Okay, great. I think we're live. Good afternoon, everyone. Thank you so much for coming. My name is Toshiya Hari. I cover the semiconductor space for Goldman. Very psyched, very honored to have the team from Marvell with us this afternoon. We have Matt Murphy, Chairman and CEO. You have Ashish Saran, Vice President of Investor Relations. First of all, thank you so much for coming, guys.
Oh, thank you.
Appreciate it. Yeah, thank you.
Thank you.
We're psyched to be here, too.
Good stuff. So Matt, on your recent earnings call, you spoke to your AI business potentially exceeding your prior targets, fiscal twenty-five, fiscal twenty-six, and I know you're not giving specifics here, but I was hoping you could kick us off by speaking to how your outlook has evolved. You had a big event back in April. How have things evolved, and, you know, what are some of the key drivers that you're focused on?
Yeah, no, great question, and happy to set the context. I think it, it's actually kind of unbelievable if you think about, where we were, and I'm gonna go back to May of 2023, so not that long ago, right? A little over a year ago. We took our time, but we sized our AI opportunity at then, and we said, "Look, we for fiscal 2024, which was last year, could do $400 million in revenue, and then $800 million for the year we're in." That was in May. And then, you know, as the quarters went on, you know, we kept overachieving and outperforming to that. When we had our AI day in April, we sized it for this year then at $1.5 billion, right?
So just in not that long a period of time, really less than a year, effectively, the opportunity had over doubled, or about doubled in size, and then we sized for next year, our fiscal 2026, $2.5 billion. So that was sort of the step up. In our most recent earnings call, we said, "Look, business is strong, both on the custom side, as well as our optics side, and so that we would exceed the $1.5 billion for this year, driven by both of those, by the way. And then that's also gonna carry over into next year." So it's been evolving really nicely.
The other thing that happened was earlier in the year, because the demand ramp had been so strong, especially in the optics initially, we had a bit of caution in terms of what the slope of that ramp could look like as it went up so fast, and the other thing we said in the recent call was that actually, every quarter this year, we foresee our optics revenue to be up sequentially, so Q1 to Q2, Q2 to Q3, and even Q3 to Q4 as we build backlog, so business momentum is very strong. Production ramps are going really well on the custom silicon side.
We have a great setup into next year and, you know, a very compelling technology roadmap that I think is really resonating well with our customers in terms of the long-term positioning we have at Marvell for this really once-in-a-generation type opportunity.
That's great. Thank you. A couple of questions on the custom compute side. Again, at your AI event, you laid out a long-term growth outlook of 45% for the accelerated custom compute market. And in this outlook, I think you're assuming that custom processor grow from 16% of the market last year to 25% in 2028. I'm sure a bunch of work went into this, the analysis. How did you and your team land on the 25% number? Do you see custom processors competing effectively with GPUs in external workloads, or do you expect custom processors to mostly address internal workloads?
Yeah, great question. So a couple things went into that. The first is, we said, and just to kind of give you a range, we said 20%-25% would kind of be the range. And we, and we don't know, right? You could sort of debate around the edges, but it's a big number, and, and we, we, we continue to have very strong conviction in the growth of that market and that sort of TAM materializing over time. We arrived at that number, that range, through a couple different ways. You know, one is obviously using industry analyst resources and, and data. We built a pretty, I think, solid and comprehensive bottoms-up model, you know, just based on the programs we know about that we have and also that are out there, and what the magnitude of those look like.
But a key part of that was we didn't put really a lot of penetration at all comprehended in the external workload side. So for us, if that were to materialize, which is the aspiration, I think, of a lot of our customers, we'll see how they do, then that would be an upside scenario to Marvell, 'cause that would just sort of increase the percent towards the high end of our range or even higher. And, you know, we've been getting this question a lot just in the one-on-ones that, you know, aspirationally, I think our customers talk about wanting to increase that percentage. Then there's just the reality of timing and when you can get there and how many workloads are gonna be available.
So our view is that was a pretty thoughtful, judged view and certainly has a lot of upside to it, potentially, depending on how things play out.
