Investors in new industries have a propensity to engage in folly often resulting in the manifestation of valuation bubbles for the pioneering companies. We intend to expose such folly in an effort to identify bubbles before they implode. Such was the case with our first two published research reports exposing micro-cap scams or "mini bubbles" which subsequently burst. Investors that heeded our warnings avoided subsequent losses of over 70%.
In financial markets participants are constantly processing incomplete information while being driven by the basic human emotions of greed and fear. Consequently, participants feeding a financial bubble tend to have a biased perception of reality making the bubble difficult to spot until after the implosion.
As the great Charles Mackay wrote in 1841: "Men, it has been well said, think in herds; it will be seen that they go mad in herds, while they only recover their senses slowly, and one by one." Nearly two hundred years later these words are still relevant as even nowadays men find it difficult to act against the herd while the herd goes mad.
Such is the case with shares of Mellanox Technologies (MLNX) which we believe are exhibiting the signs of a classic financial bubble no different than the bubbles that plagued the shares of Zynga (ZNGA) and Facebook (FB) this year and Netflix (NFLX) and Green Mountain Coffee (GMCR) in 2011. Those companies experienced sharp revaluations from the unsustainable multiples projected by the cartel of sell-side analysts when reality set in and investors realized that the businesses faced increased competition and a more modest rate of growth.
MLNX is the most expensive semiconductor company in the world trading at 55x Trailing FCF (Free Cash Flow) in an industry that trades at an average of 15.25x Trailing FCF.
Industry Avg. Excluding Mellanox: 15.25x Market Value to Trailing Free Cash Flow
When incorporating analyst estimates for 2013 cash flow into our realistic model we estimate that shares of Mellanox are worth no more than $46.17 or 61% lower providing significant downside risk from the current price of $118.00. In order to arrive at this figure we rely on very rosy estimates made by the company and sell-side analysts while having to wait over a year to see if the company can even deliver on those estimates.
Thus far MLNX has produced only $212 million in FCF since 2007 or an average of $47 million per year meaning that investors buying shares today at a valuation of $4.90 billion are in effect paying 105x the average annual cash flow the company has proven capable of earning. To place the richness of this valuation in perspective, imagine owning a business that required over a century to earn back the price you paid for ownership… Certainly a foolish proposition!
Mellanox Technologies Ltd. is an Israeli based semiconductor company that focuses on data transmission between servers and storage systems. The company's success thus far can be attributed to a prescient bet on a technology called Infiniband which it does not own (open standard) and which was overlooked by industry giants such as Intel (INTC) and AMD (AMD). Infiniband is a connection technology that is extremely fast (up to 56 Gbit/s) and provides a very low latency environment for large servers and storage devices. Infiniband adapters allow companies to increase their network performance given the incremental connection speed with additional benefits for servers running in virtualized and cloud based environments.
With Infiniband Mellanox played the classic role of a first mover getting a head start by recognizing the market opportunity and developing a suite of products and services that thrived with no competitors in sight. From the time Mellanox became a public company in 2007 (roughly the time the market for Infiniband started gaining momentum) the company has commanded a 70-90% market share of the Infiniband market. It is important to put this figure in perspective as the company's total revenues in 2011 were just $259.25 million indicating the total market for these products is still quite modest.
There is no doubt that Infiniband has been gaining prominence as a connection technology, but unfortunately for investors in Mellanox today this development should not be viewed favorably. As long as the market for Infiniband was niche and of modest size a company like Mellanox with no discernible moat could thrive due to its first mover advantage. In our view as long as the market for Infiniband continues to grow, Mellanox will for the first time face significant competition. Put differently, it is unrealistic and even foolish for investors to assume that going forward Mellanox will continue to command a 90% market share. As we will demonstrate in the next few sections, even the most optimistic sell-side analysts have warned that Mellanox has a 1-2 year lead at best on potential competitors. Specifically, our analysis confirmed an entrance by Intel which will be launching a far superior 100 Gigabit a second Infiniband product line (twice the speed of Mellanox products) as early as Q4 2013, cannibalizing market share from MLNX as soon as Q4 2012.
