| | | Mellanox Is Taking Some Hits, But The Pessimism Seems Overdone
Must Read
|
Sep. 14, 2016 1:07 PM ET
|
| About: Mellanox Technologies, Ltd. (MLNX), Includes: AVGO, INTC
by: Stephen Simpson, CFA
Summary
Intel is having an impact on the high-speed interconnect market, as it is pricing aggressively to gain share at Mellanox's expense.
Underlying growth for high-speed interconnect remains healthy and Mellanox believes it has a price/performance mix that beats Intel in its target markets.
While execution/competitive risk is high, low double-digit FCF growth supports a low-to-mid $50's fair value and Mellanox is worth another look from aggressive investors waiting for such a sell-off.
When I last wrote about Mellanox (NASDAQ: MLNX) in March, I ended with the following:
"The good news is that these shares seem to routinely post 25% pullbacks that give patient investors a chance to reload. At this point, I'm more inclined to wait for one of those freakouts than chase the shares today."
Since then, the shares have pulled back a little more than 25%, with a big drop in late April/early May around earnings, a rally, another decline in July/August around earnings, and an attempted rally from mid-August to mid-September that hasn't held.
Nothing much has really changed in how I view (or value Mellanox). Intel (NASDAQ: INTC) is making plenty of noise with Omni-Path, some potential Mellanox customers are adopting it, and that is having an impact on Mellanox. But that was always expected (at least by me) - the bigger question is whether Intel is really changing the game in a more meaningful way, and I don't think that's the case. I believe Mellanox will continue to be volatile, but I still believe in the basic story and that the mid-$50's is a reasonable fair value.
Beat-And-Lower Doesn't Help
Growth stock investors want beat-and-raise quarters, but Mellanox has guided down for the next quarter after its last two quarterly reports. While the company's "beat" in the second quarter did come close to the original pre-lowering expectation of the Street, and the guidance revisions haven't been large, it has nevertheless spooked investors already skittish about the pace of transition to higher-throughput interconnect and the potential impact of Intel on the market and Mellanox's position within it.
That Mellanox continues to post double-digit year-over-year organic growth is more or less relegated to, "Gee, that's nice. Now, about that guidance..." and that's unfortunate. Revenue was up about 25% in the first quarter and 16% in the second quarter excluding EZchip, but investors were bothered by the 2% sequential decline in InfiniBand revenue (never mind the 28% reported sequential growth and 11% core growth in Ethernet) and the small downward revision to revenue guidance.
Now the question is whether things perk up in the second half of the year. The comps for InfiniBand are getting easier, but I would describe the enterprise hardware spending environment as "skittish." Companies with very demanding performance needs (deep learning, Web 2.0, et al) are still spending, and they represent an outsized portion of Mellanox's addressable market, but storage isn't looking so great and data center/cloud likewise seems a little more restrained.
Intel Making (And Leaving) A Mark
Intel has not been shy about the claims it has made regarding Omni-Path - in how it performs relative to Mellanox's technology, the price/performance trade-offs, and how will it impact the market. Intel has made some aggressive claims about early adoption, including 30% share of 100G InfiniBand. The June release of the Top 500 supercomputer list shows that 41 of the 53 new non-Ethernet systems are using InfiniBand (and eight using Omni-Path), which suggests Mellanox gaining share overall (relative to all HPC sites using non-Ethernet intereconnects). I'd also note that 40 of those 41 new systems using InfiniBand use Intel CPUs, so you would think Omni-Path was at least an option.
Mellanox is acknowledging that Intel is having an impact on the market. In the second quarter conference call, management said they lost those eight deals that Intel won because "Intel paid their way into the market." I'm sure Intel would disagree with that characterization, but it does seem inevitable that Intel is having, and will continue to have, an impact on price in the high-end interconnect market.
How far this goes depends on where the truth lies between the various performance and price/performance claims made by Mellanox and Intel regarding their technologies. Mellanox skeptics claim that the company's offload architecture is only important in a "limited" number of applications, but they seem to neglect to note that those users will pay up to have their needs met.