Okay. That's very clear. Thank you.
But I mean, just to kind of-
Mm-hmm
... step back and say, you know, like we were talking in another one-on-one about, we got into this custom business through the acquisition of an asset from GlobalFoundries called Avera Semiconductor, which spun out, which was the original IBM custom microelectronics team. And, you know, I don't think anybody could have foreseen that that acquisition would have opened up a $40 billion accelerator TAM, where we could actually go get meaningful content in. So it's amazing how much has-
Mm-hmm
... happened in five years-
Mm-hmm
... if you think about it. So when I think about the next five years, I really don't wanna cap it, but that's the current view we have.
Okay, really helpful. Maybe a good segue into my next question, which is on sort of the competitive differentiation you guys have, again, in custom compute. You have a credible competitor here in the US, but you also compete with, you know, smaller players in Asia. How would you describe the sources of your competitive differentiation, and what factors typically sort of, you know, drive design wins, or not, in custom compute?
Yeah. I think the industry trends are really what are in our favor, and I would say would favor large, scaled-up, highly integrated suppliers to go do the job. That's basically born out of the following. Now we're deep down the road, right? We're ramping two projects right now. We've got more in the pipeline, so we kinda know what this takes now. It's a combination of things where, you know, it's not just having one piece.
So it's not just like having, well, I have the best, like, kind of, single piece of IP over here, or I've got a good set of engineers that can do a really good job on the back-end, or, "Hey, I have a packaging team, a packaging, you know, third party." So there are third-party ways that people might try to get to the final solution, but when we look at what it really takes to ramp one of these into volume, it's the combination of, quite frankly, front-end design resources, world-class back-end resources, leading-edge package technology, massive scale relationships with suppliers, including not just for nanometers and process nodes, but also the packaging side, substrates, the ability to have the OSAT capacity to actually ramp in volume, the product and test, the manufacturing engineering personnel that can drive yield improvement, the quality and FA people.
You start layering all these things in, and the ability to invest big time in not just one process node ahead, but really two, and drive process, I/O, IP, packaging, all these things together in concert. Our view is that in the end, when the TAM sort of evolves and what is shipping in volume, I believe it's gonna come from companies that have an integrated offering here and can deliver it, and can provide the confidence to these customers that we can actually get the chip into production, shipping with high quality and high reliability, deliver the TCO they need, and already be on to the next one and the one after. That's our take on the whole market. Of our large competitor, each has great capabilities, technologies. It's always a ferocious battle.
Both very good companies, but I think the market, if it goes from whatever it is today, $7-$10 billion to $40 billion, and you look at the sort of evolution, it's gonna be a, I think, a great opportunity if you've put in the work and you've got the basis now to compete, and really it's come down to a couple companies, in our view, that can do that.
Okay. Great, makes sense. At the event, you guys talked about, you know, three customers, four projects, you're ramping on two. I'm curious what kind of visibility you have into the projects, particularly the ones that are on the come, and I guess medium to long term, what kind of market share do you all aspire to as it pertains to accelerated custom compute?
Yeah, we have a very, I think, you know, very solid seat at the table, I think on pretty much every major opportunity that's out there today, and again, that's really not only 'cause we're executing on current programs, but I think on the basis of our, I think, our industry-leading and compelling roadmap. You know, we get the meeting, we get the opportunity. Now, we gotta go close it, and we gotta go win it, but tremendous amount of design activity, that's been happening in this area, especially since the GenAI thing hit early last year, so that's sort of continuing.
Our Marvell president, Raghib Hussain, who's been with me since the Cavium acquisition, he was co-founder there, and he really, and I together, kind of had architected this whole data center strategy for Marvell and the pivot here, and, and he helped drive a lot of the M&A, and now he's running the big team, running custom. I mean, he stood up there at the AI day and said: "Look, we got a $40 billion TAM, and, you know, I would be disappointed if me and my team couldn't drive at least 20% market share in that area." So he set a bogey, you know, as a long-term target for where we could get to. So if you think about 500 million-ish this year is what we signaled, right? One third of the 1.5 billion, kind of plus.