MLNX Valuation Defies Logic and Cannot be Mathematically Rationalized without Unsustainable Multiples
Investors have bid up shares of Mellanox to a valuation that reflects 55 years' worth of cash flow generation under the naive assumption that Mellanox can grow into its valuation by continuing to command a near monopoly in an industry where it has thus far faced zero competition. Mellanox's valuation is so out of whack that even if the business had commanded a durable competitive advantage, a multiple of 27.5 years' worth of cash flow (half the current multiple) would equate to a very rich $59 per share. Even at that price investors would be valuing MLNX at twice the valuation of bellwethers such as Intel or Texas Instruments (TXN).
In a way Mellanox is a victim of its own success, having made a lonely bet on an esoteric technology that is becoming the industry standard. The company's modest financial success can be easily attributed to that of a lonely entrant owning an industry that was too small for any competitors to take notice until now. It is absolutely crucial for investors to be cognizant of this point as the sell-side analyst community promotes the company with research that is predicated upon the flawed assumption that MLNX will continue to own the industry. Moreover the economic models employed by the analyst community are also flawed and can only be rationalized by projecting out income and cash flows for nearly half a century at worst and one third of a century at best. But the most critical flaw in the models comes from the lack of sensitivity for future market share cannibalization. All across the board, analysts appear to be taking the view that Mellanox will continue commanding a near 90% market share in 2013 and beyond. We believe this view is not only naïve but irresponsible given the guaranteed entrance by Intel as well as other competitors.
We have seen nearly identical trends before with other companies such as Netflix and Green Mountain Coffee. Both companies experienced rapid growth in an industry that was initially overlooked by competitors and were heavily touted by the sell-side analyst community which raised and re-raised estimates maintaining the same growth and market share variables in their models. Only when reality set in and the companies reported more realistic results did the analyst community incorporate those variables into their models. By that time it was too late as Netflix declined by over 80% from peak to trough in 2011 and Green Mountain by over 84%. We are absolutely convinced that Mellanox will follow the same path as the mathematical reality of its earnings power is processed over the coming quarters and sell-side analysts are forced to update their models.
Mellanox Meets Competition: Intel
Nearly 8 months ago on January 23rd shares of MLNX dropped nearly 11% in one trading session. The reason behind the drop was that QLogic (QLGC) (Mellanox's only competitor with nearly 15% of the Infiniband market) sold its Infiniband unit to Intel for $125 million.
With this acquisition Intel, the world's largest and most profitable chip company, decided that it would aggressively enter the Infiniband market. An article communicating the fear experienced by Mellanox investors the day of Intel's acquisition can be viewed here.
Brian Freed, an analyst at Wunderlich Securities summed up the market's reaction to the news as follows:
"These two companies dominate the market and Mellanox always had a significant lead on QLogic. Three years from now, this track record of dominating the market may change to a company that could potentially be chasing Intel."
Aviad Shrum, an Analyst for Psagot Investment House shared similar concerns:
"Intel is entering a niche market dominated by Mellanox and which is the core activity of the company. We see this as a significant threat to MLNX's future..."
The concerns shared by the analysts were valid as Mellanox would now have to compete with a company that was over 100x its size with a more solid customer base and stronger marketing and distribution channels. In a Bloomberg news article dated February 19, 2012 Radoslaw Walczyk, a spokesperson for Intel, confirmed the company's intentions:
"Our acquisition of QLogic's Infiniband business opens up new opportunities for competition and innovation in supercomputing"
Adding even more uncertainty to Mellanox's future, Intel announced that its first rollout of Infiniband products would be at 100 Gigabits per second, nearly double the speed of Mellanox's products and an entire generation ahead. Intel's decision was very wise as the industry life cycle for these products is roughly 5 years and the latest boom in Infiniband adoption took hold in 2008-2009 indicating the next big upgrading cycle will begin in 2013-2014.
