've spent all these years building this platform for this moment, and I'm going to talk about how we've been doing that and how we've been planning to scale this company.
I'll talk about it along these three pillars. I'll start off talking about our customers, where are we focused and how do we win. I'll talk about our core engineering and our systems and how we set up to scale to deliver for our customers. And I'll talk about the relationships and long-term capacity in the supply chain.
Last year, our AI event showed that analysts forecast over a trillion in total data center capex by calendar 2028, and that was showing basically a modest 20% growth going forward after a really strong growth year in 2025. I remember when we put the trillion dollars up there, it kind of stuck in my mouth a little bit. A trillion dollars, like that's a lot of money. Is this a little bit crazy? And then you fast forward a year and a half, and it wasn't crazy at all. Actually, we ended 2025 over $700 billion, and now we're hitting over a trillion this year already. Clearly, this is a fast-moving market and one that's hard to forecast, and it's growing faster than anybody expected. And as Matt said, analysts are now forecasting over a trillion by calendar 30.
And that's after a much faster 25, 26, and even 27. We're still expecting capex to moderate over time to 20% after that. And this is the capex envelope that all of the numbers we've shared with you today, total market, the billion, all the revenue targets, that's the assumption, the total data center.
So if you look inside that data center capex, the top 10 hyperscale customers are actually driving over 60% of that last year. As large as they are, they actually grow their spend even faster, reaching more than 70% of the total by 30. That's, if you do the math, that means that these top 10 companies are spending over $100 billion each. And if you look within that, the top four US hyperscalers alone were more than half of total data center capex last year. So you can see that any customer strategy in this market really needs to be hyper-focused on those companies.
Large economic, are technology leaders, okay, divided through our, and architected this company to the perfect hyperscalers.
Large and sophisticated operators. They've been designing, deploying complex infrastructure for decades. They actually define what it means to operate at scale. And they've seen in the real world how these workloads interact with the underlying architecture, and they have a point of view for how to optimize that architecture for the future evolving workloads.
What they need is a broad portfolio of connectivity technologies to stitch it all together. And at every distance, Marvell has a full portfolio of capabilities.
Our teams are able to sit together and make bandwidth, reach, power, and manufacturing scale trade-off decisions, and literally co-architect future cloud data centers.
So at Marvell, we are laser-focused on these top 10 hyperscalers. In fact, within that top 10, we can hit the majority of the goals that Matt laid out with the top US hyperscalers alone. We aren't counting on some major new customer win or a new home run design win. We need to focus on this, find that set of customers, and execute.
Journey with off-the-shelf system, a branded storage switch, optical module with Marvell silicon inside. This is how these hyperscalers get experience operating networks with Marvell chips. The next step is often to build a custom system with Marvell's standard silicon inside, optimized for their overall network design.
Actually, will often ask for a custom feature, or maybe they want to put some of their IP in an otherwise standard chip, adding their secret sauce and adapting it to their own network topology. The last step is a full custom engagement, where all the IP that they've been using those standard products for all those years is brought to market, a custom piece of silicon designed for that customer.
But we're not selling custom silicon to customers who aren't also buying other parts of our platform. We're not a vendor. Give us a call if you want a custom chip. We're not a backend services company with layout engineers hanging around, a manufacturing company if you need some wafers. We are a connectivity leader building partnerships with the largest hyperscalers in the world. In the end, this is a full platform engagement, merchant all the way through custom.
Actually, took a look at that fiscal 31 forecast that I was talking about, and I broke it down by product and by customer. We're talking about hundreds of products across this hyperscale base, just the top 10. If I look into the top four, more diverse hyperscale engagement.
We are building strategic engagements. We actually published strategic agreements with three of the five largest companies in the world by market cap. Each of these agreements is totally unique and tailored to the individual situation, built on top of years of collaboration.
Our first announcement in 2024, an agreement covering custom AI products and a broad suite of networking and connectivity products, including switching, data center interconnect, and optics. Last year, we added our packaged optics technology to the agreement after our earlier this year, we announced a strategic agreement with NVIDIA. This again was a very broad and very strategic agreement covering optics and photonics, scale-up networking with NVLink Fusion, and AI RAM.
Most recently, the announcement was with Google, and this outlined a total of $40 billion opportunity over and a half years for custom products attached to the TPU ecosystem.
That's additional and separate from all the connectivity business that we do together today.
All these partnerships are multi-year in nature, and they all hinge on trust and execution and delivering value for both sides. These agreements are a natural evolution of our partnership. Usually, when we've already been working together for years at or nearing a billion a year in business together, with a rapidly expanding design win funnel, that's typically when these sorts of discussions start. How do we move towards a win-win partnership, aligning interests on both sides? So we are very strategic with how we approach these agreements.
Now, let's talk about how strategically we approached scaling our engineering systems and organization and the company.
We have been preparing for this growth for a long time. We've been consistently investing ahead of our revenue growth. Just take an example. Look at fiscal 24, where Matt mentioned that we had done a data center repeat. We spent a billion and a half in R&D, which was into the data center market, which means we were investing well north of 50% of our revenue back into R&D to drive Marvell's growth, and it's worked. You take last quarter, fiscal Q2 of 2027, and you annualize it. We quadrupled our revenue in data center, 8.7. We are investing heavily in R&D, Marvell growth.
And we're making these investments in strategic R&D centers around the world. We're not just hiring from our competitors. These are very unique skill sets, very hard to hire people. So we're doing it right. We're building a long-term talent pipeline at Marvell. We're working with the local universities in countries where we have these R&D centers, recruiting, helping to influence the curriculum to make that talent comes out with the right skills that Marvell needs. And we're funding scholarships. More than 200 students are funded by Marvell scholarships every single year across more than 50 universities around the world.
That brings us to our internship program, which we started a decade ago, grown it every year. It's now more than 500 students annually. The vast majority of them convert to full Marvell employees over time. Marvell has become a premier talent company, people actively seek out and want to come and work for.
And recruiting talent is only half the story. When they come to Marvell, we arm them with the best toolset in the semiconductor industry. We have embraced and adopted AI tools throughout our design flows. We've deployed an ever-increasing pool of agents for common tasks, automating design, simulation tools. Marvell design tool in-house with our own data, our own decades of experience. We started over years ago and grown dramatically since. It now becomes the foundation of Marvell's AI-native workflow.
And the results are dramatic. We now see engineers with years of experience delivering the results that it used to take an engineer with 15 years of experience on a traditional tool. And it's not just making engineers more productive sooner. We're actually accelerating and shrinking the overall design cycle. We're shrinking the time from concept to shipping silicon, allowing us to iterate more quickly on designs, find issues early, and adapt quickly. This is really important in the age of ever-shrinking hyperscale design cycles.
And none of this would be possible if we just kept doing it the way we always had. And that includes moving to the cloud. One of the advantages of working with the largest cloud operators in the world is we see how differentiated and performant their technology stack is. So we went all in early, announcing our first deal with AWS in 2023 to move our EDA into the cloud. We invested ahead of the curve when it wasn't the obvious choice. In fact, my IP team told me at the time, "Chris, this is going to be more expensive if we do it this way." But I knew where Marvell was going to scale without the cloud.
So we've worked closely with our entire toolset, all of our EDA vendors, and our cloud to optimize the complete flow and move it all into the cloud. Today, more than 80% of our total compute infrastructure is in the cloud.
So you might be wondering, how does this matter, and does this actually help? Let me give you an example. Last year, we won a design win that we didn't have server provision for. Within 60 days, we had 30,000 new servers over a period of time, provisioned, engineers productive and working in the cloud. 90 days later, that was up to 70,000. I can't even imagine how this would have been possible if I had people running around looking for servers and data center space to do it today. Completely impossible.
And by the way, that same chip we were able to tape out in less than months. It's in the lab today and being delivered to our customers this year.
Now, let's talk about supply chain, a critical part of Marvell's scaling journey.
Everything I said about focusing on customers, it's about 10 times in the supply chain.
It's all about trust. These companies are buying buildings, hiring engineers, and buying equipment for products that are going to be delivered 5 years from now. So they need to trust you. They need to know you're going to be there. They need to know your demand is real.
Our strategy is to focus on a small number of critical suppliers who are the best in the world at what they do. These relationships often go back to the founders of these companies. When I took over operations in early 2021 and I went on a sales campaign, we could see where all this was going. We built relationships deep into the supply tree, not just foundry, but substrate, packaging. You name it, we needed a complete set of relationships all the way through.
And we told them our story. Where do all these chips fit? Why are they critical? Why should they bet on infrastructure at a time that they were seeing unprecedented demand from stay-at-home driven consumer markets?
We showed them how diverse our platform is, how when anyone wins, Marvell wins. Suppliers are very sophisticated, very experienced in this market. They've been burned many times, and they really know a good customer.
So these trusted relationships are also trusting what we said we were going to do, year in and out. So we made the decision again to give our suppliers forecast years. Who does that? Show you how you need to do that. World-class visibility.
Every piece of data at fingertips today. Every customer forecast, what are the products, how it evolved over the last quarters, what are all the pieces in the supply chain, how much capacity we reserved. This isn't just an MRP feed. We have architectural insights across the staff.
You know that 31 for billion that put up, that might have been a surprise to some of you, not to our suppliers.
And I visited these CEOs, shown them our forecasts, asked them if they could get us there, shook hands, and we are now hand in hand to deliver.
And capital investment reservation, this is where you start. You can't show up with a checkbook and say, how much capacity can I get you? They just tell you to go away. You needed to have built the trust, shown them that you have a track record, and then yes, we are investing ahead of our growth to reserve the capacity. This is to cement a relationship and put skin in the game for our forecast.
This is how you build a supply chain.