You know, next year we had signaled that it would be like $1 billion, but now that's kind of a plus because it's overachieving. Sometime in the future to, you know, high single billions. I think it's a great aspiration for the team, and I think it's achievable if you look at what some of these other big programs have been able to to deliver over time. So that's kind of our guiding light, you know, at some point. We think the TAM is big. We've got the programs in place. We don't know exactly how this is all gonna play out in terms of which program's gonna go to what volume and the exact timing.
And along the way, I know our investors, because it's such a competitive area, there's a lot of noise in the system at times, we tend to just kind of try to look through all that with the first premise I talked about, which is what it's gonna really take to ship the chip in the volume and make it happen, and we think we're gonna be there. And then I think we've set a reasonable share target. So if the TAM materializes, which it's on its way to do, we should be in great shape-
Mm-hmm.
and have a great growth business in front of us.
Got it.
And if it can upside along the way, that would be fantastic, because there's other things that may come into play. Like you said, how is this monetized? What happens with the workloads? So we're just gonna let that play out, but right now, we're heads down on execution, and making sure, again, we have the best-in-class technology roadmap for accelerated computing from a platform offering for our customers who have their own silicon ambitions in this area.
So Matt, to your point, the timing of projects, the volume of projects, there's always uncertainty. When you and Raghib and sort of the broader team sit down and sort of think about various projects, and particularly new projects, what decides whether or not you pursue a project and follow through with the project or take the project, if you will? Is there a margin bar or a ROIC target? What determines, you know, that decision-making process?
Yeah. I mean, we look, you would expect we have a very thorough and sophisticated suite of financial metrics, right, that we bounce, quite frankly, all of our investments against, and they would be traditional financial metrics as well as some specific Marvell internal ones. But in the end, you know, we do try to make a judgment call on risk, on are we really delivering the value on the project or not? I mean, can we really bring the best of Marvell, or is it more of a job shop type of thing that maybe won't go anywhere? You know, we try to triangulate from an executive standpoint, what's the seriousness of it?
It's not really a factor anymore, but, you know, geopolitics played a role over the last couple of years in terms of geographically, which projects you might want to take or not. You know, we ended up indexing and basically focusing on the North America customer base and not internationally. I think that was a good call, given a lot of the export control tightening that's happened. So that was a conscious decision, as an example. We could have done some of those projects. It turns out we're glad we didn't, just because that would have been out of our control. So we do think about those things, you know, and I would've never thought as a CEO that I took this job eight years ago, that, like, geopolitics was gonna be a factor in, like, which project I do or not.
I mean, it was just incomprehensible even four or five years ago, but that is another factor, but look, in the end, these projects we wanna make sure at the end that these projects are, like, mission-critical for our customer, and so having that executive buy-in all the way, that it's not a sideshow or a science project, and I think all that's gone. I mean, it's too serious now. You can't just experiment anymore. You've got to go make it happen, and so, yeah, we feel really good about, you know, the position we've got at the moment.
Okay, great.
Yeah.
Maybe shifting gears a little bit, on the interconnect side of the business, you've got a really strong market position across the back end, the front end, DCI as well. In your long-term model, you're assuming a CAGR of 27% for the interconnect TAM. You're assuming flat market share off a high base.
Mm-hmm.
Any specific product groups or opportunities that you and your team are particularly kind of excited about and investing in, or is it across the board?
Yeah, well, look, I'd say that in each of the areas, we've got very dedicated teams that are driving really a best-in-class roadmap and execution. So certainly, the PAM team and optical DSP team has done, like, an amazing job in terms of the market share we've been able to get, but they're absolutely driving that roadmap, you know, at the bleeding edge. You know, the DCI team, you know, at ECOC a year ago, basically in October, we announced 800 ZR, you know, kind of way ahead of where I think the market thought it would be. They're heads down already on the next generation, right, which requires silicon photonics, coherent DSPs, complex module design. I mean, so again, we're not just working on the next one, we're on the one or two after.