Here lies another overlooked but very important point: Intel made a strategic decision to basically delay all efforts to gain market share with the existing Qlogic product line, instead focusing the newly acquired business on the development and rollout of the next generation products. This resulted in Qlogic/Intel products predictably losing market share in Q1, Q2 and most probably Q3-Q4 of 2012.
Wall Street Loves Bubbles
This graph shows the share price of MLNX since its IPO on 2/7/2007 at $17 a share. Note how this already expensive company began to form a bubble right around January 23rd the date of Intel's announcement. We view the parabolic uptrend as the "last hurrah" before reality sets in as early as Q3 2012 (10/26) and affirmatively by Q1 of 2013 as businesses begin to scale down their orders in anticipation of Intel's superior line of next generation products. Investors that pull up graphs of NFLX and GMCR will notice almost identical patterns with both companies experiencing a rapid subsequent decline as the bubble around the shares burst.
On January 23rd 2012, the day Intel announced its acquisition of Q-Logic, formally entering the Infiniband space and becoming Mellanox's first true competitor, MLNX shares traded at $31.35 for a market value of $1.23B. As we had mentioned in the previous section that day saw no less than 7 sell-side analysts distribute warnings and cautious notes with essentially the same theme: "The party is over."
Exactly 2 days later on January 25th, MLNX reported its fourth quarter numbers which beat estimates by roughly .02 cents a share amounting to net income for the quarter of .11 cents a share. The next day shares of MLNX rose by over $5 (over 10 years' worth of income per share) to $36.16. This activity can be easily explained as Wall Street participants simply did what they know best: They forgot all about the game-changing development and continued to buy shares of MLNX based on its predictable growth in earnings. As each subsequent quarter came and passed MLNX delivered the predictable results and the bubble began to form.
In our view the only reason shares of MLNX have traded from $31.15 the day of Intel's announcement to $118 today is because investors guided by sell-side analysts on Wall Street completely kicked the upcoming Intel threat down the road, choosing instead to focus on MLNX's short-term results which ironically are being fueled by Intel's strategy to enter the space (shelving efforts to market current generation products allowing MLNX to gain additional market share). But what troubles us most is that investors have now bid up shares of MLNX to levels that leave no margin of safety. At a $4.9B valuation investors purchasing shares of MLNX have to be certain that not only will the market for Infiniband grow at 15% CAGR (as the company predicts) but that MLNX will continue to maintain a 90% market share in 30-50 years.
Fortunately our thesis does not live in a vacuum. Recently other astute investors and business writers have written about the conundrum that is MLNX shares. In a piece for the UK Register, Timothy P. Morgan wrote an article that does a good job at outlining the guaranteed headwinds facing MLNX:
"These will likely be remembered as the salad days at Mellanox some years hence, when Intel has taken over networking much as it has done with processing. Perhaps during the 100 Gigabit InfiniBand and Ethernet rollouts maybe in 2014 or 2015"
The bottom line is that MLNX thrived in a niche industry with no competitors because the industry was in its infancy, being too small to attract attention. As the industry matured and became a real opportunity the world's largest chip maker marked its entrance. Those are simply the facts of life. MLNX does not have a moat (aside from a 2 year head start on the 56 gb/s products) and has absolutely no chance at competing with the likes of an Intel as it prepares to launch superior technology at the end of 2013.
Now it is imperative to remind readers that all these details and speculations would be irrelevant if MLNX shares were not trading at a bubble valuation. Had market participants not made MLNX the flavor du-jour cheer-led by the sell-side community, MLNX may have made for an attractive acquisition candidate for another entrant to the field. However at the current valuation MLNX investors are caught between a rock and a hard place as the reality of competition sets in, at which point the opportunity for a sale may have already lapsed.
In this graphic from the Infiniband Trade Association which plots the growth of Infiniband, as well as the projected evolution of the technology, we superimposed dates of significance. This graphic does a superb job at visualizing why MLNX shares are primed for a significant decline. As investors can clearly see: Intel simply decided to skip one generation forward before launching its products. That time is near, at which point MLNX will lose any competitive advantage and be valued at multiples closer to its peer group. That is under the assumption that MLNX will still be able to compete with the likes of Intel.