As you can see at Marvell, we are not short-term thinkers. We are building the long term. We have deep, trusted relationships with the largest customers in the world. We've built a talent pipeline, tools to feed our growth and keep the company scaling. We have a supply chain that we built based on trust with the critical suppliers, the best in the world.
My team and I are ready to deliver that billion that I laid out. We've grown the company 4x in the last years. We've got all the demand and the supply I need to hit the billion next year. We can certainly 4x from that with the wind at our back. We don't need any new home run design wins or any new sockets. We have what we need today. We need to execute, deliver them to market, and put them into production. And I'm here to tell you that my team and I are going to deliver that forecast. Thank you.
Morning, everyone. I'm very excited and delighted to be here today.
Every dollar that gets spent in the next 5 years by the AI hyperscale data centers and our customers will run into the same four constraints or walls: bandwidth, latency, power or energy efficiency, and data center reach or the distance data has to travel.
Before I get into that, let me tell you in the next 15 minutes why Marvell is best positioned to those barriers and those challenges.
But I want to give you my origin story. 8 years ago, Matt invited me to dinner exactly 8 years ago, and he laid out his vision for how he is going to transform Marvell to be a data infrastructure company. This was even before AI.
He told me about the opportunity, the technology, and to take it to the next level. I signed up to the challenge, and my journey at Marvell began in Feb 2019.
In 2020, go back to the investor day, I presented a plan to transform Marvell to be a leading provider of data infrastructure.
On the process side, we had to because Marvell was really a process follower. We were on technologies one generation or two generations behind, but impacts such that it was already energy efficient, but a fast follower. It had architecture to do Wi-Fi and storage products about 200 square mm, but we had to pivot the team and the technologies to do reticle-size chips.
Then came design and IP, analog mixed-signal IP that needed to be for the data infrastructure market, had to be high-speed, high-bandwidth serdes. We already had 28G and 56G that had to transition to higher data rate, more like 100G, 200G at advanced nodes. And let's see where we ended in 2026. And I'll go from the process technology.
This was a very important challenge. Matt wanted me to get the team to the next node. We were on 12 and 16 nanometer, and this was a big challenge. We went to 7 and went to 5 nanometer.
So not only did we succeed in the years 2020 and 2021, we did it with leading-edge process and first to node with all the IP shields with our foundry partners at 3 nanometer, at 2 nanometer, and we are already on 1.4, 14A. And not only that, we have transitioned from less than 50% A0 to production.
What does A0 to production mean? A0 to production means you need to get it right the first time. Measure twice, cut once. But we measure many times, and the mask gets cut once. So we deliver market to the company.
Again, next architecture, we want to deliver reticle-size dice, the ones that will, and I'm going to talk to you about 200 square mm, and the reticle-size die is about 856 to 858. This is defined by the lithography limit and process technology that we can do.
And here to say we have industry-first monolithic 100 square mm, nobody has that. They will talk all about it, and we have about greater than 10 products at 600 square mm and beyond shipping in volume and production. And Will is going to talk about his XPU and XPU attach where we have accomplished that. So we have completely transitioned not only on process technology but also architecture.
And when you take a look at large, why is it hard? It's hard because let's say you have an 856 square mm, you may get maybe 30 or 40 die, and you need to get it first time right because you need to have the methodology and the team to execute flawlessly to ensure that they account for defect density, signal integrity, and many other laws of physics in the silicon to optimize these architectures and get it first time right.
Now let's talk about design. We were best-in-class IP portfolio, but fast follower. But we had to invest significantly in our technology to get leading-edge IP at 28G, 56G, 100G, 200G, and 400G. And just in OFC, we have demonstrated in silicon. This is just not theory. It is actually done in multiple process technologies, and we have industry's best-in-class serdes at all these data rates. So we have delivered and proven at every step, and this is foundational for Marvell.
Let's talk about the dynamic, the latency, the reach, and what is optimizing for total cost of ownership is really about delivering the maximum tokens per megawatt.
But you also want to have the performance, the interactivity of the user. Whenever a query is served, you need responses immediately on the tokens, whether it is on research, on LLM, or even other applications that people are interacting with the LLMs.
So what happens here is AI is no longer a silicon problem. It's a system, a connectivity challenge that Matt talked about. Now, in order to increase performance, we need to take a look at all the constraints, and let me talk about the walls that you will run into.
Starting with the first one, the memory wall.
The best way to think about it is an example of the human brain. The plasticity of the human brain gives immediate access to memory with the compute. So, let's say for short-term, you will remember what you had for breakfast today, or from long-term memory, you will know what you studied in your university. So, in the ideal world, you want infinite memories, uniform latency, and a very high bandwidth interaction.
Now, how is that achieved in the real world? In the real world, the memory wall is because of the hierarchy of memory that exists in the compute architectures and system architectures. XPU has a distance to get to the HBM memory, or when you have a pool of accelerators or XPUs, they need to access not only HBM but pools of memory. So, memory wall is a real bottleneck for performance for AI infrastructure scaling.
And how is Marvell positioned to take advantage of that? Memory wall as a shared resource, we have multiple technologies. We have controllers that can spread and reduce the latency for memory access, Compute Express Link technologies in the CXL family for memory expansion and pooling, again bringing memory as close to the compute as possible, photonic interconnects for memory disaggregation.
And they will talk to you all about those applications. And not only that, inferencing accelerators and the performance needs very high density, highly capable ramps very close to the silicon, and we will go into further details about that. So we are connecting systems using all the technologies is what Marvell is excellent at, bringing memory closer to scale-up fabric from forever, from all the history that we have had with these products.
Now let's take a look at the next wall. It is the reach or the power wall. Matt went into considerable detail on this. But what I'm here to tell you is the power wall is extending all the way from clusters of AI compute to coming to rack and tray.
Recent transition, optical, and the scale-up optics is here and now. So solving the scale-up bottleneck is the biggest gain we can drive. Will is going to talk about the scale-up optics, and scale-out and scale across. She and Russ are going to go into further more detail.
Now let's take a look at how Marvell is positioned to bring scale-up optics. We have the silicon on the silicon to enable the scale-up optics revolution.
If you just take a look at the entire gamut of technologies we have at our disposal, we were already leaders in different modulation schemes. Optical modulation schemes are very important for getting the reach, bandwidth, distance.
and the energy efficiency that every application needs. Very different, it's not a one-size-fits-all. So you have Mach-Zehnder modulation for distance greater than 20 km plus. Not only that, for within the silicon, if you want different modulation for highest energy efficiency, you have microring modulators, different modulation schemes. She is going to talk about the electro-absorption modulation, different modulation schemes, and scalable interconnect have new micro-emitter technologies like micro-LED and micro-pixel. She is going to talk all about it. So what is this actually telling you? What it is telling you is data movement challenges for those constraints require multiple technologies. So this is the core business Marvell is making, and we have the industry's largest optical talent. Chris went into how we have enabled talent and building these, and this is just a decade in the making. So we have the most brilliant optical engineers making all these technologies happen at scale, proven in silicon. Now let's take a look at the common technology.
reach because ultimately to solve those constraints it is not just about compute, it is about connectivity because if the connectivity is not solved, intelligence or access to compute means stranded compute, unused intelligence, not usable intelligence. So let's take a look at what are we doing on the process technology.
It is the table stakes to play in the market. However, we also have advances in silicon germanium, indium phosphide to implement and produce highly capable TIA and driver. She is going to talk all about how this is a key core part of its portfolio. Silicon photonics, this is where scale is being enabled, including distances from 2 km to 1000 km. Russ is going to talk all about it. Advanced material science research, lithium niobate. We work with advanced foundries across the board to bring technologies of the future, and this is here and now. Five years later, we are taking a look at what's going to be for high bandwidth energy plasmonics.
that is also invested in today so that we can have product 3 to 5 years out. So this is about technology enablement even before products are released 3 to 5 years out. So we are fully integrated, process agnostic, boundary ready, and future proof. Now, analog mixed signal is the heart of the data movement challenge. Why? Because high bandwidth and low latency requires 30, and it's a very critical technology. And if you take a look at what we have today, we have demonstrated and shipped high speed, whether it is optical or electrical, at 100G, 200G, and 400G has already been enabled in products which will be shipping in volume next year. That's 8x growth in data rate in 6 years. And SerDes is just not the only technology that is important for delivering on our products for XPU and XPU attach because there are more than 10 inside the package. You need highly capable die-to-die interfaces, and you might have heard pJ/bit as a currency for high energy efficiency. We have advanced die-to-die interfaces to enable communication and data movement between these dies. And bringing it all together at the architecture level, you use these so that there are important constraints to be understood at the system level or the tray level or the package level when we have to connect XPU devices to the XPU. That requires advanced system-level simulation and technologies that we already have built and enabled and shipped in volume. So Marvell's 30-year leadership in advanced mixed signal is in here now, proven in volume for every reach. Let's take a look at the most important technology also in not only compute innovation about J 2025 in the 2D package. Very simple to do. Then it went to 2.5D, wherein you have multiple dies on chip, on wafer, on any other silicon interposer material. But now where we are seeing die-on-die, bringing multiple layers, 12 all in critical because these substrates are greater than 100 mm on the side, going 150 mm on the side, and some of the samples I'm going to show you. Not only that, because of the wafer scaling comes into the package, we also innovate on co-packaged optics. So full integration and design to scale and volume, and these are the capabilities that you need in every single product that we ship. So I have talked about a multitude of technologies. Easy to say, well, we have, easy for a company to say, well, we are excellent at silicon germanium or we are excellent at silicon photonics. But what I'm here to tell you is it requires just not one technology. It requires advances in packaging, all the capabilities in boundary, analog mixed signal, all under one roof. And that's how we are well positioned to capitalize on the opportunities that Matt talked about. This is foundational, and this is technology at its very best and bleeding edge. Industry-leading core technology across all these means it's process and packaging IP and architecture and system-level solutions. And you will hear the details from all my colleagues, Russ, Dave, and she. Now it's proven execution and deployed at scale. And to put it all together, let me start by saying it's just physics. I know when just physics, physics doesn't negotiate with strategy. The company that solves the physics problem of bandwidth, latency, reach, and power is the one that is best positioned to go after the opportunity that Matt outlines today. And I am here to tell you that the team that we have built at Marvell is exceptional. Marvell, industry-leading solution. Are you all up? Thank you. Wouldly. 15 minutes.
