Marvell (MRVL) is an undervalued, underappreciated investment opportunity that offers 50% upside over the next year. The shares have been hit by a series of negative events recently, all of which should reverse over the next year. Removing these overhangs on the stock will provide ample catalysts for price appreciation. The shares are also supported by a truly massive buyback (17% of the total share count was bought back over the last year), a 2.2% dividend yield, and a clean balance sheet.
MRVL operates in three main business lines: storage, wireless, and networking.
- Roughly half of MRVL’s total revenue comes from storage.
- MRVL’s main products are controllers, which manage the reading and writing of memory on HDD and SDD devices.
- Gross Margin for storage is likely in the 60% range.
- 60% market share in the HDD controller market, selling into WDC (24% of total revenue), STX (10%), and Toshiba (10%).
- 50% market share in the “merchant” SSD controller market. 2/3 of the total SSD controller market is merchant, with the remaining 1/3 being “captive” production (Samsung).
- Leading edge designs in the new hybrid drive space that combines the HDD and SSD technologies into a single device.
- MRVL has been taking share in both HDD and SSD from LSI… although most of the share gains in HDD are winding down.
- The latest HDD share gains come from the 2.5”, 500Gb platter design, which MRVL has the only viable controller solution. These leading edge HDDs offer large storage capacity in a small format that is ideal for small laptops and Ultrabooks.
- 50% of HDD revenue is from non-PC end markets; this is up from 35% in 2008. Non-PC growth has kept HDD TAM flat, to slightly down, in the face of rapidly declining PC sales.
- They don’t break out the HDD/SSD split in storage revenue, but SSD is likely <15% of storage revenue. That figure will grow on the back of 60%+ TAM growth in SSDs, plus share gains for MRVL.
- Overall digital data storage is expected to grow at a 33% annualized rate through 2020. It isn’t reasonable to think that this growth will only come from cloud based storage solutions. The type of data that needs to be stored on consumer devices is shifting from text documents, to photos, to videos. With that shift, local storage needs are growing quickly. Both HDD and SSD will benefit, with a premium paid for solutions that are small and cost effective.
- We expect 6% y-o-y growth in storage revenue for FY’14 (ends Jan. ’14).
- Below are some interesting thoughts on the storage market from WDC’s most recent earnings call:
Keith F. Bachman - BMO Capital Markets U.S.
Right. So philosophically, if the market's down 5% to 7% on a sustained basis, you think you can have positive unit growth even against that PC backdrop?
Wolfgang U. Nickl - Chief Financial Officer and Executive Vice President
Stephen D. Milligan - Chief Executive Officer, President, Director and Chairman of Executive Committee
If we go back, Keith, I mean I think that there's a couple key points to emphasize. I mean, one, we are not anticipating a meaningful change in terms of what's happening in the PC market. We're not counting on that. We're being very cautious with regards to our production plans in that regard. The other thing is that we talked about, at our Investor Day, was a longer-term unit growth rate in terms of HDDs of 3%. That continues to be our current assumption, but that growth is not coming from PCs. It's coming from other areas.
Stephen D. Milligan - Chief Executive Officer, President, Director and Chairman of Executive Committee
So what's interesting in terms of what we've seen is that the issue in terms of hard drive TAM, right, because it's obviously down, it's down from 170 million pre-flood to 135 million this past quarter. Clearly, we've been impacted in that regard by lower PC sales. But it's not because of SSD replacements within PCs. In fact, we've actually seen that pure SSD as an alternative for hard drives has not changed significantly from what we expected or from what we've seen historically. Now you are seeing dual-drive configurations where you've got SSD in a hard drive. And so that -- but we're agnostic to that. That's okay from our perspective. The issue is that there are competing devices, for example, tablets, that are more compelling to customers that they're buying instead of PCs. And so if you go back to the 5-millimeter story or the hybrid story, what we're trying to do as a company is provide a more compelling storage alternative in a thin and light package that our customers can work around, design more compelling products to have them buy a PC or a PC-like ultraportable device and versus a tablet.
- Wireless is roughly 20% of total revenue; down from ~45% several years ago. The revenue decline came from the loss of RIMM as a customer in new models. MRVL designed the processors in a number of the hugely popular Blackberry phones in the 2005-2010 time frame. They were replaced by QCOM in the newer models in 2011, just when RIMM started to fall apart.