Insider Sales Confirm Bubble
As investors drove MLNX shares up by nearly 300% in 8 months, insiders have opined by unloading an unprecedented amount of stock. Unfortunately for long investors there are an additional 3.8 million options outstanding at an average exercise price of $17.09, guaranteeing even more dilution and potentially adding an additional 4-500 million dollars to the already bloated market value of $4.9B.
Since the beginning of the year MLNX insiders have unloaded over $31 million worth of shares. From this figure over $17 million worth of shares were sold in the last 30 days hinting that even insiders believe that MLNX shares are overvalued.
This table presents the disclosed insider sales for 2012. It's interesting to note that with the exception of CEO Eyal Waldman (who is the largest individual shareholder of the company, owning nearly $200 million worth of stock) every single insider has been unloading with increased tenor over the last 30 days.
We were surprised to see Michael Kagan CTO and Chief of Architecture liquidate nearly $2.5 million worth of Mellanox shares in the last 30 days. Surely as the CTO of the company he would be in a favorable position to gauge whether shares of Mellanox are cheap or expensive. Even more insightful would be Marc Sultzbaugh, VP of Worldwide Sales, who has sold over $2.5 million worth of shares in 2012, with over 73% of that amount being sold in the last 30 days!
The insider sales merely confirm what is obvious to any objective investor making reasonable estimates of future cash flow for MLNX: The valuation is a bubble! It is highly unusual to see insider sales amounting to 70-100% of total yearly sales being executed in a few short days. Clearly, they have decided to unload their shares into the frothy valuation.
On a related topic it is interesting to note that CEO Eyal Waldman's Facebook Page is open, allowing anyone to view the fast paced life of the tech millionaire. We highly recommend that readers of this report take a peek, as you will surely enjoy the plethora of fast cars, yachts, and bikini-clad women. At the very least investors can appreciate that the proceeds from the share sales are being spent wisely!
Mellanox is Overvalued Even According To Mellanox
Slide From Mellanox Presentation in August of 2012 Showing Total Available Market @ $5.3B
|(click to enlarge)|
Slide from Mellanox investor presentation on June 3rd, 2011 showing Total Available Market @ $4.7B
Since at least June 2011 when its market cap was only $980 million Mellanox has been touting its Total Available Market (TAM) as an opportunity worth $4.7B. Since then investors have bid up shares of Mellanox to a bubbly valuation of $4.9B. In the updated investor presentation from August Mellanox now pegs the market opportunity at as high as $5.3B. But even this figure highlights a major red flag for investors betting on the future of the company. According to the company's own assessment of the potential size of the market, Mellanox is extremely overvalued as its fully diluted market cap is now larger than the market opportunity itself.
We could understand why a very optimistic investor would make a bull case for a company tapping a $4.7B opportunity while its market cap was less than $1B but we are puzzled as to why any logical investor would pay more for a company today than their current market opportunity. We also believe that MLNX is not accurately presenting the TAM as it does not subtract its cumulative revenues from the TAM as we have done in our models.
Our Fair Value Model for Mellanox
As part of our research we decided to build a fair value model for shares of MLNX. In an effort to be as objective as possible we used Bloomberg Analyst Estimates for future revenue and FCF figures incorporating some very rosy estimates. Additionally we used the company's own data for growth rates for the market opportunity as we wanted to monitor the true size of the TAM.
Our model simply takes the FCF estimates for 2012 and 2013 projected by Wall Street but applies a more realistic 15.25x FCF multiple which is the peer average. Additionally we apply sensitivities to the projected FCF number based on potential market share penetration. Contrary to the sell-side firms that love drinking the Kool-Aid we don't just assume that MLNX will command a 90%+ market share for the remainder of the year or 2013. Finally we tallied the cumulative revenues for MLNX since inception, as well as the cumulative cash flows, in an effort to decipher the actual TAM. By subtracting the cumulative revenue from the TAM the model projects the remaining TAM. Our model then takes this figure and uses Mellanox's own growth rate projections for the TAM to estimate the TAM in 2013 and then once again subtract the cumulative revenues in 2013 (incorporating analyst estimates for revenues in 2013) providing a remaining TAM at year end 2013.
