Barclays
Mellanox Technologies
Analyst day reinforces investment thesis – MLNX still leads high speed interconnects
Stock Rating/Industry View: Overweight/Neutral
Price Target: USD 60.00
Price (27-Oct-2016): USD 42.50
Potential Upside/Downside: 41%
Tickers: MLNX
Highlights from MLNX’s 2016 Analyst Day
Just a day after reporting 3Q earnings, we attended MLNX’s analyst day in New York. We believe the company laid out a strong case for why it can continue to grow in the 15%-20% range over the next couple of years and hit our $4.00+ EPS estimates. 24% growth in 100G Infiniband revenues in 3Q16 and share commentary from the analyst day should ease investor concerns over the threat from Omni-Path. Ethernet was weak last quarter but as hyperscale customers deploy 25/50/100G, MLNX is well positioned to grow. We estimate that MLNX is on pace to grow revenues 15% organically (the 2016 results include EZCH which is not disclosed separately) and that the level of growth can continue in 2017 – we currently estimate 15% growth.
Almost all growth in the high speed adapter market is in 10GE+, where MLNX the majority of share currently and Broadcom is really the only other player in the market. We view the Spectrum switch offering as a promising development in diversifying the revenue base but we don’t expect revenues until 2017 and still don’t model a significant contribution initially. Success in this market which will come from both direct sales and through white box vendors would mean 10s of millions of revenue in 2017 as a starting point.
Market share data presented on Friday strengthened our view that Omni-Path is not an Infiniband killer. Intel will get its fair share and needs to be monitored, but MLNX will do just fine in our view. Inifiniband maintains 90% share versus Omni-Path and has 41.2% of the overall TOP500 supercomputer list. This should be confirmed when the new Top500 list is released in mid-November. Within that dataset, MLNX has 70.4% of the TOP500 HPC platforms, which is where Omni-Path is focused. Though Omni-Path pricing seems to be aggressive, Infiniband’s 20%-50% higher application performance drives a lower total cost of ownership and a 10-30% lower datacenter cost/performance.
The engineering point in the MLNX vs. Intel battle positions MLNX’s offload capabilities versus Omni-Path’s integrated on-CPU approach. MLNX doesn’t have to wait for CPU or GPU and can analyze data as it moves which allows a customer to do things in real time. MLNX argues that even if Omni-Path costs less, customers have to pay more because they will need more CPUs as a result of the onload/offload debate. Mellanox documented public user testimonials report performance and scalability issues with Omni-Path with users unable to use all CPU cores due to the Omni-Path architecture. We continue to believe 100G Infiniband will be able to grow in the face of Omni-Path in 2017 and we forecast revenues up 36% (albeit off a lower base than other segments), part of this growth will be fueled by the intro of 200G HDR Infiniband next year.
The day highlighted a testimonial from Albert Greenberg, CVP for Microsoft Azure. MSFT has been a key customer for MLNX’s 40G Ethernet product. Mr. Greenberg laid out the value proposition for Azure and several of his comments on the architecture needs of MSFT were directly in-line with the MLNX strategy. He stressed the need for open architectures, the need for low latency and high performance across the “spine” of the data center – meaning all the layers of the data center for the in-data center “East-West” traffic needs to be at high speed. He highlighted RDMA performance and that the MSFT “likes” the move to 25/50/100. In all we viewed his presentation as a solid endorsement of MLNX approach and position in the highest end of the datacenter, which is where we believe MLNX offers the most value.
CFO Jacob Shulman laid out a broad financial overview detailing opportunities for revenue growth (transition to 25/50/100G, Ethernet switching and LinkX, Infiniband growth, BlueField/NPU) but also a focus on driving EPS expansion and cash flow generation. Opex as a percentage of revenues has typically run around 50% but the company has implemented cost reduction programs to drive this down and we do believe there is significant leverage in the business model if opex becomes a focus. We grow opex at a mid-teens rate in 2017/18 and keep it at 49.6% of revenues. Bringing opex down to just 49% of total revenues and keeping all else equal adds roughly $0.10 to EPS in 2017 and 2018. Gross margin will come down in the future due to mix but all of the new product introductions should help keep them elevated at the high end, or even above, its long-term model range of 65%-69%. Mr. Schulman also emphasized the strong cash flow of the company. Debt associated with the EZCH transaction has already been reduced by $27 million and the company is cash positive. We expect the company to use its cash to look at future deals (it has made 7 acquisitions) and CEO Waldman expressed his view that with the strong growth opportunity ahead he did not believe that a share buyback, even to cover the dilution of stock based compensation, is likely in the near term.
We reiterate our OW rating and see the current stock price as good place for investors to build positions in the stock. |
