It’s not really fair to say that Marvell (MRVL) is “doubling down” on its growth opportunities in the data center and 5G, but management definitely structured their analyst day on October 8 to highlight Marvell’s transformation into a growth story. While this will come at a modest cost to margin (historically an underappreciated driver of semiconductor stock multiples), this is a growth-driven market today and the messages coming out of the analyst day shouldn’t do any harm to the already-robust multiples for these shares.
Strategically I like what Marvell is doing. I like the opportunities to leverage Cavium IP for data center DPUs and likewise the opportunities to leverage its Cavium and Avera IP into custom ASICs. Auto Ethernet opportunities are shaping up, and 5G should provide strong growth for several years. The issue is the price and the multiples. I was already expecting double-digit revenue growth over the next five years (and close to 9% growth over the next 10 years), and the presentation didn’t support a strong outlook for margins. At over 31x ’22 EPS and 26x ’23 EPS, it’s tough to say that the growth potential is going unappreciated.
An Accelerating Top Line
Marvell management raised its growth target from mid-to-high single-digits to 10% to 15%, but a lot of analysts and investors were already in that ballpark. While management did maintain a mid-30%’s-plus operating margin target, the gross margin target is coming down a few points due to the greater anticipated mix of ASICs and semi-custom in the revenue line.
Marvell is focusing on three primary drivers for this well-above average growth – data centers, 5G, and autos. Management also highlighted a strategic shift from a “fast follower” approach to a “first mover”. That’s a little hyperbolic (I think Broadcom (AVGO) would be amused by Marvell asserting itself as a first mover at 5nm), but I do still believe that it is an important shift in the company’s approach. It’s a move that carries more risk, but if you want to maximize your revenue growth potential, you can’t really wait around to see what your rivals are going to do next.
Multiple Targets In The Data Center
Seemingly every semiconductor company wants a piece of the data center market, and it’s not hard to see why with hyperscaler customers seemingly willing to pay exceptionally rich prices for bleeding-edge performance.
I like the breadth of the opportunities Marvell is targeting. Data center DPUs will be a very competitive space (Nvidia (NVDA) among others), but Marvell can likely get some traction with its Cavium LiquidIO. I also like the shift in focus away from PC storage toward DIY SSDs, and I think the 30%-35% market share target in cloud DC storage is at least credible (ambitious, but credible).
I’m also still bullish on the custom ASIC opportunity, as Marvell looks to leverage technology from the Cavium and Avera deals and jump from 14nm to 5nm. I don’t see Marvell displacing Broadcom in custom datacenter ASICs, but Marvell scored a recent win over Intel (INTC) at 5nm and I think a lot of Marvell’s would-be rivals will have a harder time making that leap.
5G Developing Nicely
Marvell management believes they’ve secured “full platform” wins with two Tier 1 basestation customers (Samsung and Nokia (NOK)), and that includes transport processors, baseband processors, massive MIMO radio processors, PHYs, and assorted switches and components). While the value of winning Nokia’s business is perhaps debatable given the execution challenges that company has had lately, Samsung has been gaining share and recently won a multiyear 5G RAN deal from Verizon (VZ) worth about $6.6 billion. What’s more, with Huawei increasingly unwelcome in several large markets, I see meaningful scope for additional wins.
Marvell has a diverse customer line-up for 5G. In addition to those large-scale platform wins, Marvell has lesser design wins with two of the other five Tier 1 providers, and has also been securing embedded processor wins with Tier 2 providers.
Auto A Nice Add-In
Marvell continues to add wins for its auto Ethernet business, now having 24 OEMs wins, including seven of the top 10 OEMs. Content growth from Ethernet adoption should be significant, growing from less than $50 (as low as $5 in some cases) to $100 or more. As a reminder, auto Ethernet is a way of managing the increasing bandwidth demands of various auto systems (infotainment, safety, monitoring, et al) without adding to the wire harness/cabling requirements of a car. As wiring/cabling is typically the third-heaviest component of a car (after the chassis and engine) and hybrid/EV designers are keenly focused on lightweighting to improve performance and range, Ethernet is an attractive solution to that bandwidth-vs-weight issue.
I liked what Marvell had to say at its analyst day, but honestly … not much of it was new to me, and it doesn’t really change my view or my model that much. I like the “philosophical” shift and the sense that management will be more aggressive in pursuing growth, but my view on the company’s opportunities isn’t much different. I will say, though, that I do believe management did a good job of highlighting the value it is leveraging out of its past deals, and I do think the company has an attractive IP portfolio – if Xilinx is buyable at around $30B, Marvell too could be a target for a larger semiconductor company that wants to boost its growth potential and maybe leverage some gross margin and operating scale.
The Bottom Line
Absent some tweaking (mostly on the gross margin and operating expense lines), not much really changes in my model, nor in my valuation. And therein is the problem, as Marvell was already trading at a robust price relative to my cash flow and margin-based fair value estimates. I realize growth semiconductor stories (Nvidia, Inphi (IPHI), Lattice (LSCC) and others) trade by a different set of rules, and I do still see some potential for outperformance to drive higher estimates. But with Marvell already trading at an Inphi-like multiple on FY’22 EPS, it’s tough for me to see these shares as undervalued.
Disclosure: I am/we are long AVGO. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.