- MRVL designs ARM-based processors for various smartphone and tablet functions. The true value-add is an all in one solution that offers various processor and connectivity solutions in one package, at an extremely attractive price. This has been very popular with white label Chinese smarthphone manufacturers.
- Gross Margin is likely in the 25%-35% range, depending on the end market.
- MRVL has targeted designs based the China Mobile based TD-SCDMA 3G standard.
- In the past year, the company has changed direction and now offers a unified platform that works on TD-SCDMA and W-CDMA.
- Besides saving on development costs, this allows MRVL to compete more effectively throughout most of the world on the 3G standard. A new LTE offering was recently introduced and should generate numerous design wins in emerging markets.
- For those who haven’t followed the recent AAPL threads on VIC, I’d encourage you to read through the extensive comment sections. You’ll see numerous references to mid-range Chinese smartphone designs that match (or exceed) iPhone specs at a fraction of the price. It’s hard to quantify exact design wins, but this is the sweet spot for MRVL’s wireless business today.
- If there is one part of the MRVL story that is the biggest question mark for investors, it is the growth of wireless. Q2 revenue guidance is up 35% sequentially (off a seasonally weak quarter), and informal projections for the 2nd half are extremely strong. We expect wireless revenue of $673M for FY’14 (ends Jan ’14), a y-o-y decline of 18%. Our numbers aren’t too far off many of the sell side projections, which leaves plenty of room for an upside surprise.
- Networking is roughly 25% of total revenue, with gross margins likely in the 45% range.
- There’s quite a wide product mix within this segment including processors for routers, set-top boxes, gaming consoles, and carrier grade network equipment.
- Revenue is remarkably stable (growing 2% per year the last two years) considering the volatility of the carrier spending cycle. The nearest comp would probably be BRCM.
- We expect to see 4%-6% growth over the next two years based on management’s outlook and the pent up demand in carrier spending. Our projection for F’14 revenue is $749M.
Balance Sheet, Use of Cash, and Insider Ownership
- MRVL has $1.7B in cash and no long-term debt.
- Management has used the cash to make a serious dent in the share count over the last 2 years. MRVL has spent an average of $230M each of the last 5 quarters on buybacks. The share count has fallen from 606M to 505M over that period. This rate of buyback should continue over the next few quarters, which is not accounted for by most of the major sell side analysts in their EPS projections.
- A $0.06 per share quarterly dividend was initiated last year, for a current yield of 2.2%.
- Dr. Sehat Sutardja (CEO), Dr. Pantas Sutardja (CTO, brother of Sehat), and Weili Dai (VP and Sehat’s wife), together own 31% of the equity.
- MRVL hasn’t been very acquisitive, but there are several interesting possibilities. At the top of my list is ADNC, which would add nicely to the wireless offerings.
- Gross Margin target is 50%-55%.
- Opex guidance is flat y-o-y.
- Tax rate is minimal; MRVL is “based” in Bermuda.
- Our revenue projections: FY’14 (ends Jan. ’14) $3.2B, FY’15 $3.4B; which is basically in line with consensus. As mentioned earlier, there is a lot of upside if wireless plays out the way management is suggesting. We’ve been burned on this before so we’d rather wait and see.
- Sell side uses non-GAAP EPS, so we’ll stick with that for comparison purposes. Stock based comp has run at about $120M per year vs. $900M+ per year spent on the buyback.
- We see non-GAAP EPS at $1.03 and $1.34 the next two years… consensus is $0.84 and $0.93.
- FCF per share tracks slightly above non-GAAP EPS: $1.06 and $1.41 the next two years.
- A significant part of the EPS boost will come from the continuation of the share buyback, which isn’t included in the sell side estimates. The company has sufficient cash to continue buying back at a $900M per year rate. FCF should finance more than half of the buyback, with the rest coming from a gradual reduction in balance sheet cash.
- In December 2012, a Pittsburgh jury ruled against MRVL in a lawsuit brought by Carnegie Mellon University. The jury awarded CMU $1.17B in damages for the violation of 2 patents used in MRVL’s HDD controllers. Because the violation was deemed “willful”, the judge could triple the damages awarded to CMU. The judge has heard all the post trial motions and should issue a final verdict in the next couple months.