On the other side, Mellanox bulls/Intel skeptics note that integrating Omni-Path onto the CPU (expected next year) could still prove harder than expected, with further delays and/or underwhelming improvements in cost and latency at least possible.
All told, Intel is going to get some business that was otherwise going to go to Mellanox. The real question, at least in my view, is whether Intel is going to get more than is already factored into expectations - Intel's Omni-Path has been a looming threat for some time now, so it's not "new news." Thus far, it doesn't seem to me as though Intel is doing notably better than expected, so unless Intel can drive a real sea change with an integrated Omni-Path offering, I think Mellanox is still on pace for high single-digit long-term revenue growth and double-digit free cash flow growth.
How Is The Market Moving?
Here's another point to consider about the Mellanox versus Intel battle. Right now, only a small portion of high-performance computing customers are buying 100G interconnects, and numerous surveys would suggest that 56G can/does meet the needs of projects on the books. With Omni-Path at 100G, that could make for a pretty attractive opportunity for Mellanox in that 25G to 100G space.
I'd also note that a lot of companies seem to be moving from 10G to 25G and holding off (for now) on 40G. The 10G to 25G move is less disruptive and cabling can be reused between 10G and 25G (but not at 40G). In fact, even with Mellanox's growth forecasts, the company still expects 10G to be the largest portion of the high-speed adapter market in 2020 - from 81% today to about 43%, with 57% of the market at 10G or 25G.
Last and not least, I also want to mention that for all of the concern about Intel's impact on the InfiniBand business, Mellanox does still have something to worry about with Broadcom (NASDAQ: AVGO) in the Ethernet business. The company's Ethernet business has been growing nicely of late, and Mellanox's Spectrum Ethernet switches cover the high-speed waterfront (from 10G to 100G), but I wouldn't just dismiss Broadcom's Trident and Tomahawk platforms.
The Opportunity
The biggest risk I see to Mellanox is that Street expectations don't fully reflect the potential price/margin impact of Intel's aggressive price-based push with Omni-Path. My revenue numbers for both 2016 and 2017 are below the lowest published estimates, but that doesn't mean there couldn't be additional downside.
I am still looking for long-term revenue growth in the very high-single digits (or double digits if you round up), and that is not conservative in my view, even if the high-speed interconnect market is due to grow on further spending on deep learning, Web 2.0, cloud, and advanced storage.
I have shaved a little bit out of my margin expectations and that reduces my resulting long-term FCF growth estimate from a little above 15% (10-year CAGR) to a little under 13%. With that, my fair value drops less than $3, to a little under $54.
Risk factors remain numerous and varied. Adoption of high-speed interconnect may underwhelm, whether from lower speeds being "good enough" or fewer than expected high-end deployments. Rivals like Intel and Broadcom may take/keep more share than expected, or entirely new rivals/alternatives may emerge to disrupt the market. And, of course, there's regular old operating risk to consider - many tech stories find it difficult to deliver on growth and margin expectations and maintain their rich multiples.
On the other hand, Intel has a record of over-promising and under-delivering, Broadcom may no longer be willing to reinvest enough to maintain R&D leadership, and a company like Qualcomm (NASDAQ: QCOM) could see Mellanox as a valuable piece toward building its own end-to-end solution.
The Bottom Line
All of my valuation methodologies (DCF, ROE, OPM-EV/Rev) come back with estimated fair values in the low-to-mid $50s, and that's basically where I was back in March. While the recent trend of lowering guidance has to stop for the stock to work, I think the basic fundamentals of the business are still intact, as is the long-term growth opportunity.
Holding these shares into earnings has been risky of late, but it looks like expectations are lower this time around. I'm certainly interested in hearing management's updated commentary around Intel's pricing behavior and the sort of market share wins they're seeing, but I think the risk/reward equation is a lot more interesting now and I think aggressive investors should take a closer look.
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. |
|