And then you sort of have these really, I think, highly leveraged, call options for growth, where we've taken that same PAM technology and we've leveraged it for AECs, right? Which was really an emerging market that started a few years ago. It's very much a real market now. We're seeing use cases for that, that we hadn't even comprehended. And again, with the state-of-the-art AECs, those have all moved to PAM-based modulation, right? So that's sort of a leverage point for Marvell. And so we are now ramping those products into production. That's gonna be a very large market. We have a very good competitive position there. Same with PCIe on the retimer side. You know, again, highly leveraged from our PAM investments already.
So from a connectivity standpoint, we just have this, a very, very broad offering with a lot of shots on goal, both defending our position where we have product leadership, but it's very competitive, and then trying to drive a lot of innovation in these new categories, and even things that we haven't even talked about yet that are coming. And I think connectivity has become, quite frankly, one of the leading drivers of kind of how these data centers are architected and how these chip architectures end up manifesting themselves. It used to just be, do you have the nanometers? And then, by the way, what's the I/O? Just bolt on an I/O.
Quite frankly, now, it's not that the nanometers are trivial, but if you don't have the best in class, you know, IO and interface roadmap and be able to execute that, it's a very, very complex technology at very, very high speeds now. It almost doesn't matter if you have the big digital monster chip if you can't actually get all the IO to work. So that strategically, as a technology point for us, has become very relevant. Our customers all know that, and it provides just this very large TAM that, like, is growing 27% a year, you know, to a very large number.
Got it.
Really, really pleased with the positioning there.
Great. Great. To your point, Matt, the optical DSP team has executed really well, home run acquisition. I'm sure you get this question a lot from investors as well, but the competitive landscape is, you know, again, as of now, you guys are very strong. There are concerns out there in terms of market share from a technology disruption standpoint. How are you guys thinking about things like co-packaged optics, LPOs, the others that are working on optical DSPs to disrupt your position?
Yeah. Yeah, well, you know, certainly we're going to stay very hungry and very paranoid, okay? We're never going to take this position we have for granted. I think we're addressing it in a couple different ways. The first is, as I alluded to, kind of on the core pluggable DSP, you know, roadmap. We're just heads down to execute faster, better, in time, with the best technology over everybody. That team is just off to the races on that one. And that's ultimately, especially in a high cadence, high beat rate type of market, you really need that time to market. It's like it's just mission critical. So that solves a lot of problems from a competitive standpoint. It's not everything, okay? Then there are always, especially in this optical area, there's always these technology disruptive things that are lurking, that eventually manifest themselves, right?
Like, PAM was out there for a while, and, like, a bunch of companies tried, and then sort of Inphi became the winner, and then we acquired them. But even that took a long time, right, to make that NRZ to PAM transition. I mean, even when I was at Maxim, we had an NRZ-based optical transceiver business. I was running it, and this PAM thing was out there, and it was like: Yeah, it's going to happen eventually. And, of course, I was gone by then, so that wasn't my deal. But anyway, on the other technologies, we are investing in those. I mean, you know, things like, you know, linear optics, right, or CPO or silicon photonics. I mean, we're never saying these aren't going to ever happen.
I think what I just want to make sure is clear to the investors is, the preponderance and the bulk of the market in terms of revenue, shipments, you know, where the puck is heading for the next few years, foreseeable future, is going to be in pluggables. That's still what we see, and we talk to these customers all the time. But that doesn't mean we're not going to invest to make sure we're ready. So we'll be ready if there's LPO applications, right, where they don't need the DSP, but they need a really good broadband analog solution, we can be there, you know? Same thing on Silicon Photonics and CPO. You know, we are shipping in volume today with Silicon Photonics.