Will please, as possible again. Willly, please be seated as. PR will please. Willly, please be seated as soon as possible. Please. Welcome, Vice President and Manager, Connectivity.
Morning everyone. Welcome from the break. That Marvell is a connectivity-first company. I'm thrilled to be here to talk about connectivity-first. Just like Matt and Chris, this year marked my 10-year anniversary at Marvell. Although that 10 years, I spent five years at Inphi before acquisition. Today I want to show you what we actually mean when we say Marvell is a connectivity-first company and why that is becoming more critical as the infrastructure scales. Let's start with what connectivity-first looks like in practice. Marvell has the broadest connectivity portfolio. We organize our portfolio into four categories: scale across, scale out, scale up, and scale in. And together they cover every distance an AI signal has to travel. That is 1,000 km between the data centers down to 10 cm inside the package. At each of these distances, Marvell has a solution. Module and gear for the longest reach, TIA, drivers, and AECs for ACC, NPO, and CPO for the rack. And today I'm also going to share more about what we are doing with optical I/O chiplet using next-generation sources as we bring optics inside the package. No other companies cover this entire stack. Our competitors may try a piece of it, but Marvell offers all of it. That is the breadth, exactly what lets us support our customers' architecture with whatever implementation they go with next. In my presentation, I will focus on scale out and scale in. Russ will talk about scale across, and Dave will talk about scale up. Before we talk about the product and business, I want to take a moment to show you the complete connectivity technology portfolio we have built over the past 30 years. As a connectivity-first company, we have placed dedicated focus into being the first to solve tomorrow's AI infrastructure challenges. We built the best franchise in high-speed analog with expertise spanning across all compound semiconductors. We developed our analog IP that is tailored to the specific requirement interfacing optics. We built our DSP portfolio with advanced nodes from the very beginning. We developed expertise in DSP algorithms supporting both time and coherence modulations, and also with a variety of IP options. We also developed photonics in-house as well as forging partnerships with optical component vendors in the ecosystem. We work with a wide range of optics from silicon photonics to micro-ring, micro-discs, and plasmonics. The result is a technology portfolio that covers every single connectivity. Not by chance, not because of one product, it is 30 years in the making. I am very excited to be in this market. Personally, I earned my PhD in Wuhan with the focus on photonics. I also spent years on high-speed analog with LEDs, and I developed solutions before I joined Inphi. It's exciting to be able to bring all this technology together to support the market we are in today. There is one aspect of this enabling technology portfolio that does not get enough attention: the TIA and the driver. If you take a look inside the optical module, has been the analog sits right at the boundary between the optical domain and the electrical domain. The job of the TIA and the driver is to convert electrical signal to optical signal and back again. The performance is everything here: noise, linearity, and power. If you got it wrong, the whole link suffers. You also have to do it at high speed, generation after generation, using circuit innovation because you know analog does not scale with process. This is not a single product category. Each generation will support a wide range of optics which have different requirements. The market is also going to expand to support emerging form factors like NPOs and CPOs. This is a component category where Marvell has built quietly this industry-leading performance and is foundational to everything else they do in optical interconnect. Marvell has a long history here on the product side. They have led the industry since the 40-gig generation. That is where we got our start. Today we are at 1.6 generation shipping in high volume, leading our competitor by two years and counting. On the technology side, we have invested across multiple processes, across a wide range of compound semiconductors, developed together with our foundry partners, because different optics again require different requirements on the bandwidth, swing, and noise. That multi-boundary, multi-process leadership combined with our capacity enablement. All that technology developed in-house supports our flagship DSP portfolio. We have been leading since inception of a time with a fully co-designed DSP, driver, TIA solution. They have been leading at 200 gig, 400 gig, 800 gig, and 1.6. 800 gig is the workhorse of AI today. It is what's powering all AI models that you are using right now. There is a strong continued demand from the market for our economic solution. This is what helped us to drive this business to multibillion dollar annually. We are leading at 1.6, first to sample, first nanometer, first to 3 in volume shipment. Marvell is the powerhouse here. We are in the middle of the fastest run of 800 gig and 1.6. Interconnect is critical to AI, and the way we're being engaged with the customer has evolved right along. 10 years ago, it was very simple. You got a spec from a standards body and went with it. However, early on, we realized that model wasn't sustainable for the market. Very heavy. So Marvell established a multi-generational co-design model, working directly with the cloud customers, system integrators, and optical component vendors in the ecosystem to build it together years ahead of deployment. The result is a solution that is aligned to the customer architectural need from day one and is ready to run when the moment is needed. This is even more important in an AI era where GPU and XPU refreshes occur every 12 to 18 months. Fast forward to today, and our approach to 1.6T reflects just how much AI workload has diversified: training, inference, networking, security. So rather than one solution, we are executing multiple optimized 1.6T products in parallel, each purposely built. Being the foundation, optimized for power, for introducing optics, RX for link robustness, RM for security, building, and PTP for flexible network development. Customer need this multiple optimized solution so they can have optionality and flexibility to deploy the network and compute. Our market has outgrown an approach long that one product for each generation is sufficient to support this market. We expanded our portfolio to support it. We have the parallel engineering resources to deliver all of them to our customers. And scale. Looking ahead, the jump to 3.2T brings a new level of demand. Even hyperscalers are deploying 1.6T. We are already working on the next step: double the bandwidth at much larger speeds, more power, and still in the pluggable transceiver factor our customers know how to deploy. We started working on this requirement two years ago. Marvell was first to show a live demonstration of 400G per lane in 3 nanometer at 2025. By working with our customer, we learned something important. PAM4 alone wasn't the optimal answer, and time-four modulation has its limit. It's very simple physics. PAM4 has propelled the industry for more than a decade, but at 400G per lane, the reach maxed out around 500 m to 1 km. Most of scale-out interconnect today was 300 to 500 meters, but larger deployments require reach up to 2 km, and that calls for a more advanced modulation scheme. Our answer is Coherent Light, where Marvell is first. It bridges for optics with traditional coherent technologies with a balanced approach to power and cost, giving customers the best of both worlds without forcing a choice. We don't just talk about it; we showed that three weeks ago at the EC, European Conference of Optical Communication, Marvell demonstrated industry-first 2 nanometer technology, live, low-power 2 nanometer 3.2T solution with 400G per lane, 3.2 coherent for center and campus reach with balanced power and cost. We started this journey years ago. As a matter of fact, we are in volume production and shipment of our first generation 1.6T Coherent Light with open optics for campus application. We are bringing Coherent Light into the adventure, and our 2 approach reflects our long-term understanding of the technology and the market and the uniqueness of Marvell to support both time and light. This is the same platform leadership story extended. We build our DSP, TIA, and driver technology with our own leadership IP, and it is what carried us from 200 gig through 1.6T, where we are first to ship 3 nm in volume. At the 3.2T, we are leading with 2 nanometer light and 2 nm 400G, and the race continues on to 6.4T. We have the technology depth and breadth, and we develop it years ahead of when the market needed it. Scale out is a large revenue driver for real interest, continues to be strong in the future. Now let me talk about what we are doing to bring optics into the package, which is what we call scale in. Why is this important? As we continue to scale, interface between chip-to-chip and chip-to-memory will become the bottleneck. At the same time, the sizes are increasingly getting bigger. So not only the speed is getting faster, also need to transmit longer, in some cases meters. Traditional interface shown on the left side is built upon electrical circuit. Electrical die has been a workhorse, workhorse, but it does not scale to what is coming next. It could meet the bandwidth requirement by increasing the link speed over time. However, the distance has its limitation. Together with our customer, we identified a solution to it, designed to replace traditional service for all future generations. We take next-generation light source technology, micro-LED and micro-pixel, and integrate them directly into the package and the silicon with optical service. This bridge rule isn't the light source or the laser itself. It is the integration, the system-level design, and the advanced packaging that make it all work, all based on high-volume component that is ready to scale for uniscale interconnect. This level of integration takes extraordinary engineering. We are already working with customers who are looking to bring this solution into their system. Over time, this could be yet another multibillion-dollar business for Marvell. So let me bring together. Has built a strong position across three things: technology, customer engagement, and scale. These are not siloed with threat; they are one connected advantage. Out, this is our most established fitness and existing leadership position with new designs coming, continue to drive growth. Our DSP revenue has already grown to multibillion annually, and our analog business is headed toward billion-plus annually. We see strong growth continuing in scale-out demand, with the double data rate, double generation to generation, accompanied by an increased account for. In addition, we are seeing new opportunities opening up for scale-up with our TIA and drivers, a significant revenue ramping next. As a result, I expect DSP and analog business that I manage to remain the largest contributor to an overall interconnect, which is in aggregate we expect to reach approximately 37 million in 31 at the midpoint of our forecast. Across the investments we are making today in technology, the customer design, and the manufacturing scale are what let us keep.