The first data point that is uncovered by our model is something MLNX fails to mention in its own presentations: At the end of 2012 MLNX will have tapped nearly 25% of the TAM given their cumulative revenues hit $959mm as of June 30th, 2012. By the end of 2013 MLNX will have tapped over 30% of the TAM, begging the question why any responsible analyst would project out 40-60 years' worth of cash flows.
Additional data points from our model serve as the nail in the coffin of our thesis:
- MLNX fair value will be $36.59-$47.04 a share by the end of 2012 (assuming it meets analyst estimates) based on a peer group avg. multiple of 15.25 years' worth of cash flow (INTC only trades @ 13.4).
- MLNX fair value will be $44.22-$56.85 a share by the end of 2013 (assuming they meet their numbers over a year and a half from now). Keep in mind that any option exercises will reduce this figure as we did not make an attempt to project a fully diluted share count.
- At the end of 2013 the remaining market opportunity is projected to be no more than $4.77 billion, which is about $200 million less than MLNX's current market cap!
- Applying the historical FCF to Revenue Margin of 22.7% we arrive at a potential cumulative FCF for MLNX of just $1.08 billion from the TAM which will have to be earned over nearly 8 years and assumes that MLNX will retain a 90%+ market share. Put differently, investors purchasing shares of MLNX today will have to wait nearly a decade for the company to earn only 20% of its value under the most rosy scenario.
- Using the blended average for the 2012 and 2013 numbers we arrive at a fair value estimate of $46.17 for MLNX over the 6-12 months, projecting a nearly 61% decline in the share price.
It Ain't What You Don't Know That Gets You In Trouble. It's What You Know For Sure That Just Ain't So. - Mark Twain
In conclusion, we urge long investors to stay away from MLNX shares as there are significant downside risks that the bubble around the stock begins to burst. Speculative investors should consider a short position capitalizing on the asymmetric risk/reward proposition at these unsustainable levels.
Summarizing our view on the name we highlight the following red flags and catalysts for a sharp revaluation to the downside
- MLNX growth rates and profit margins are unsustainable and have only been possible due to a temporary monopoly whereby the company commands 90% market share. Given that the current share price reflects the rosiest growth and profit projections as much as half a century into the future we believe that MLNX shares will experience a sharp revaluation to the downside when the company's numbers begin to confirm our thesis. As with NFLX and GMCR by the time they do it will have been too late for investors.
- Intel is entering the market in less than 12 months with a far superior product a full generation ahead of MLNX's, ending the days of operating in an industry with no competitors and guaranteeing market cannibalization in the bull case and possibly a full threat to MLNX's existence in the bear case.
- Sales of Infiniband products account for 90% of the company's revenues while 49% of revenues are derived from just two customers. As the lifecycle of Infiniband products is 5 years, businesses will begin to weigh their future investments in upgrading to Infiniband over the next 12 months (the end of the boom in first generation Infiniband products) most likely choosing to wait for the superior next generation of products from Intel.
- Unusual and aggressive insider selling over the last 30 days by key employees confirms our thesis as the people who have the best insight into the future of the company vote with their pocket books. Additional dilution is guaranteed with almost 4 million options at an average strike price of $17 to be exercised and sold into the market.
- MLNX shares have become an unsustainable bubble trading at a valuation of $4.9B now exceeding the entire market opportunity as stated by the company.
We leave you with a chart of GMCR and NFLX superimposed. This chart perfectly visualizes the bubbles forming and then bursting as reality set in and shares of both companies declined by over 80% in less than 5 months. We believe that shares of MLNX are destined to follow the exact same path with the bursting of the bubble occurring imminently.
Disclosure: I am short MLNX.