I mean, in all of our, you know, hundred gig COLORZ modules for DCI and now four hundred gig and eight hundred gig, that's a Marvell-specific design, silicon photonics PIC, with all the associated circuitry, external components, full module design, et cetera. So we're not afraid of those things. I think, you know... So I never want to come across as defensive. I think we're just trying to be objective in saying, like, where's the market going to be, and where's the bulk of the revenue going to be? It's going to be in pluggable optics, for sure. But it doesn't mean we're pooh-poohing other things, or we're going to not put our-- we're going to put our head in the sand. So we intend to have the most broad suite and complete portfolio for this interconnect opportunity, the one that's growing at 27%.
If one of those things hits in some big application because it's ready in time or it actually works, then we fully intend to be there. We're not taking that stuff off the table.
Okay, very clear. Maybe a question on Innovium and sort of the data center Ethernet switch opportunity. Here, you're more of a challenger-
Yeah
... if you will. I think you're in production with your 51.2 T product. Maybe spend a couple of minutes on the growth opportunity here, again, in data center, and what's your value prop relative to the competition?
Yeah. Well, yeah, we, as you mentioned, we got into this business, this higher layer switching for cloud applications and AI, through the acquisition of Innovium, which was a venture-backed startup that we acquired. Marvell had a switch, and has a switch, business, that actually came from an acquisition Marvell did in 2000. Some of you guys were around back then and remember Galileo, which, at the time, there was no cloud, right? So the market was really for enterprise and carrier class switching applications. Very strong team, very successful. When I became CEO in 2016, they were pitching me on, "Hey, we want to go do, you know, higher speed, you know, cloud-based switching." and the effort internally always looked just sort of insurmountable, given the strength of the competitor and the cadence they were on.
We made the decision, kind of post-Inphi, or almost in conjunction, that, you know, having the custom ASIC, the interconnect, and the higher layer switching was going to be kind of a key set of technologies that would be very important and strategically relevant, with the switching one being farther behind, obviously. We acquired Innovium. We got a great team. We introduced our first product with them this year. It's the 51.2 T product. We took our time to get it right, and we moved it all to Marvell Technology, so it was done on our design flow, Marvell SerDes. It was done with our, I think, very thorough advanced verification techniques and suite of things that we do. The chip worked right out the chute, and it's now going into production this year.
And there's a lot of interest in it. We're a challenger, for sure, but what's happening is, I think the whole higher layer switching market, as you've heard kind of from everybody, is definitely opening up in a different way because the AI use cases are somewhat different. There's new feature sets that need to be added. So it's not like you need to do one monolithic chip every four years and then just see who can get the chip out. So I think there's gonna be a wide variety of opportunities there. We have strong customer interest. We have a large, formidable competitor in that area, so we've got to go earn our way in.
and it's gonna take a little bit of time, but we're absolutely focused on this area and have a very strong roadmap, general manager to run it, and we've actually taken the whole Marvell team and combined it with the Innovium team, and the resources of that combined group now, which is a lot of engineers, are really focused almost primarily in this area. So we're investing here big time, and we're, and we think it's very complementary with our optics business, and again, kind of that system-level view with the ASIC as well.
Got it. Great. Maybe one last question on the data center business, perhaps on the non-AI side, in storage. That business went through a fairly extended correction. It's been, you know, on a nice growth path or recovery path over the past couple of quarters. Enterprise SSD, Nearline HDD, you know, customers sound fairly constructive at this point. What are you seeing in that business, and what are your expectations going forward?
Yeah. Now, we've seen a nice recovery in that business. It was one of the first businesses at Marvell to downturn in the economic cycle, which always happens, by the way. It seems like storage always kind of volunteers to go first for us for some reason, but they went first. They corrected, you know, at the beginning of calendar 2023, pretty severely, and that business now, really for the last probably six quarters or so, has actually, in a very kind of steady and somewhat programmatic way, just kind of grown, and what we said was, "Look, normalize that data center storage business from Marvell was about $200 million a quarter." It definitely overshot during the pandemic and the supply overhype and everybody being worried. But if you just sort of said normalized, our view was, like, $200 million a quarter.
