Marvell – Great Long-Term Secular Drivers, But Sentiment A Threat

This has been a challenging year for Marvell (NASDAQ:MRVL). As fears have grown about a big reset for growth semiconductor stocks in 2023, Marvell shares have lagged the SOX index this year, falling more than 50% year-to-date and underperforming about 10%, and trading down more than 15% since my last update.

I find Marvell shares a little more challenging to evaluate now. I love the company’s long-term leverage to growth opportunities in the data center, particularly as hyperscalers move to 400G and 800G in the coming years, and I also like the exposure to 5G and enterprise networking. What I don’t like as much is the valuation and the level of expectations heading into this next quarter, as sell-side estimates for FY’24 revenue range from $6.3B to $7.8B and this is still a consensus “buy” call.

I do think Marvell is a high-quality growth name, and I think paying up for growth is fine, but I also see more risk than upside to sell-side estimates coming out of the next quarter, and I’d rather risk missing out than buying in ahead of that potential reset.

Not All Data Center Exposure Is The Same

Among the many worries about semiconductor demand in 2023 is an expected slowdown in data center investments. With a weaker growth outlook, the market cares more about cash flow now, and that is likely to lead to more restrained investments in new capacity and a slower growth rate in 2023.

A slower growth rate is not the same as a decline, though. What’s more, Marvell has strong leverage to high-end data center customers that are likely to continue investing in areas like 200G/400G/800G upgrades. As I discussed in reference to Broadcom (AVGO), that means healthy demand for high-ASP, high-margin components, and I expect Marvell to see strong demand for its PAM-4 DSPs, 400G ZR interconnects, SSD controllers, and custom ASICs, as well as emerging opportunities like active electrical cables (or AECs) and compute express link (CXL).

I expect overall data center spending to slow to low double-digit growth in 2023, and I don’t rule out a decline into the single-digits if the economy slows more than I expect. For much of Marvell’s data center business, though, I think 20% spending growth is still a plausible expectation, and I don’t see a tremendous amount of competition in these core high-end products.

Marvell’s Overall Mix Isn’t So Risky

Looking at Marvell’s share price performance, you’d think the company was more leveraged to areas likely to see a sharp deceleration in 2023, but I don’t believe that to be the case.

I do see weakness in the consumer sector (PCs, mobile, consoles, et al), but that’s less than 15% of Marvell’s business now. I’m less sure what to make of the carrier infrastructure market over the next 12 to 24 months. Deployments have been delayed by component shortages, and I would expect healthy catch-up deployments in 2023, but I’m not sure how much content Marvell has already sold through to these customers. Beyond an uncertain 2023/24 environment, I do see longer-term upside from Marvell’s exposure to Samsung.

Enterprise network is another market where I expect weaker overall trends in 2023. I like Marvell’s leverage to switching and multi-gig PHY, but I do have concerns that enterprise could shrink at a double-digit rate off of a high starting base.

There are a lot of product segments within Marvell that are relatively small today, but that I think could become more significant contributors over the next three to five years.

Marvell’s custom ASIC business is only about one-tenth the size of Broadcom’s, but the business is growing nicely and I think this will be a meaningful growth driver for Marvell as hyperscalers look for high-performance offload options.

CXL is a new business for Marvell, acquired with the Tanzanite deal, but as hyperscale loads increase, memory performance limitations become more problematic. My understanding is that CXL interconnects memory over PCIe fabric and effectively allows CPUs, DPUs, GPUs, and accelerators to share DRAM, debottlenecking that limitation.

With AECs, Marvell has a business that complements its existing strengths in 200G+ speeds, allowing data center operators to use more cost-effective copper cabling for 100Gps/lane products – traditional passive direct attached copper cables can’t support the higher requirements for signal integrity, and active optical cables are more expensive.

The Outlook

I believe that Marvell is going to produce strong revenue growth for this current fiscal year (FY’23), but see a sharp slowdown in FY’24. That’s not the consensus view today, and my FY’24 revenue estimate is $6.3B versus a sell-side average of $7.1B. This is my biggest concern about the stock right now – the market certainly seems to be pricing in a slower outlook, but I think there could be downside risk to sell-side estimates and sentiment over the next two quarters if/when management guides to that slowdown.

I still expect strong double-digit long-term growth from Marvell (around 11% annualized), driven by the company’s high-end capabilities in data center, carrier infrastructure, and enterprise and complemented by those emerging opportunities I just mentioned. I do see a risk of a year-over-year decline in margins in FY’24, but I think the business will rebound quickly and 40% non-GAAP operating margins are not that far away. While adjusted free cash flow has been hampered by high stock compensation expense, I do expect significant adjusted FCF margin acceleration over the next decade, with long-term margins in the 30%’s driving very strong FCF growth (from less than $1B this year to close to $4B in the 2030’s).

The Bottom Line

Despite the weak recent performance of the shares, Marvell doesn’t screen out as particularly cheap yet, and that’s a time when many high-quality semiconductors are trading at meaningful discounts to fair value. I do expect growth names to trade at premiums, but I’m less comfortable paying a premium when I still see a real risk of further guidance/estimate reductions.

If there is a meaningful reset to expectations and Marvell shares move back into the $30’s, this is definitely a name I’d come back to. As is, I find the investment case more equivocal – I think longer-term investors will be fine, but I do think there could be some turbulence along the way and I would personally rather risk missing out than jump in ahead of the reset that I’m expecting.