1)      Highly complex technical software that is difficult to understand unless you are a semiconductor expert or spend substantial time doing industry diligence.  MENT has never been posted on VIC and none of the big industry players have ever been discussed on VIC except for SNPS in 2001.

2)     Generally poor analyst coverage / lack of good comps in the parts of their business where most the value is, and weaker businesses in areas where larger public comps play (SNPS & CDNS). 

3)     Poor appreciation for the differences in their revenue recognition model vs peers and how timing of software renewals and M&A activity can adversely impact results in a quarter without having a material impact on the long-term business.

4)      Bad Q4 and first half guidance that sent the stock down from 27 to its price today, reasons for which have been very poorly communicated / have little to do with the strength of the business.

5)     Temporary weakness in the hardware part of their business which garners much analyst attention but does not move the needle meaningfully on valuation.

6)     Management are good operators, but very poor at investor communications.  They also have arguably managed the business to make quarters rather than maximizing long-term profitability.

Given the valuation, conservative balance sheet, and strong fcf, we view downside as limited, and envision scenarios where the stock could appreciate 40-60% as temporary issues abate, with higher values possible in a strategic transaction or if the business was managed more for profitability than for growth.  We think the situation is ripe for a PE takeout or constructive activism (for a variety of reasons we think overly hostile activism would not work well here).  This is a beast of a company from a research standpoint, we look forward to answering those questions we are able to and learning from others in the VIC community who have looked at this or decide to look at it because of this write-up.

Business Overview - why this is a good business:

Mentor provides EDA software, which are a series of tools and software that allow semiconductor and electrical engineers to design, test, and verify semiconductor and various electrical designs.  In short, if you are an engineer who designs semiconductor chips, printed circuit boards, wiring systems, or other electronics, chances are you use Mentor software or that provided by one of their two primary competitors, Cadence & Synopsys.   In many respects, the best comps to Eda software are applications that professionals in other industries rely on every day to do their jobs: Adobe (creative professionals), Technical software used by engineers (Autodesk, Dassault, PTC, Ansys), or software applications many of us are probably familiar with in the finance community that we rely on every day to do our jobs (Bloomberg / Factset / CapIQ).  This software categorically, and Mentor’s software specifically, is core to the everyday function of the people who use it.  There are huge learning curves to getting familiar with a certain software and substantial switching costs due to engineers familiarity with certain software and the disruption that switching can bring in terms of productivity. Further, in many cases, there are reams of data and designs stored in the software, and switching to different software would require you to manually move the data, running the risk of having errors when you transfer (competing software does not play nice together here, generally on purpose). For all these reasons, switching is rare, and when it does occur on tends to coincide with a new design, often times resulting in several years of transition before you can completely phase out usage of a product. 

Eda Software is generally sold to three different end markets - fabless semi companies, semiconductor foundries, and "systems companies" that make electronics within their products (eg automotive and mil/aero). Mentor's revs are split about 50/50 from semiconductor and systems companies, vs cadence and Synopsys who get the vast majority of their revs from semi companies. We think distinction is lost on the market, and it is important: systems business is generally less competitive and less price sensitive, and it is growing faster rate as the number and complexity of electronics increases in most products on the market today. It's also worth noting that neither market is particularly cyclical, with spend generally tied to semiconductor r and d, which is much more resilient and less cyclical than semiconductor revenues or profit margins. 

Unlike some of the aforementioned software markets, the Eda software space is made up of about 10 broad markets, and in many cases 5-10 sub-segments within each broader market.  In general, the big 3 control almost 100 percent of the market in each niche, and the leader typically has market share in the 60s:

[Trouble pasting image, see p.9 http://images.mentor.com/investor-relations/FINAL_MENT_ISS_Presentation.pdf)

The above actually understates the market share each player truly has in the broader markets, as in many cases within the broader categories there are sub-segments where one player may dominate and another player may dominate, so when you aggregate them together the share looks more even than it actually is.

[Trouble pasting image - see p. 11 http://images.mentor.com/investor-relations/FINAL_MENT_ISS_Presentation.pdf)

Mentor focuses efforts on products where they are the leader or a close number 2. As such, the vast majority of their revenue comes from areas where they lead in market share or are a close second.  I think this is probably overstated by a few percent, but regardless the point is the same:

Overview of Mentor products - Dominant products:

Rather than focus on explaining everything mentor does, we think it makes the most sense to focus on the franchises where they have leading positions and very high margins. This classification does not match up particularly well with how mentor breaks out it’s revenues, but you can match these up to the right category reading through the 10K.  It is worth noting that although ebitda margins for the business run in the low 20s, we believe there are certain areas mentor dominates where their margins are substantially higher, and other areas where mentor is not nearly as dominant where they are breaking even or losing money.  This is one of the many opportunities we see for constructive activism. We include some estimates of revenue and profitability below, which are primarily based on conversations with industry experts, former employees, and industry data from Gary smith Eda, the leading independent firm that tracks the Eda space. We have no doubt these numbers are wrong, but they should be directionally close, especially on revenues.