It's definitely on its way back up, and it's pretty much grown, like I said, every quarter for six quarters, and sometime next year, I don't know when, we're not saying exactly, it's gonna get there. It's gonna get there, and those trends we're hearing are great, 'cause they do sound constructive, and certainly, you know, if that continues to strengthen, that's great, and if we can go do better than the 200 a quarter, that would be even better, but I think for modeling purposes and just the way to think about it is, it's on its way back up. Inventory is being digested, new things are ramping up, and our bottoms-up shows we're on our way.
Got it. Okay. On enterprise networking and carrier infrastructure, I hate to lump them together, but in the interest of time, I will.
Yeah, sure. Okay, I don't mind.
So good, good to see the, the stabilization. I know you've shared sort of normal, steady state revenue levels for, for both businesses. But what are your thoughts going forward? When do we get there for, for both businesses, you think?
Yeah. Yeah, so we lump them together just to try to make it easy for investors as this is coming out of its cyclical correction. But basically... The good news is that the first half of this year, which we thought was really gonna be the bottom, and quite frankly, had to be the bottom with how much those businesses had corrected, the good news is we signaled Q3 up, you know, mid-single digits for carrier and enterprise combined, which is nice and healthy, and it's good to see it coming off the bottom. Q4, we've already got backlog building, and orders have really picked up so that the growth rate from Q3 to Q4... We try to be helpful to investors on this, because to your point, people are wondering what's the slope back up.
That'll be better than mid-single, so that'll actually accelerate, so that's even more positive, then I think it's gonna keep chugging away, but to your point about their different markets for different reasons. I think the enterprise one is more just a balance sheet drawdown by our end customers, the channel inventory clearing, whatever needs to clear out with the customer base, and then that'll just kind of recover, and the commentary is pretty constructive, I think, out of the OEMs in terms of the end market looking better, so I think that one's gonna have its path. The carrier one, you know, definitely CapEx in that area has been depressed.
Now, that being said, inventory is being burned through, and we have some new content on the wireless side that's gonna roll in, really starting in Q4, and we have those orders now, and we have that backlog, and that's gonna roll into next year. So for kind of two different reasons, one is more market, just normalization, one is maybe assume the market stays weak, but we have our own kind of goodness coming.
Mm-hmm.
That's what gives us confidence that it's coming back. To the extent that everything finally clears, and we have a better economy, and people are spending money on this sort of thing, great. Maybe that's gonna be upside to the sort of billion each that we talked about.
Got it.
But the good news for investors, if you can wrap your head around the fact that this is the slope, and we're right, then basically, there's about another $1 billion that at some point, of annual revenue, that's gonna flow back into Marvell. 'Cause right now, these are at about a $1 billion kind of run rate today.
Yeah
... as of this time. And, you know, we're not gonna have to put a lot in to go get that. That's just more business that is gonna come back-
Mm-hmm
... needs to come back.
Right.
So that's. Sometimes your headwind always becomes a tailwind at some point, assuming it comes back, so.
Of course. Sounds good. Great. On gross margin, you're guiding October to 61%. I think, you know, you guys and Willem have talked about over the next couple of quarters, sort of expect that sort of range... There's a long-term target out there, 64%-66%. Is that still the right kind of relevant target? Or just given how mix is evolving, we should think of 61% plus or minus?
Yeah.
Yeah.
I'll start, and I'll ask Ashish to add after. But look, I think what we said on the call, and Willem said, too, is probably hangs out in this zip code for a couple of quarters, which I think in a way is kind of a win in that you've got custom ramping really strongly through this period, right? But you've got some of the other businesses coming back, too, and on higher volume. That's a helper, right, in terms of just the absorption being spread over a larger revenue base. The long-term target, you know, we're not ready to kind of adjust that yet for a couple reasons. The first is we're still in a cyclical recovery on things like storage, enterprise, carrier, so we don't really know where that's exactly gonna come back to. That's in the future.
The second is we don't know how big the custom's gonna be. What if they're really, really upsides? What is that gonna do to the mix? So how do you sort of take one off the table or not? And then, depending on the revenue scale we get, though, you're also gonna have ... if you're really upsided, you're gonna have a lot of revenue that's then gonna really make your manufacturing overhead absorption look even better, right? And so there's a lot of factors, so it's not like we're trying to be cute, it's just we don't know enough yet. So why, and this is, these are long-term targets we set. I think the main punchline, independent of the gross margin %, 'cause that's really mix dependent at this point.