Morning. Marvell. Comply. My name is Nick. The highest bandwidth, high-speed optical fiber. Now a lot of 10 anniversaries from Marvell today. Chris, Matt, she, this is also my 10th anniversary, my 10th month anniversary here. Not familiar faces out. Industry for 30 years. My graduate work in electro-optics and high-speed optics, and then progressing to some of the largest IP and optical OEMs in the market, including Cisco, Nokia, most recently Inphi. I could not be more excited today to be at Marvell and lead and have the privilege to lead the most experienced coherent team in the market. Marvell has connections for AI infrastructure at all distances. We heard describe our advance and our leadership in scale out and scale in, and will cover next scale up. I'm here to talk to you today about scale across. Now this market is very dynamic, and data center interconnect is changing. So let's jump into scale across. Now the DCI market has a lot of really positive tailwinds today. First, power efficiency. We heard Sandeep mention this in his talk. Been a transition over the last 10 years from boxes to pluggable to interconnect data centers. This is nothing new to Marvell. Hyperscaler customers were and invent verification. Data center, more power, more fibers. What this, more fibers, more connecting. Data center proving. Plug another tail. Third line system density, millions of fibers are going in annual. Line systems, tremendous innovation here to all the bandwidth generated from coherent. Plug another. Plug it. Tail is very strong last years. And this is scale across. This is defining the data center interconnect market. This is the direct connection of AI clusters to each other from distances from maybe 10 km on a campus all the way up to 1000 km. However, the bandwidth here is not the same. 15er. Of. Time. This market. L the pl a decade ago moving from boxes, smaller boxes. From box. 75% reduction. Hyperscalers. Each generation provides a 50% power reduction per generation. As we can see here, this significant power savings and power efficiency, but we can also look the time frames are increasing. The pace of innovation is pulling the DSP. And this requires extremely. Generations. Have DSP generation now multiple. Now jump further across the AI clusters. 15 times the bandwidth. What this, the market is very profound. Only more bandwidth, it drives the unit capacity of plug. Unit. Buiness. Unit. Number. 30, 31. The and coherent plug. Data center grow over 3 million units. This is a steep ramp. Marvell strength. Powerider. We are toep. Andadovate. Now for the ofation with speeds. 1.6. We did. Now 3.2 is required. We must continue to invest and keep the of innovation rolling. Now Marvell also unique in another. We have twoct. To market. Number one, our module business. We've been this for a decade, and this allows us to very closely hyalers to deler theseules based on performance and power and speed and security to. Iso market and areading merchant coherent. Dspider in the market. This us to add all the accelerating demand from hyalers. Nealers or carriers between these two very distinct and unique paths. Now let's look inside a little b deeper into the module. Tell you these modules, they look innocent enough, very small in your pocket. I will tell you these run at the highest bandwidth, highest speeds across every technology Marvell provides today. So all the advancements that s mentioned, as she mentioned, are inside of these modules. Marvell is a powerh here. Everything you see in blue is designed from the ground up by Marvell. About laser in there, we design and develop this laser. Leading laser manufacturers who can provide high quality, high and highest. Now also very fortunate, I feel privil to lead the largest coherent team in the market. We have the largest and most experienced dspam in the. They've been 2006. Now. Leveraging Marvell power advanced 30 walked us through third. We have the best, fastest and deepest converters to convert the analog to digital and digital analog back into the DSB. And last but not least, the most experienced and deepest photonics team to do integration. To. More 20 years. Put all these technologies together is crucial, but I will tell you one of the most difficult pieces with high speed, high bandwidth optics is the high volume manufacturability, and this is where Marvell excells. You know, been in lots of different optics companies. I can see with high confidence this is a superb that allows us to deliver. Marvell more. Service over 15 billion field hours. Been since we started. We continue to innovate and lead today. And so for me, look at all these core competencies, deep technical. This was a very compelling reason. I feel very privileged to leave this team and join Marvell. Now multiply. But multigenerational scale across requires leadership from a long time. We got into this business at 100 gig ZR about 10 years ago. We then innovated with our customers and brought 400 gig ZR to market. Then we have 800 gig ZR. We partner with our customers and they said we would like you to please put encryption and higher performance in 800 gig. We need this now. So we built another DSP for them, had these new features inside of it. As you can see, 1.6 multiple DSPs are required for campus, metro, long haul. These are optimized for performance, power, and distance. And 3.2T is under development as we speak. This requires not only more DSP, but much like 1.6T, new materials. 1.6T required thin film lithium. 30 years later, excited to work on lithium again. Now plasm is here. This requires higher bandwidth, lower power, shorter modulators, and we are making the investments in this technology to ensure that it's ready when needed for 3.2T, 2, and also be first with 14 DSPs. Marvell is the leader 2 nanom technology, and we didn't do one, two new DSPS, new modules. We demonstrated these public for the first time weeks ago at the European Conference of Communications in Spain. In our industry, this is rare to be able to deliver two DSPs at the same time. And I would contend the first time it's been done in the industry. That on leading node in the market, two simultaneous coherent DPS built, brought up, demonstrated with our technology and our partners as a leading merchant at the same time. This is the pace of innovation that our customers counting on Marvell to continue to deliver. Very, very proud of the team who executed this flawlessly with the first out on the first shuttle in January. But like she said, demonstrating it is work. Having modules, having little samples, great showing it to our customers on time, first time, right, is what they count on us for. This is the live demonstrations in our booth and in our partner booth publicly at the European conference. I could not be more excited about the tails in this business and the team we've assembled over decades to deliver for our customers. We have very strong tail. We see our cam expanding rapidly both in units and revenue. In addition, we are shipping to the top hyperscalers today. We are deeply embedded with design with them and making we deliver the right technology at the right time with the right features. As a result, we on track to drive revenue from business from 500 million in fiscal 26 to safely over 1 billion in fiscal 28, with a clear path to multiple billions in the subsequent financial years in the strategic period 29 through 31. To our customers, our partners, and the Marvell team foring to deliver innovation.
Dave morningskm networing Marvel in February of this Ai and CEO of AI this is my first Marvel day and I couldn't be to be here. I'm excited about the opportunity here before usilled by the magnitude of the opportunity with scale up switching and in optics. I feel that Marvel is layering strength on strength with our inorganic growth strategy. As you heard from that, we have invested as a company over0 billion over the course of the last 10 years in innovative technologies and in leading teams. And my business group is the beneficiary en large park for a lot of those a lot of those investments. So I'm excited about the position that we're in and couldn't be happier to be a part of Marvel right now. So let's dive into the data center networking business group. The group is comprised of three different business units. Cloud optics that you just heard from Rus that includes plug data center transavers and DCI and then switching and integrated optics. Switching includes both scale up and scale out. And integrated optics is NPO and CPO. I'm going to focus my talk on these two areas of integrated optics and switching. Let's start with switching.
So as Matt mentioned, we have an established and growing business in Ethernet switches based on our terlink platform. We expect that business driven in large part by the ramp right now in 51 switches to exceed north of a billion and significantly so next fiscal year. In addition to that, what we're most excited about is a new flagship product that we have recentlyed which is based on much of the techology that deep introduced in his talk earlier today. Carolink T1 product and introduces multiple leading edge disruptive techology capabilities including it is the first and only monoltic fulltical limited die that is operating a 100abit per second of bandwid how weve achieved this is asandy pointed out Marvel has the world's leading 200 gig and 400 gig have 200 gigs in a limited die in this device. Why do that matter? It matters because if a uniform monolic device that means that the latency of this device isicantly lower than a multip module package. Our latency is leading edge we on the order of 20% or more lower latency and in fact lower power than any leading competitor in this space. And why that matters is because we can use this device uniquely to serve both scale out applications for conventional as well as scale up this emerging market that is that is taking off right now. So we're thrilled about the opportunity for the 100 switch and we expect the 100 switch as a product in of itself to layer on incremental multiple billions of dollars of annual revenue in the coming years. So a huge opportunity from this switch alone.
Right behind that we have a series of additional Ethernetbased switch solutions that include the 200 switch includes the 400 switch and we have Et e Ethernet scale up network solutions that include an E115 115T ethernet optimized for scale up similarly 230 and 460 in the pipeline. So a broad suite of Ethernet based switches and ebased solutions for scale up and scale out. In addition to that, Marvel is developing for our customers UA linkbased switches. We're building now and we be releasing next year a UA link U115 switch that is being deployed at one of the the industries leading hyperscalers. And right behind that we have the U115, excuse me, the U230 and a U460 switch. a broad portfolio the industry broadest portfolio of scale up and scale out switch technologies we have within Marvel excited about this opportunity a growth for the company going forward and the opportunity extends also to not just UALbased protocols and ethernet or ebased protocols but as you as Matt mentions uh we have partnered earlier this year with Nvidia and we now offer link solutions have to solutions all our customers up to meet their requirements on application by application basis and on a customer by customer basis. So let's let's discuss what I'll do is I'll go into much more detail on scale up. You heard from G and you heard from Ross scale the scale across, scale out and scale in markets. I'm going to dive into scale up in detail not just talking about the products and technologies but about the application drivers at the artificial intelligence workload level that are driving requirements for the next wave of data center infrastructure specifically in scale up we feel that scale up is going to drive a tremendous opportunity for the company inflection point and it's inflection point in particular with this transition fromer to optical interconnectivity that Marvel so well positioned to lean.