Revs and ebitda in dominant products: (56% of revs, 90% of ebitda)

Physical verification / Calibre suite of products (34% of revs / 56% of EBITDA):

This area is the crown jewel of mentor, and where we think most the value is. Calibre is a suite of software that is used in "physical verification" of ic designs. In layman's terms, there are three basic parts of the semiconductor design process: ic design (actual design of the chip), functional verification (testing the logic and making sure the chip does what it's supposed to do), and physical verification (making sure the design can actually be manufactured and done so efficiently). Mentor's Calibre suite addresses this last challenge, and their market share in these products generally ranges from 50-80 percent. This product line can be further broken out into a few key segments: Design Rule Checks (or DRC, core of the Calibre franchise), RET, silicon test and yield (Tessent product), and design for manufacturing solutions. 

Nearly everyone we have spoken to in industry refers to Calibre as a “franchise” and “industry standard”.  In the interest of time, we will focus primarily on their DRC software (the bulk of revenue and profitability here), but can answer questions on other areas in follow up.  Once engineers have completed the design for an IC, they will hand off that IC to a foundry, who actually produces the IC.  The foundry has a list of thousands upon thousands of design rules, which are manufacturing rules that must be abided by in order to make sure the product can actually be manufactured and work as it is supposed to.  People at the foundries are semiconductor experts, not software experts.  In the mid 90s, with years of work and investment, Mentor partnered with TMSC to translate their design rules into a programming language that could interpret semiconductor designs and check them against the design rules of the foundries.  Foundries are not software experts, so they relied on Mentor to help them basically code these rules and, in many cases, find workarounds to various rules that would create issues.  As process nodes advance new design rules are added, and this process once again happens between Mentor and the major foundries.  In essence, for the last 20yrs, the semiconductor industry has basically outsourced this problem to Mentor, who codes these rules in their software and in turns sells the software both to the foundries and, more importantly, to semiconductors companies who must use this software if they want to make sure their design can be produced efficiently.  Both Cadence and Synopsys have, over the years, tried to make inroads, and here, and have largely failed because of the stickiness of Mentor’s DRC software.  There are a couple issues here that make it difficult for a competitor to break in

1)      Mentor owns the format / coding language that they have developed for translating DRC rules.  A competitor would basically need to recreate all these rules with the foundries, which would take a tremendous amount of time and duplicative effort on the foundry’s part.  Mentor has people on the ground at foundries whom they have worked with for years, if it ain’t broke, don’t fix it…

2)      There is a very high risk of messing something up, and if this happens it can be very expensive.  The cost of producing a single bad design with substantial yield / performance issues can run in the hundreds of millions of dollars (basically more than the entire market for DRC software) so there is little incentive for the foundry to engage with another provider as long as Calibre works well (and it does).  Further they would need to translate the existing design rules (which work and have been iterated over time with mentor) into a new language with new people.  Again, for this reason it just doesn’t make sense for the foundry’s to help another player be competitive.

3)      There are natural network effects that are at play here.  Because Mentor has DRCs for all the major foundries, anyone who wants to manufacture at the foundries needs Calibre for DRC.  In most cases, especially at the more advanced process nodes, you don’t have a choice.

Printed Circuit Board (PCB) design solutions (17% of revs, 26% of EBITDA):

PCB software is used by PCB engineers to design PCBs for use in consumer electronics and industrial applications. ICs are pieced together onto a PCB that in turn allows multiple ics to be connected together to form more complex tasks or logic. There are about 12 different subsegments of pcb design, and mentor has the most comprehensive and integrated solution. This space has a few more competitors, but Ment is the leader.  MENT estimates their share at 42% in one of the prior graphs.

Altium is publicly traded in Australia and trades at 18x EBITDA FY16 EBITDA and over 5x revs. Mentor predominantly competes with Cadence on the high end and Altium on the low end. Although Altium has been successful in capturing share at the low end of the market, their software "breaks" as you get into more complicated pcb designs, which generally can only be handled by cadence and mentor. Like many of their software products, switching pcb software is a huge nightmare. Not only is there the requisite learning curve issues, but there are substantial issues, risk, and time associated with porting your design library from one piece of software to another. For this reason switching is very uncommon, unless you have to. Altium’s EBITDA margins in PCB are in the low 30s, and generally mirrors what we have heard as this being a very profitable business.