Yeah.
It's really based on this custom thing. Second driver would be sort of the core business coming back, and the third, probably the absorption thing. But regardless of all that, when we look out to next year and we look at our plan, because of the outsized sort of leverage we get on the business that's rolling in next year, particularly on custom, the OM line we feel really good about, right? Because there's tremendous fall through as we get this revenue growth for next year. We don't have to spend to keep up with it. There's NRE that's coming in the door because of these custom programs. And the OpEx, we can control, right? So we've been going through our whole capital allocation, what we're gonna spend. So that's a knob.
You know, the top line looks really good, and so if the GM moves around a little bit, I still feel very good about where the OM is, and that's really what we're driving for next year, is that thirty-eight to forty target. Like, how do we get there? How do we get there as fast as we can? How do we sustain it there? And then we'll adjust around it, right? If GMs are up a little bit, we can pull a knob. If GMs are down a little bit, probably means we have more EPS, to be honest. It's a little bit of a weird thing, but the lower the Marvell gross margin goes, in some ways, it probably means there's a big fall through, 'cause the core business margins are holding very strong and are very consistent, right?
I think net-net, we would use, like, a future investor day, you know, whenever we would do that again, to probably say, "Okay, let's take stock. Let's articulate the future. Let's look at the OM and the GM in concert and the growth rate," and that's how we would think about it.
Makes sense.
Yeah.
Maybe in the last minute, on capital return, and M&A as part of that. You've obviously been quite acquisitive since joining the company. A ton of very good acquisitions. Do you feel like you've got the right portfolio, the right assets in place to pursue the long-term objectives? And, if you feel like M&A is still an option, what would you look for?
Yeah.
And from a buyback perspective, what should investors expect?
Okay. Yeah, real quick. Yeah, we did a lot of M&A, we did a lot of divestitures, a lot of churn organically to get where we got to, because we had to. I was not handed a beautiful gift, you know, like, in terms of, like, "Hey, awesome, you got this whole portfolio, Matt, of all this killer stuff for, like, the future, so go, just go run with it." We had to make this happen. We had to bootstrap this company from a basically sub-$3 billion enterprise value at some point to where it is today. And so we had to go chop all this wood, take all this risk, do all this stuff. But I said in 2021, "Look, we did what we had to do. We got the portfolio we want.
By the way, given the regulatory environment, I'm thinking there's this M&A is gonna get really hard going forward. "We got to control our own destiny." So our organic investments are really paying off. We've focused on the internal stuff so we can fund our future. And we're enjoying the benefits of not having done a bunch of M&A recently. Like, to your point, we're buying back stock. We're buying back a bunch. The board and I did, you know, $3 billion dollar authorization, right, just fairly recently. Willem's articulated that we've been increasing our buyback every quarter. We actually said we're gonna take share count down in Q3, which would be fantastic.
And the P&L looks much cleaner, 'cause a lot of the stuff that happens when you do M&A is starting to finally roll its way through, and we're generating a lot of free cash flow and able to really fund our future. So we have a very strong organic plan. We believe it can get us to where we need to get to. We'll always look for technology that can help accelerate that, no question. And if we had to do another M&A, we're actually good at it. We know how to go, like, do this stuff, but what's actually actionable? What makes sense? You know, that's all these question marks and, you know, there's not more Inphi's and Cavium's and Innovium's and stuff floating around.
So we, we've kind of had that recognition, like, three, four years ago, and we really like how we're positioned now, with the investments we've got. And so yeah, expect on the buyback, we're gonna keep going, 'cause we have conviction in our future.
Great. Unfortunately, we're out of time.
Okay.
Thank you so much for coming.
Yeah.
Yeah.
You're welcome. Thanks for hosting me.
Congrats-
I appreciate it. Thanks, everybody.
Thank you, everybody.