So let's begin with the workloads. The workloads that we've been addressing an industry over the course last several years began with the ability to train multitrillion parameter models. And this is this is hard. See, it puts a lot of pressure on all of the resources on compute, on memory a lot of network intensity. But over the course of last couple years ag there a new wave of inference models which are reasoning inference time comput and what that did it introduced a set of requirements for higher capacity bandwidth higher capacity high bandwidth memory and and it introduced a need for scale up networks to increase the amount of memory semantic interconnectivity with scale up domain XPU to XBU connectivity those workloads were driven by open AI3 and1 for example in Deepseek R1 that was released couple of years ago and then more recently you see this push toward mixture of experts in longer contextbased workloads this was the result of a very important innovation that took place in June of 202 years agy and Berkley a group introduced inference engine vlm and one ins vlm was kb and kb has now proliferated rapidly through the industry and enabling much more efficient uation of memories i'm going to spend some time in this talk talking about the impact that kb is having on the data center infrastructure that we're building with all of our hypale customers but all of these innovations have have come and a shi at the time a ship from a training centric environment for data center infrastructure to increasingly inference centric infrastructure about 60% ofal data center infrastructure is inference and as you all know inference is the return on that roi equation training is the investment so we're seeing not just an increase in inference we are seeing an increase in efficient inference so what that is in turn is drive a significant increase in the operating profitability operating performance operating profits of the cloud service providers years ago there were some cloud service providers that had operating profits in you know middle digits today almost all of them are seeing operating margins that are approaching 40% the math is working the flywheel is spinning those profits are being reinvested in data center infrastructure decruified because investment is taking place across compute across memory across networking and across connectivity market is coming to us we extraordinar extraordinar di into specifically happening with reasoning why we care about this group of researchers that did years ago but we don't really care so much about that as much as we do about the utilization the efficient utilization of memory what we care about is how the model is building the context token by token. So again this group introduced the concept of KV cash. So what is KB Cash? KV Cash is the storing of key and value matrices rather than requiring the need to go recompute that stuff. So you have the you have the information available in high speed memory so that it makes the interaction in reasoning models rapid. wait for recompute. Humans are inherently impatient right so the problem though is that the amount of KV cash is not unlimited. So puts a lot of intensity on memory capacity on memory bandwidth and it puts a lot of intensity on the network latency. But at some point you exceed the amount of available KB cash in the system and at that point one of two things happens you either need to reload from a different location in the memory hierarchy those key and value uh parameters or you have to go back and recuteuse them. In both cases, you have to make room in the KV cash for the new key value indices and it and you have to reject or evict information to make room for it which can potentially result in hallucinations. So in all cases this reloading and recomputing results in inefficiencies. It drives higher memory latency. It drives a need for memory and compute transactions. Overall a negative impact on the efficiency of these systems. So ideally we have we have systems that have ideally infinite amount of KV cash highspeed memory that's available that can be addressed by any XPU. The problem is it is not infinite. The KV cash is fine. So how we're addressing this issue is at the system level we addressing it first with maximizing the amount of high speed F RAM on the chip. We then increase the amount of high band memory in package if you look for exle state of the RTP from Nvidia Ruben has roughly 300 gab of memory capacity in the package. The problem is that that 300 gab is orders of magnitude too small to deal with advanced AI models. So that is driven Nvidia and the rest of the hyperscalers to build systems that interconnect multiple XPUs together in a memory semantic scaleup domain. In the case of Nvidia, they use NV link and NV switch. Right now, there's 72 interconnected XPUs inside of that rack. The problem is that the memory requirements are inable. They're continuing to grow and they're spilling over what will fit in the rack. indry needs multi 576 how you get and what are the challenges right some the challenges are today 100% of all interconnected x inside of all racks are all of it that supposed to challenging the opportunity for our so why is that so why not just take thater and extend it to a multi 224 gab per second going by the way to 448 gab per second but at 200 gab per second the reach that you got is 2/ it forces you inside the rack so the only way to overcome this limitation within the scale of domain is optics rather fighting the laws of physics knowing that moving information overer electronically dealing with electron unfortunate el mass the beautiful thing about optics is that they have no resting mass right so it gives us the ability to leverage physic use it to our advantage thaty to fight the physic it gives us the ability to increase fully connected scale of domains with optics larger numbers of directly interconnected xus sharing the same memory with extremely high bandwidth and with extremely low latenc but the the bars It's very different right than a scale out network. If you think about it, the XPU on the left here has to be able to operate do a load transaction and address the Xbu on the right the memory in that Xbu and pull it across the network and ideally it is operating as if it's the same the same performance in terms of bandwidth and latency as the
Matthew Murphy 2:11:39 ↗
Memory on the package. That is challenging, right, because HBM has very high bandwidth. It's tens of terabits per second. Right now, HBM4 has about 16 per second. And you do not want the network to be the bottleneck for a memory transaction. You also want to be able to minimize latency. Again, latency for these applications and reasoning translates directly to revenue for our customers. The latency inside of the package is less than 100 seconds. Latency going over a scale-out network, if you had to use a scale-out network for this, would be hundreds of times greater than the local memory latency.
So what you really need is the latency down around 200 nanoseconds. Really, really fast memory transactions, also with low power. Sand pointed out earlier that the amount of data traffic going across scale-up networks is 85% of the total data center data traffic. 85% of it, only 15% is scale-out. What that means is that if you can reduce power here even a few picojoules per bit, you can repurpose that energy that's being used to move information and use it to compute information or tokens. Tokens aren't generated moving information, right? So the options that we got for addressing this set of challenges are unfortunately not pluggable, as you just heard from...
The leading pluggables are going from 1.6 to 3.2. It's an order of magnitude lower in terms of bandwidth. The power is not there. The latency is not there. They're not designed for these applications. This application requires integration. It requires NPO and CPO.
The good thing about the circumstances for Marvell is that this just comes to us because we have a full suite, as pointed out, of optical technologies that we have built over the course of the last 25 years with the best on the planet building these technologies, and we bucket them in three categories.
The category of working with module providers in the way that we have with pluggables, with supporting these providers in the overall optics ecosystem with drivers, which we're doing today. And then we have two categories of fully integrated, we call light engines. One of those categories we call Fast Pipe, which are 200 gigabit per second going to 400 gigabit per second per lane in optics, and the other we call Flat Pipe. Flat Pipe optics are designed specifically for scale-up networking. Much lower latency, much lower power. We have two flavors of Flat Pipe. One is based on industry standards that are emerging now, OCSA, a big set of acronyms that stands for Optical Compute Interconnect Multi-Supplier Agreement. That's basically the industry converging around base optical scale networks, and they've landed there. So the set of optics that the industry has standardized on or is just working on standardizing on, we have that as an option. The other is the technology we developed, the photonic fabric, which is fully optimized, again, ideally for scale-up networks where we've got strong customer demand.
We put checks in the win column as it relates to customer demand across all three of these categories. We're winning in delivering drivers to module providers today. We have won a major socket recently. We're winning multiple sockets in Fast Pipe optics, and we have won sockets, including the world's leading hyperscalers, in Flat Pipe as well.
So let's double-click on this and take a look at the level of complexity of these devices, right? What's in here, right? So in the case of the Fast Pipe light engines, here's our 2x400G. You know, we're able to leverage things like our optical DSPs, the range of optical DSPs that we've got that are fully retimed or transmit-retimed optics. And then if you blow open the light engine, you can see a lot of the innovation that Marvell developed over the course of the last 25 years in advanced drivers, including silicon and germanium. And then we have again into the silicon photonics with Mach-Zehnder interferometers, and all of the optics and packaging and scale that you heard from Russ and Guy, leveraging all of that in an NPO form factor. So a large opportunity, Fast Pipe optics, excited about that opportunity.
On the Flat Pipe side, very similar, other than the fact that this technology has been designed from the ground up for this application, scale-up. And what we've done is, if you look at the light engine, it is far more integrated, right? A deeper level of integration in the optics. So all of the discrete components are now integrated in an advanced package. This is the driver, the controllers, and most importantly, analog. These systems are designed to eliminate the need for digital signal processors. We're using analog equalization and eliminating the DSP, take the largest power contributor out. Game-changing performance, game-changing latency. The ultimate optimal solutions, by the way, that Flat Pipe is applicable for both. I'll jump back because it's important.
It's applicable for both silicon photonics as well as electro-absorption modulator-based photonics, sharing the same control electronics, the same advanced ASICs. So this is where we brought the team together from all, have come from Marvell over the last 25 years with our team from Celestial AI. So now, some customers are interested in NPO-based form, are interested in CPO. From our standpoint, the answer is yes. We're agnostic. If a customer's operational priorities are the replaceability of the optics, worried about blast radius of a failure, or if they want reconfigurability, you plug it into the board, and if there's a concern, you swap it out. Not a problem for those customers. On the other hand, that are looking for the ultimate lowest latency, lowest power...
This is the deepest level of integration, which provides the most superior performance for scale-up network solutions. We have developed both, and we are in the process of deploying both. So again, Marvell in a position where we don't have to force our customers to make a choice. We can embrace them, and in fact, that's how we do it. So if you think about how all of this technology comes together and how we work with customers, our customers are operating with us as an extension of their own teams, engaging with them to understand what their requirements are for developing, architecting, and developing the right scale-up network across the entire link from the XPU interface, going in some cases going from copper to the right optical I/O, determining what is the right physical layer, what optics do they want to run, what is the right logical layer, what protocol do they want to run to the switch, all of it. And these things are highly integrated and interdependent.
Most of these programs that we have with the largest hyperscalers in the world have started years ago, right? So we're working with them to develop solutions in networking that takes time to get developed and implemented and deployed, right? And it enables us to design and optimize for link performance, including power, latency, bandwidth. It enables us to accelerate time to market and time to first token generation within. And it also helps them de-risk the operation of these devices by leveraging things like telemetry, so they can see the stability and the resiliency of their links, and employ things like security to ensure that in the world of agentic AI, their links are secure.
So these are examples of how we're working together using the leading-edge technology that Marvell has brought together to address these rapidly evolving architectures in scale-up networking. Now, what I'm about to show you is how we are adding new capabilities and tools to the toolbox to help our customers disrupt this, meaning help them craft system architectures that can provide next-level capabilities for dealing with things like KV sizes, to decouple memory scaling, high-speed, high-capacity, high-bandwidth memory scaling from compute.
