NVIDIA (NVDA -3.87%) has been all the rage this year. The company's graphics processing units -- traditionally used for high-end video game graphics -- are picking up steam in data centers. It acquired complementary networking hardware company Mellanox, announced its intent to also acquire chip licensor ARM Holdings, and announced new data processing units (DPUs) to further its momentum in the massive data center and cloud computing industry.
But while NVIDIA remains my favorite semiconductor stock, it won't be the only game in town as digital data continues to grow at a fast pace. Marvell Technology Group (MRVL -2.64%) is making moves of its own and looks like a relative value, so I recently made an initial purchase.
Image source: Getty Images.
Big data isn't going anywhere
It's been more than a decade of rapid cloud computing development, and now the advent of 5G mobile networks is accelerating the growth of digital data during the pandemic. And in the decade ahead, other technologies that create and require the use of even more information will develop -- including AI, network connected industrial equipment, and advanced driver assist and autonomous vehicle technology.
Marvell has a hand in all of the above. The company has a diverse portfolio of chips, including networking equipment and mass storage devices for data centers, 5G and communications infrastructure, and intra-network chips for the complex systems built into most cars these days. Marvell also makes custom designs to handle customer-specific tasks involving AI, as well as infrastructure processors like the above-mentioned DPU chips NVIDIA will begin selling.
This portfolio makes Marvell an ideal partner for enterprises looking to upgrade their operating systems for the modern era. Unlike graphics processors (GPUs) designed to accelerate the computation of lots of data with a single set of instructions, DPUs are similar to a CPU (like the chip that handles basic tasks in your PC or laptop) in that they can handle multiple tasks at once -- but are more specialized than a CPU as they can efficiently move very large packets of data.
In a world where complex systems need to move information around very quickly (like in 5G network infrastructure, a data center, or an autonomous vehicle), I think Marvell's DPUs and related hardware will be in high demand.
A relative value on a solid chip portfolio
Marvell made a couple of acquisitions to build out its suite of data management and processing gear, including the takeover of data processing chip outfit Cavium in 2018 and 5G hardware company Avera Semiconductor in 2019. According to Marvell, its resulting lineup of products and supporting software has a total serviceable market of some $16 billion in annual spending, and it expects that market to grow at a 9%-a-year pace through 2023. With revenue of just $2.8 billion over the last trailing 12 months, that leaves plenty of room for Marvell to muscle its way further into the market.
Even after going shopping, the balance sheet remains in good shape too, with total debt of $1.44 billion and cash and equivalents of $832 million. And as it laps acquisition costs, free cash flow (operating cash flow minus capital expenditures) is also rebounding -- running at positive $420 million over the last 12 months (good for free cash flow margin of 15%), even though some of its key customers in the auto and industrial sectors have been deeply affected by the pandemic.
Marvell is showing signs of accelerating, though. Revenue grew 8% during the first half of the year, but management forecast a 13% year-over-year increase for Q3. Shares trade for nearly 70 times trailing 12-month free cash flow, a seemingly steep premium that accounts for the purchase of Avera and other pandemic-related disruption. But Marvell stock trades at a steep discount to high-flying NVIDIA using the price-to-sales and price-to-book value metrics.
Data by YCharts.
At this juncture, I think Marvell trades for an attractive price and is another way to play the rising demand for computation and management of massive amounts of data. I recently purchased a starter position (less than 1% of my total portfolio) and will buy more of the stock if it is able to capture some of the recent momentum in the data center, 5G, and connected industrial equipment markets.
