Marvell (MRVL) is a semiconductor company addressing the storage controller, mobile and wireless and networking markets. Marvell reported second quarter earnings on August 22. The financial results and Marvell's recent announced wins of slots in new, mass-market devices demonstrates its likely long-term value to investors.
The last time I wrote (Marvell's Opaque Second-Half Opportunities) I explained how CEO Sehat Sutardja's predictions of increased application processor chip sales in Q2 were hard to analyze because, as a chip supplier, Marvell usually cannot announce its specific device wins in advance of the device releases.
The list of device wins announced since that time has been impressive, and mainly came within the last month. Chips ship before devices are released, so it is fair to assume that much of the 10% sequential revenue gain in the second quarter came from these and other device wins. In no particular order they are: the 7" Samsung Tab 3 tablet, Google Chromecast, China Mobile's first self-branded smartphone, Lenovo smart TVs and Yulong Coolpad smartphones.
A few years ago the smartphone and tablet world was dominated by Apple. As Android-based devices began to dominate the smartphone market and picked up market share in tablet devices, Qualcomm (QCOM) was the dominant chip supplier, with Marvell and many others trying to keep alive in the space. Marvell had acquired Intel's (INTC) ARM-processor cell phone division in 2006, and at first that division was tied to the fate of BlackBerry (BBRY). But just as Intel is still alive in the tablet and smartphone space (with Atom processors) largely because it has so much cash from its CPU operations, Marvell remained profitable during the recession based on its data storage controllers (where it dominates the market) and its networking and wireless chips.
Marvell stole a march on competitors by working closely with China Mobile to develop smartphone chips specifically for the Chinese market. That strategy has had its perils, but it has paid off and set up Marvell for great things in the next several years. There are low-cost, low to middle range smartphone chip challengers in China, and there are high-end rivals like Qualcomm and Apple, but Marvell has stayed ahead, innovating while keeping costs in line with what is appropriate for the Chinese markets. In China carriers do not heavily subsidize smartphones, so price is a key issue.
Now even as 3G networks are building out in China, LTE is about to be introduced. While Qualcomm is a formidable rival, LTE is going to level the playing field, at least a bit, since Qualcomm's CDMA patents will no longer be an issue. Marvell has put together an array of smartphone chips that will allow OEMs to build phones that can hit price points that are attractive both in the developing world and in the U.S. We should begin to see these phones in the U.S. in the second half of 2014. They will run the Android operating system. They will have Marvell's superior wireless communication technology for both 3G and 4G. They will be made in China (just like iPhones), they will be reliable, and they will be reasonably priced. I would bet they will sell through just fine.
The Q2 numbers confirm the story. It is important for more detailed analysis to note that Marvell's second quarter fiscal 2014 ended on August 3. That means it caught some second-half inventory build at OEMs in July. Q3 is typically seasonally up as chips go into devices meant for holiday sales. Q4 and Q1 are typically seasonally down.
Revenue was $807.1 million, up 10% sequentially from $734.4 million but down 1% from $816.1 million in the year-earlier quarter. GAAP net income was $61.8 million, up 16% sequentially from $53.2 million, but down 36% from $96.2 million year-earlier. GAAP EPS was $0.12, up 9% sequentially from $0.11, but down 25% from $0.16 year-earlier.
Excluding stock-compensation expense $41 million and other non-cash items, non-GAAP EPS was $0.23, down a penny from $0.24 a year earlier. Marvell continues to heavily invest in R&D, with GAAP R&D at $293 million in the quarter, up from $265 million a year earlier.
Marvell's stock has been, in my opinion, undervalued for about 3 years now because of the thesis that SSDs (solid state drives) would totally replace HDDs (hard disk drives) for data storage. Obviously that has not happened. Marvell derived 52% of its revenue in the quarter from the storage controller chip segment. HDDs are still going strong and Marvell has emerged as the leading maker of controllers for SSDs. Sehat specifically addressed a new anti-Marvell thesis: that NAND flash makers would make their own controllers and squeeze Marvell out of the market. That has not been happening. In fact, the best prediction is that Marvell eventually will dominate the SSD controller market the same way it dominates HDD controllers.
One of the big secrets of SSDs is that they are actually less reliable than hard drives. Go to any good tech site review of an SSD and you will see reliability is a major issue. This goes against our usual experience that semiconductor chips are reliable and moving parts like hard drives are much more likely to fail. Typical NAND cells are rated to be reliable only for about 3000 write cycles. To make the SSD last without data corruption, the controller chip has to look for failing cells and then stop using them. Fortunately many cells are written to once, say when the operating system files are installed, and then experience only reads of data. But some sectors of the drive may get written over fairly frequently. Even at one write per hour, you can expect cells to start failing regularly within a year.
Marvell's controller chips use a variety of methods to make sure the SSDs don't fail. Having an inferior controller chip results in higher failure rates. The better tech reviewers have software that tries to make the NAND fail by trying to rewrite to the same cells frequently. If the reviewers find the controller is not doing its job of spreading writes across cells and blocking out cells that have already had too many writes, they are going to let people know. Marvell has been gaining market share, showing SSD manufacturers understand this critical issue.
Of course, the thesis that Marvell will dominate is based on the idea that it will extend its current lead by continuing to innovate faster than competitors.
Guidance for Q3 is for another sequential revenue ramp, this to between $850 and $890 million, with corresponding non-GAAP EPS in the range of $0.23 to $0.27. While I would expect Q4 to be seasonally down, I think it is likely that Marvell will reach $1.00 per share in non-GAAP EPS for calendar 2014. If you can think that far ahead, you might consider the August 23, 2013, closing price of $12.19 to be well under what the stock will be priced at when (or if) my 2014 EPS estimate proves out.
Note too that Marvell pays a dividend of $0.06 per quarter, and has had an aggressive share buy-back program that will continue as long as the stock is undervalued.
One wild card in the Marvell story is the patent litigation initiated by Carnegie Mellon University. A local jury awarded $1.2 billion in damages last year. I think the jury was sadly lacking in the capability to understand the issues of HDD controller algorithms, and just went with the local hero. I believe the jury decision will be overturned on appeal, but of course can't be certain of that.
As in all investing in technology companies, there are a lot of very smart engineers and business people out there, working for a lot of contenders. Nothing about the future is certain. The competition is particularly fierce in the smartphone chip market, where Qualcomm has both market share and a significantly larger R&D budget than Marvell, and Intel, which has very deep pockets. I would not write off Apple yet, and Nvidia and many other companies, including upstart Chinese companies, are all out there trying to find niches to stay in the competition.
Nevertheless, I believe in the long run Marvell Technologies has a strong chance of dominating SSD controllers and a fair chance of dominating smartphone chips. Being strong in SSD and mobile chips would also mean strength in tablet computers.