So what you're looking at right here is what we call the photonic fabric memory module. These modules, 8 to 16 of them, are integrated in a photonic fabric memory appliance, and what it does is it provides, for the first time really in history, the ability to decouple memory scaling from compute. Sockets that are currently available for HBM reside 1 mm away from XPU, a GPU. So what that means is that our customers are forced to use their GPUs as the world's most expensive memory controller. This technology changes that. It changes fundamentally the economics of memory, and we're excited about where we stand with this. This is a product that we taped out earlier this year, being brought up now and sampling with multiple customers as we move into next year. So very excited about this opportunity.
So in closing, I'd like you to take away three primary things from what I covered. The first is that we have an established scale-out switching business. All of our business switching today, all of it is scale-out. Now that business is growing, doubling year over year this year, more than double again next year, to well over a billion annual run rate, and that will grow to a multibillion annual run rate in scale-out switching alone. In addition to that, we are leading and enabling this new emerging market segment of scale-up switching. And there we have significant customer traction. We expect scale-up to layer in an additional multibillion annual recurring revenue opportunity for Marvell.
And in addition to that, we have scale-up optics. And what's interesting about scale-up optics is the fact that there is significant Marvell content. Content for a few reasons. One, because the volumes are so high. The other, because you saw the level of content in these highly integrated devices like the Fast Pipe and the Flat Pipe light engines. A lot of content.
So if you think about it in terms of the dollar content for scale-up optics per XPU, you should think about this in context of on the order of thousands of dollars of scale-up optics per XPU. So it's large, valuable technology, but it's bigger than that. The optics are required on both sides of the link, and our customers in large part are looking for comprehensive solutions on the XPU side as well as the switch side of these links. So we're excited about the magnitude of what the scale-up optics market brings to Marvell, the fact that this market just comes to us. So that we are in the right place at the right time with the right technology, and we've got the right trusted team to execute. I'd like to say thank you.
All right, good morning. My name is... I've been with Marvell for years, and I lead our custom business. At last year's Custom AI Investor Event, I talked about Marvell becoming a leader. It turns out since then, we've grown dramatically. We increased our fiscal '29 revenue outlook from more than 10 billion to more than 12 billion, and we're on a path to approximately 30 billion by fiscal '31, at the midpoint of our long-term target range. Today, super excited.
Let me start with how we built the custom business from the ground up. Very early on, we saw that hyperscalers were going to need significantly more custom solutions. We recognized that Marvell already had the...
core IP to address this market and world-class engineering teams with 30 years of experience in storage and custom, 25 years in networking, compute and security. We also had decades of experience taking the core IP from our merchant products to create semi-custom and full-custom solutions. My mission has been to integrate these teams with ours and build a custom solutions leader. We are focused on XPUs and a diverse set of XPU attach solutions, and on customers that leverage Marvell's entire technology platform. Our strategy is working.
At last year's custom AI investor event, I talked about the 18 sockets we had already won. Since then, we've continued with many more designs and with multiple multigenerational wins across the top four hyperscalers, along with early wins with many others. That continued momentum is what's driving our strong revenue growth.
Hyperscalers were initially interested in custom XPU. Our initial XPUs were already complex devices, one to two reticle size, with integrated HBM using 2.5D/3D packaging technology. Today our XPU business is on track. We've already shipped in high volume and we're executing very well across all our programs.
Looking forward, XPU complexity will continue to increase as hyperscalers push for more performance. You see more compute, more HBM and standalone copper IO chiplets all integrated using 3.5D packaging. By the N+2 generation, we expect even more compute and HBM, and that these XPUs will approach the copper wall and need to integrate optical IO.
This is why at our custom AI investor event last year we highlighted the critical technologies you see here. Our deep investment in connectivity IP is another big differentiator and serves us extremely well as XPU architectures continue to scale. Hyperscalers need a partner like Marvell that already has a track record of taking complex XPUs into high-volume production and leading NPO and CPO technology. As Dave highlighted, we are focused on customers that leverage our complete end-to-end portfolio of XPUs with our IO chiplets.
That combination of connectivity leadership and XPU experience makes Marvell up to be a big winner as AI architectures scale, and our partnership with NVIDIA further expands that platform. NVIDIA and Marvell are engaged with customers on NVLink Fusion solutions, combining Marvell custom XPUs and NVIDIA's scale-up connectivity and rack infrastructure.
Now our opportunity extends beyond the XPU, and there are a number of exciting XPU attach opportunities. The XPU attach opportunity really evolved from the traditional data center market. The market is a well-established ecosystem of merchant solutions using standard interfaces like PCIe to support x86 CPUs. Marvell participated in that market for many years. We developed networking, processing, security, storage and memory products and shipped millions to the hyperscalers.
As the industry transitioned to AI, hyperscalers initially leveraged those merchant solutions, but as XPUs became more complex, they needed increasingly optimized solutions around the XPU. We saw this trend early and engaged our hyperscale customers to address these new requirements with our IP expertise, customer credibility, and the experience of turning our merchant IP into semi-custom solutions.
What makes XPU attach very attractive is that these sockets are very sticky. They persist across multiple generations of XPU, they leverage Marvell IP and expertise, and they extend our reach across the entire AI stack, even with non-Marvell XPUs. Let me take you through a few of them.
In traditional data center, a vast majority of the market is merchant NICs. Foundation NICs are the most prevalent. There are smart NICs with integrated CPUs, and IO speeds are typically 100 gig. Marvell has shipped NICs for decades and millions of parts in production. Marvell invented the smart NIC back in 2005 and has been in production for over 20 years.
There are a few merchant solutions on the market today, but all hyperscalers are moving to custom because standard solutions lack either the optimization or the performance they need. Hyperscalers are looking for a specific mix of IO speeds, bandwidth and configuration. They need optimized data path accelerators as well as custom networking protocols, integrated CPU dimensions, right for their applications.
All these requirements tie directly to how efficiently data moves across the network to support their specific workloads. Marvell is a trusted partner for hyperscalers to solve these challenges. We have leading 35 at 100 gig, 200 gig, 400 gig, decades of experience with mixed high-performance data path and accelerators.
We are doing well here. We have design wins with multiple hyperscalers across multiple generations that are ramping in execution. Now take a look at memory.
In traditional data centers, workloads were CPU-centric. CXL was originally developed for traditional data center applications, not AI. We at Marvell had early investments in CXL, but this market developed slowly. AI changed everything. Workloads like inference and KV cache created a very early on we saw the memory wall challenge. We worked closely with a leading hyperscaler to define a differentiated solution leveraging our early investments.
Today Marvell is a leader in memory computing with multiple hyperscalers and with multiple programs in execution. Our structural X products support DDR5 and DDR4 for memory recycling, industry-leading memory capacity and compression to extend the capacity further.
It brings compute closer to memory to accelerate memory-intensive workloads like LLMs and offload the CPU. We also have a broad ecosystem across memory, CPU and GPU providers. This is a memory adding card developed with a leading memory vendor, and the opportunity is expanding beyond CPUs.
XPUs need more memory capacity and bandwidth as well. We're now developing custom memory expanders that connect directly to the XPU at very high speeds using our high-speed 30 IP, giving the XPU direct attach to expanded memory. This creates growth for our memory expansion business, and with that, storage.
Historically the storage controller market had many merchant solutions, often bundled by the flash vendors with their NAND. There are also merchant accelerators that offload storage and data management functions from the CPU. Again, Marvell brings decades of storage experience with millions of SSD controllers and accelerators shipped across multiple generations.
Looking ahead, hyperscalers want to control their storage firmware and software so they can leverage their investment across multiple generations of SSDs and multiple NAND suppliers without being tied to a specific drive vendor. AI workloads are making flash IO speeds and storage offload performance increasingly important.
This ties directly to our storage heritage. We leverage our merchant technology to build a leading portfolio of semi-custom SSD controllers and storage accelerators. For hyperscalers developing do-it-yourself SSDs, our multi-vendor controller delivers advantages. For accelerators, we optimized our merchant technology for an initial hyperscaler and already secured design wins beyond that initial engagement.
Our accelerators increase storage throughput and provide confidential compute and virtualization offload, freeing up the CPU. Now I also see storage requirements increasing with each generation. This is creating a win for our custom storage business.
Now I'd like to talk about a couple of emerging XPU attach opportunities that we've never talked about before. The first is a new class of AI infrastructure management silicon. We talked before about NICs, memory and storage. Today we're introducing an entirely new XPU attach category driven by a fundamental change in AI infrastructure.
In traditional data centers, BMCs monitored the health of a relatively simple server, typically one to two CPUs. These servers have approximately 150 sensors generating relatively small amounts of data. Failure requirements are pretty and the impact of a failing server are also relatively contained, typically one to two CPUs. AI changes this dramatically.
AI servers are some of the most complex servers ever created, with more than 1,500 sensors and generating hundreds of megabytes of data per day. Just managing this complexity requires 100x more performance, and the impact of failure is much greater. A single failure can disrupt the work of thousands of XPUs. Traditional BMCs weren't designed for this.
AI infrastructure needs an entirely new level of performance, processing and IO. At Marvell, we recognized this challenge and actively engaged with multiple hyperscalers to build a new infrastructure management solution. They needed a much more powerful processor to manage their servers and switches across their AI infrastructure. We already had deep relationships with these customers and a long history of delivering infrastructure-class processors.
So we built an optimized solution leveraging our technology, proven 5-nanometer IP and experience in high-volume mission-critical applications. The Marvell semi-custom processor delivers a 100x increase in performance needed for these AI infrastructure management workloads. Today we have multiple hyperscalers creating an attractive attach growth opportunity ahead of us.
Now to another emerging attach opportunity. The next opportunity is inference acceleration, and this is an exciting emerging category that could expand the XPU attach market. Inference, as you all know, is growing rapidly and architectures are evolving to deliver significantly more performance.