- From an investors prospective, there are 4 main issues to sort out; 1) Did MRVL violate the patents (maybe), 2) If they did violate the patents, is the damage award appropriate (no, it should be far lower), 3) Will the judge reduce the damages award (very possible), 4) Will the appeals court reduce the damages award (extremely likely).
- The 2 patents governing read channel technology were based on research done at CMU in the ‘90s. MRVL developed an internal solution for read channel technology that was similar to the CMU research. The capabilities that this technology enabled in MRVL’s controllers turned out to be useless to MRVL’s main customers. In fact, WDC engineers testified that they requested MRVL remove these features from the controllers because they were not effective at reducing the signal-to-noise ratio.
- We’re not engineering experts, so we’re punting on question of patent violation. Both sides have compelling evidence and the jury sided with CMU.
- The $1.17B damages award was based on 2 flawed assumptions: that all MRVL controller sales take place in the US, and that the patents allowed MRVL to generate gross margins in excess of the internal goal of 50%.
- The location of sales is a major issue because patent royalties would only apply to sales made in the U.S. CMU successfully argued that MRVL used the patented technology over a long sales cycle… that involves modeling and testing in the U.S. before securing a design win. MRVL argued that royalties should be based on the location of the final product sale. Using MRVL’s method to determine the location of sales would reduce the damages to $278M. But, even $278M is overly generous because it still uses CMU’s “fair” royalty rate of $0.50.
- CMU argued that MRVL was able to charge a $1.00 per chip premium based on their technology, which was worth $0.50 of gross profit. Hence the $0.50 royalty rate. The $1.00 per chip premium was based on one customer, Maxtor, which bought 0.37% of MRVL’s production. Maxtor was the only company that paid a premium for this feature. Never mind that WDC, who bought 47% of MRVL’s production, asked for a discount because of the disputed technology. The jury applied the $0.50 royalty to all 2.34B units sold to arrive at $1.17B in total damages.
- To summarize, not only did the damages apply to global sales, but they were also based on a ridiculous $0.50 per unit royalty rate. That royalty rate was concocted by looking at the price paid by a 0.37% customer… and ignoring the fact that WDC (47% of sales) asked for a price reduction based on the patented technology.
- The more common method of determining a patent award is the Georgia-Pacific Method. The court proposes a hypothetical negotiation between the counterparties at the time of the initial infringement. The negotiation should be arms length and based on comparable transactions when available.
- CMU licensed similar patents to IBM, STX, MMM, and INTC; all in the same time period as a hypothetical deal with MRVL would have been struck. These other patents were licensed for an average of $250,000 per year!
- Sometime over the next 2 months, the judge will announce her final decision on the damages award. We believe there is ample evidence for her to arrive at a far lower figure than $1.17B. However, she may uphold the jury’s decision since the case will inevitably go to appeals court no matter what she decides. The next step would be the Federal Circuit Appeals Court in Washington DC, where MRVL stands a far better chance. This court handles almost exclusively IP and patent cases and should be far more even handed than a hometown jury.
- Another potential outcome is a settlement between MRVL and CMU. We’d expect a settlement in the $200M range would be well received by the market. It seems like a $200M settlement would be attractive to CMU, given their low probability for success in the appeals court.
- Bottom line, the lawsuit will likely cost MRVL far less than $1.17B… and will likely take several years in the appeals process to resolve. The court may ask MRVL to post a bond to cover the damages award during the appeals process. This could affect the share buyback, but the details would be at the discretion of the court.
- EPS of $1.03 this year, and $3.36 of cash per share puts the stock at 7.2x.
- EPS should grow to $1.34 for FY’15 (ends Jan ’15).
- The multiple is depressed by the lawsuit, questions about the PC market, and worries over wireless design wins. All of these concerns are overblown and should abate over the next year. We don’t see why this company shouldn’t command at least a 10x multiple given its growth profile in wireless and SSDs; and steady earnings streams in networking and HDDs.
- A 10x multiple on next year’s earnings would value the business at $13.40. We expect about $2.75/share in cash at that time, which would value the shares at $16.15.
- Our estimates are fairly conservative… there plenty of potential for better wireless revenue. Consensus EPS estimates are lower than our numbers, leaving considerable room for upside surprises over the next year.
I do not hold a position of employment, directorship, or consultancy with the issuer.
I and/or others I advise hold a material investment in the issuer's securities.