One of those changes is the use of specialized inference accelerators to attach to the XPU to offload specific workloads. These are highly complex, large devices that require the most advanced process nodes. This represents another significant silicon opportunity around the XPU, and Marvell was already investing in the technologies needed to address it.
Our 30 connectivity technology can deliver 17 times more bandwidth at 23x lower standby power than standard solutions, and advanced 3.5D/3D packaging technology integrating 3D-stacked memory and logic to enable even more performance. Together these technologies allow us to optimize inference accelerators for performance, bandwidth and power.
We believe inference accelerators can become a significant new attach market, and we're already engaged with multiple customers. Marvell Technology.
To summarize, we have all the major hyperscalers. Since last year's custom AI investor event, we've deepened our customer relationships and have won additional multigenerational designs. All of our XPU programs are on track, and XPU attach has multiplied opportunity.
Standard products are moving to high-performance and more semi-custom solutions for NIC, memory and storage. New attach categories are emerging, including AI infrastructure management and inference acceleration. As AI performance scales and XPU complexity grows, we see higher attach rates and higher ASPs, significantly expanding our opportunity.
You can imagine one to two sockets attaching to each XPU with ASPs around $1,000. Add all that up and you can see how each attach category can represent a billion-dollar opportunity with some even bigger.
So as these ramp, we're building a large and increasingly balanced custom business across both XPU and XPU attach. Very pleased with how quickly our custom business is growing and diversifying. As XPU attach ramps, it will drive greater balance between XPU and XPU attach.
We are on track for our custom revenue to more than triple from fiscal 28 to more than $12 billion in fiscal 29, well above our prior fiscal 29 target of more than $10 billion. Further ahead, I'm personally very confident to grow custom to approximately $30 billion in fiscal 31 at the midpoint of our long-term target range.
My confidence is driven by the momentum I discussed today across all the tier-one hyperscalers in both XPU and a diverse set of XPU attach programs. As you heard from that, we don't need new wins to achieve our targets. Our technology platform is in place to deliver on the custom opportunity ahead of us.
So like I said at the beginning, it's been a busy year, and we're just getting started. Thank you.
Welcome, President and Chief Financial Officer.
Good morning, everyone. Thank you all for coming today. I'm Dan, Chief Financial Officer, and I've been at Marvell a little more than a few months. From a career perspective, I got my start in the military. I was absolutely honored to wear the uniform of my country for a dozen years. Subsequent to my time in the military, I spent a little more than half a decade in and around the semiconductor industry, first as an investment banker with Goldman Sachs in their M&A group covering semiconductor companies, then in private equity investing in semiconductor companies, and now as CFO.
I want to talk about how excited I am about what we have at Marvell. We're at a unique inflection point. We believe we're on the cusp of increasing the magnitude of our revenue by 10x, from $8 billion to $80 billion of revenue at the midpoint of our long-term target, doing it in a very short period of time. And it's all predicated on a vision, one that was laid out 10 years ago by Matt and the team, and it hasn't changed. A decade later, it is the exact same vision.
So what we've done is we've taken that vision, that star for the company, and a point of where our end markets are going, and we backed it up with conviction and a decade of investment. We invested billions of dollars to bring this vision to life.
We are purpose-built for this. We're entering the largest waves of infrastructure investment in human history, and we're doing it with critical enabling IP and foundational technology. And it's IP and technology that directly addresses, strikes right at the heart of the biggest rate limit in data center performance today, and that's connectivity.
How do you efficiently move, store, process and secure data? How do you do it at an absolutely massive scale and power-efficiently? At Marvell, we built for this moment, and it's been a decade in the making.
As CFO, I take in the underwriting of growth. You heard from the team today, the growth is diverse, diverse across products, customers, technologies and multiple generations of designs. We've got many shots on goal, and no mega-socket is needed to deliver on our long-term targets. I can't think of a better time to be at Marvell. With that, let's dive into the financials.
Before looking forward, I'll do a quick recap of how Marvell has performed to the long-term targets that we put out a handful of years ago, which sets the foundation of our journey ahead.
At the time in 2021, we outlined three key priorities, and Marvell has delivered on each. Revenue: we delivered a 23% compound annual growth rate from FY21 to FY26, exceeding the high end of our long-term revenue growth target of 15 to 20%. Operating margin: we delivered a 32% compound annual growth rate. Finally, capital allocation: we returned more than $5 billion to shareholders through dividends and share repurchases.
We've invested in growth, we've returned capital to shareholders, all while maintaining a strong and flexible balance sheet. With that behind us, now let's look ahead to what we see today.
First, we see a huge revenue opportunity in the next five years, and largely coming from our focus on connectivity and compute for hyperscale customers. $80 billion in calendar 2030, a 4x increase. We see this growing at a 45% compound annual growth rate, even when factoring in a moderation in data center capex, which we have growing at a 35% over this time period. So we window back out data center capex at 35%, and we see 45% growth, 55 to 60% for our data center business.
Second, we have an array of connectivity and custom technologies to address the massive TAM that we see. That includes our leadership positions in DSP and optics, analog, mixed-signal. These capabilities are foundational to our success, truly differentiated and truly world-class.
Third, the combination of this massive TAM and our differentiated technology allows us to accelerate our revenue growth in the years ahead.
Now let's look at the going forward. We're focused on the same three metrics: driving strong revenue growth, delivering earnings leverage, growing cash flow. And I'll go deeper into each of these in the subsequent slides, but I just want to share a few quick highlights. Revenue: we expect the revenue to roughly grow at 55 to 60% CAGR. On earnings, we expect to grow earnings meaningfully faster than revenue and deliver operating leverage and do it at scale. Finally, we expect to generate strong and growing cash flow that gives us positive optionality for capital allocation.
For calendar year 2030, Marvell's fiscal year 2031, with that we can grow our revenue from $20 billion to $80 billion of revenue, the midpoint of our long-term target in FY31. We expect the communications and other business to grow at the same low single-digit rate that we've discussed in the past. But as you can see, data center business is the primary driver, 65% of revenue, growing faster for longer.
Consistent with the data center splits that I highlighted earlier, we see strong growth across all three pillars in the years ahead. Interconnect: custom DSP, switching and storage for interconnect, with the move to optical connectivity supercharging growth. We forecast a revenue CAGR of about 65% at the midpoint, driven by scale-out, scale-up and scale-across.
For switching and storage, we see a revenue CAGR of about 45%, with our switching business expected to grow even faster, and that's driven by scale-out and scale-up. Finally, custom: we see a revenue CAGR of approximately 80% at the midpoint, and that's driven by both strong acceleration of our XPU business and strong acceleration of our differentiated XPU attach.
Key takeaways: we've got growth in the years ahead. We have large, diversified, high-growth engines emerging.
So let's talk about gross margin. As I said, Marvell's gross margin is mix-dependent. Custom silicon carrying a relatively lower gross margin has attractive operating margin due to customer-funded NRE and scale of the business. Merchant silicon, on the other hand, has a relatively high gross margin. Semi-custom variations have gross margin somewhere in between.
We expect to grow operating expenses at a slower rate than revenue growth, and that will deliver nice operating leverage as well as sizable upside to the operating margin. And in FY31, based on this combination of revenue growth, strong operating discipline and execution, we're going to deliver a structural increase in the cash flow generating capabilities of Marvell.
We're going to do it with a greater than 50% free cash flow margin over time, and we'll maintain a strong and flexible balance sheet.
To summarize all this in a table, our new long-term target model for FY31: we expect to deliver revenue of $70 to $90 billion, and that's driven by a diverse set of customers, products, multigenerational design wins, and that's across all three pillars of the data center business. We expect a gross margin of 56 to 59%, with the range being mix-dependent.
It's also important to keep in mind this mix dependence, it acts somewhat like a hedge. The low end of the gross margin range likely requires a significant upside to the base case revenue forecast for the custom silicon business. Conversely, if custom revenue nicely outperforms, our gross margin will sit at the higher end of the 56 to 59% range. So the mix dependence is a bit of a hedge.
We expect an operating margin of 44 to 46% as opex grows at roughly half the rate of revenue. And taking a step back, putting that in our new operating margin target, it's points higher than the prior long-term target we put out a handful of years ago. This significant increase in profitability is a function of the increased scale and differentiated positioning we have as a company.
We expect a tax rate of 15% and a free cash flow margin of greater than 36% over time. And I know that you guys all took a stab at what the earnings power of the company a couple hours ago and previewed and laid some comments. So to demystify it, with these assumptions baked in, I can confirm that you guys got it right, which I'm sure you did. Here's the math.
The combination of these metrics imply a non-GAAP EPS of greater than $10 a share in FY31 at the midpoint. So if I back out and think about the power of this company and how to frame that, I think $30 billion of revenue, greater than $10 EPS in calendar 2030, which again is equivalent to FY31, greater than $10 EPS in calendar 2030, 30 by 30.
So to back up, in Marvell, you heard about technology from Will, you heard about markets from Dave, and I want to give you the financials. $80 billion revenue opportunity in calendar 2030, 4x increase. $20 billion revenue guide FY28, $12 billion plus revenue guide FY29, up from $10 billion plus. $70 to $90 billion revenue target FY31, built bottoms up with a conservative view on data center capex. $30 billion earnings power long-term, greater than $10 EPS in FY31, 30 by 30.
We at Marvell, we've worked hard to earn the trust of our largest partners. Similarly, we appreciate the trust you, our investors, place in us. On behalf of the entire Marvell management team, thank you. With that, welcome back up, Q&A.
Matthew Murphy 3:02:20 ↗
But before you do that, I think Matt just took it off. Anybody want to do Q&A? And before that, I give recognition to the team here. You guys have done this relentlessly for decades. And Chris joined me in 2018, and Marvell was not in the place that we are today. He was very brave to take this on.
And you know, being CEO is a lonely job, and I think one of the few roles and people that understand that implicitly is the head of investor relations. And so in my darkest times and deepest criticism and things not going the way we wanted, look, it hasn't been all to the right for the last 10 years, although if you look at it, it's pretty good, right? 200 to 300, not bad. But along the way, we've had a roller coaster, and I just have everybody stand up.
But we're keeping him on the payroll for a little bit longer. We just can't go any further. With that, why don't we kick off and we can take the Q&A. To get to you, so those of you who have been in your seat, so please listen to this last thing.
Yeah, sure. Yeah, that's a great question. I think these programs that we're executing are all proprietary in source. It's Marvell IP technology, especially when you get to the more complex products. And on XPU attach, it actually starts with Marvell, and in some cases it's our entire design, what you call build-to-spec, which means we basically get a specification from the customer, we work with them, and we deliver it to them.
And on the magnitude of the opportunity, when you go out to 2030 calendar, fiscal 31, and you're looking at the $30 billion, what I said is I think the mix, we still don't know exactly, but think of it as largely balanced, right, between the existing XPU programs we have coming and then XPU attach. I could see that one way or the other. I think it depends on how some of these programs take off.
But with respect to the XPU attach and the large strategic agreement that we announced in August, the way that we've got that modeled right now in fiscal 31 is very pragmatic and judged. So don't think of this as the max case when you built your model on what the opportunity is.
As usual, the way we do it, we're not conservative and not aggressive, pragmatic. We use the most reasonable outcome and we start from there. And then as we build over time, you know, we adjust, right? We adjust our view. So that represents not the full value of the warrant, but some portion of it that we view is a very appropriate, judged view for this time.
So what that means is you've got a very healthy set of other sockets coming in. As we talked about, we had 18 of these already won like a year ago. We've layered in more, and they're across all the different hyperscalers, across all those different technologies. So it really will be a very diversified, rich business when you get to custom in 2030.
And it's actually a great way to play custom because you're not rolling the dice on one giant socket and hoping and seeing what happens. You know, we've developed quite a portfolio here, and I think that's going to serve us well as we grow through the next few years. Thanks for the question. Next question, we go to Tim, and I'll traffic on the question, I'll farm it out.
Hey, thanks. So Matt, one emerging theme is sort of powered shell and whether there's going to be enough powered shell for all these XPUs. And Jensen would say that there'll be enough for his product, but there won't be enough for anybody else's. So I don't know how much does that, I know that you're being, you know, very conservative in your, you know, XPU model for, you know, fiscal 31, but like, how do you think about that? Like, you know, when you're talking to hyperscale, you know, CEOs, is this something that's like coming up?
Matthew Murphy 3:07:31 ↗
No, not in the context of our XPU programs or our custom business or, quite frankly, anything to do with our current revenue. You know, the way we have our capex, I think, set aside, land, power and shell, just take a step back. If you look at our model, which is 35% capex over our period from 26 to fiscal 31, that's not a linear capex. So right now it's growing like 70 or more, right? And so we anticipate at the tail of this is probably growing more like 20%.
So we think we built in our model just a lot of cushion for the AI business to kind of digest over time. If capex does better, actually we do better, but we don't see that as any kind of constraint or any kind of customer discussion.
Every GPU, XPU, accelerator on the planet, not just for us. Go to Tom.
Hey guys, hello, hello, thanks for doing the day, appreciate it. For the interconnect, I was curious if you guys could break that down a little bit further into calendar year 30. I think it was $38 billion. Is that more DSP share, more DSP volume, NPO, CPO, any kind of breakdown that you could give us? Because that's obviously a really large number, and where you see the most growth.
Matthew Murphy 3:08:58 ↗
Yeah, I'll give you some of the assumptions behind it. Maybe Chris, you want to chime in after that. I think on the growth is that, and I think Dave did a great job, a very compelling case, right, for why scale-out needs to perform. That's our largest business. That's going to continue, right? So you can kind of know what that model looks like today. I think you should assume that we at least can grow with the market. We don't intend to lose any market share there, okay, just to be super clear.
I think Russ did a great job outlining the scale-across, which has already been growing, but with this sort of scale-across use case, right, with the data traffic increasing and the number of units going up, you should assume that we scale with that, right, and definitely hold our own, by the way, in both of those businesses.
Each of those generational changes that we go through, let's say it's from 800 gig to 1.6 to 3.2, all of those come with ASP uplifts, right? So it's not just units, but it's also the dollars. And so I think you can make a model on that. You kind of know those numbers.
And then you should just assume in interconnect that scale-up optics, which Dave talked about, that's going to layer in the rest. And obviously that's going to grow at a much faster rate because effectively we have zero revenue there today, you know, multi-hundred million starting next year. But that's going to ramp, you know, extremely hard.
And then who knows, there's a call option on scale-in, as talked about, with some of this micropixel technology that's not really material at that point in F31, but it could be. So that's kind of the bucketing I would think about today. It's, you know, I think we're not going to give a lot more precision than that. But Chris, is that about right? You got it. Cool. Thank you. Next question.
Analyst 3:10:53 ↗
Thanks a lot, Matt, thanks for doing this. Matt, I was wondering in the guide through F31, is anything in there for the AI labs, and how are you working with them? Is that potential upside, and if so, which category would it be in? Would it be in custom attach, optics, and how are you thinking about your relationship?
Matthew Murphy 3:11:16 ↗
Yeah, I think the answer is yes, and I think the opportunity set is very broad for the labs as well as the current customers we have. As Chris pointed out, I think here's the good thing in our model today: the core driver and engine of the revenue growth through the period that's currently forecasted is highly leveraged to the big four, but very diversified.
Like I said in my opening, we're talking about selling 15, 20, 30 products each to some of these big companies. So very deeply embedded in the infrastructure. That growth of capex, you know, is robust, but then we see the top taking more and more of that capex. At the same time, I think it's a real call option, some of the things that we're doing as well with the labs, but we would be embedded in that infrastructure no matter what.
Every model so far has been created, been trained using Marvell connectivity, as an example. That's going to be broadly used across all model makers around the world. The upside could be, you know, they look at more custom, other types of solutions. That's an opportunity as well. We go to over there.
Great, Mark Ever ISI, thanks so much for the great presentations, really appreciate it. Really appreciative of you guys taking a stab at what you think about your business in 2030. I think, you know, that's something we all try to do, and it's all appreciated. The transparency is really appreciated.
At the same time, it's really hard to go out that far. And I think, you know, if you look at, you know, historically, you know, when you have high growth markets like this, your customers will often make a guess at their demand and give you a forecast that is more than they want, so they get what they need. And I'm wondering if you could just talk a little bit about how you think about managing that risk.
And what's different this cycle versus, you know, last kind of super cycles? And I'd love to hear Matt and Dan, you've been around for a while, you have a perspective on that. I'd love to hear that. Thank you.
Matthew Murphy 3:13:38 ↗
Yeah, great. No, thanks. I'll start off and then I'll let you comment. I think a couple things. And yeah, you're right. We've both been around the block, you know, almost 33 years in this industry and kind of the front line. So definitely seen the cycle, touched the stove before. I'm not saying I'm perfect, but I understand.
A couple things give us comfort on this. The first is that while we're framing this a five-year, and we are, right, we're taking, last year was our first full, last fiscal year was the first full year to start the measurement period. Go out years, typical, but think about we have, right now, $20 billion in revenue. Next year, 2027, our forecast is out 2030. So you got, and we feel really good about next year. So then you got to go 27, 28, 29 to 30. That's it.
So it's not, you know, I'm actually not calling the ball like years from today. I'm calling the ball with 27 to 30. I think we feel really good about that. Now, look, we got it. We're all exposed if the world turns upside down and something happens. And I can give those, like anyone would always do in the semiconductor industry, well, what if something big happens? But at the current moment, we don't see that.
These programs are completely on track. AI investment continues to be extremely robust, outperformed all our expectations. When I go back to April 24, market thought, right, now continued to run. And I think built in from a hedge point of view for us, one, we have capex moderating at the end of the period, and then the other thing is we've given you a range, right, a range of outcomes, which I think should give you, you know, a sense of where we can land.
But if you just take the 20, even to the low end of the range, it's not... So I know these numbers are big, and I think you're going to try to catch up. But as Chris said, like we've had our head wrapped around what this could look like for a while. We got the supply chain lined up. So we have a lot of conviction on what we can go do, but certainly market dependent and certainly dynamic. But I don't, I think kind of pick not a year you want anything.

14 more exchanges in this transcript

Sign in free to read the rest of this interview. No card required.

Sign in to read the full transcript

Cite this transcript

APA, MLA, BibTeX
APA

Murphy, M. (2026, October 9). Marvell Technology Investor Day 2026 | AI Custom Silicon & Electro-Optics | Oct 06, 2026 [Interview transcript]. i101. CEOInterviews.AI. https://ceointerviews.ai/interview/3204931/

MLA

Matthew Murphy. "Marvell Technology Investor Day 2026 | AI Custom Silicon & Electro-Optics | Oct 06, 2026." i101, 9 Oct. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/3204931/.

BibTeX
@misc{murphy2026_3204931,
  author       = {Matthew Murphy},
  title        = {Marvell Technology Investor Day 2026 | AI Custom Silicon \& Electro-Optics | Oct 06, 2026},
  howpublished = {Interview transcript, i101. CEOInterviews.AI},
  year         = {2026},
  month        = {oct},
  url          = {https://ceointerviews.ai/interview/3204931/},
  note         = {Speaker-attributed transcript with timestamps}
}