Wire harness software (5% of revs, 8% of EBITDA):

Mentor's Capital product is used by engineers in automotive, aerospace, and defense to place wires in vehicles and airplanes. With the increasing number of electronics in these areas, there are now literally miles and miles of wire in a car and multiples of that in an airplane. Figuring out how to route and design the wire harness system is exceptionally complex, and requires specific software like that provided by mentor. Mentor is number 1 in automotive by a good am mourn and a close second in aerospace to IaI.  Overall according to Gary smith Eda they have about 50 percent of the market. The consequences or messing up wire design are dramatic. Airbus lost billions of dollars and encounter a couple years or delays on the a380 due to issues they encountered here that could have been avoided with the right software. See: http://calleam.com/WTPF/?p=4700

Revs and ebitda from products where Ment has greater than 20% share but is not number 1 (~30% of revs, EBITDA ??):

Functional verification software (14% of revs)

This is software that is used to verify that IC designs perform the functions they are intended to perform.  To use the example of an Iphone, you will want to test each individual IC to make sure it does what it is supposed to (e.g. that the camera works, that the antenna works, etc.).  Mentor is the #3 player in this market behind Cadence & Synopsys.  This is probably the most competitive of the EDA markets, with Synopsys & Cadence each having about 37% share and mentor having low 20s share.  Mentor is a decent number 3 here but definitely a laggard for a variety of reasons.  It’s unclear to us the margin profile of the business here considering mentor has less revenue to spread their R&D across and also since they tend to compete more on price here.

Veloce (~10% of revs):

Veloce gets a disproportionate amount of investor and analyst attention, largely because it is the fastest growing part of eda (teens growth) and also has the most fight for share (about 45% Cadence, 40% mentor, 15% synopsys).  Veloce is a hardware based emulation tool that allows for functional verification of an entire product design.  Functional verification software allows you to make sure individual components work as they are supposed to, but the amount of simulation horsepower required to verify a whole system (e.g. a whole iphone) works together as intended is beyond the realm of what software can handle.  These emulation boxes cost a few hundred thousand to upwards of a million dollars, and are used largely by systems companies to make sure products are likely to work as intended.  A famous functional verification failure was the antenna on the Iphone 4, where the antenna worked on it’s own but in the context of the entire design, had issues on certain bands.  That recall cost hundreds of millions of dollars and could have been caught with an emulation box.  Because software revenue is reasonably predictable, sales of emulation hardware are one of the more likely factors that enable an EDA company to beat / miss a quarter.  That said, we’d point out that we believe gross margins here are in the 50-60% range, and for mentor operating margins are probably single digits if that, so even though it’s a decent chunk of revenue, we don’t think this business has a lot of standalone value justified by profitability.  That said, this is considered by many to be a very attractive space and mutliples in acquisition have been high.  In 2012 Synopsys purchased Eve, a smaller player in the space, for about 3x revenues despite minimal profitability, which many in the industry argued was cheap (note that our est. includes substantial litigation liabilities which increases the multiple above the reported 2.5x):  http://www.deepchip.com/items/0517-05.html

Computational fluid dynamics software (30M+ of revs, should have ebitda margins of 20%+):

This refers to a suite of “Flo” products products produced by Mentor, of which there are several.  These products include Flotherm (software for measuring heat produced from electrical designs to make sure there aren’t overheating issues), as well as a products Flowmaster & Flowvent which model airflow and fluid movement in industrial design applications.  Flowtherm tends to integrate with mentor’s PCB software, whereas Flowmaster & Flowvent integrate more with MCAD software offerings like those by PTC or Dassault.  This market is small, but growing, and competition is relatively limited.  Major players in this space include Ansys, which has 40%+ EBITDA margins and trades at over 7x revs, and Cd-Adapco, which was recently bought by Siemens for 5x revs.

Revs and ebitda from subscale product areas: (~10% of revs, likely losing money)

Olympus and rtl software: This area is traditionally dominated by cadence & synopsys and involves software used in the actual design of IC chips.  Mentor has some decent offerings that target the low end, but they generally are subscale and likely losing money here.  They tend to be low cost and are a thorn in the side of cadence & Synopsys, who would benefit greatly if this business was shut down as it would likely alleviate some modest pricing pressure they can face in this area.

Embedded software:  We have not focused too much here, we believe this is about 50M of revs, generally pretty competitive, and modestly unprofitable.  The closest comp here was bought for 2.5x revs by Intel in 2009.

Q4 guide down and why the stock move was an over-reaction:

One of our favorite software investments to make are situations where people misunderstand variability in license revenues that can optically create the impression a business is substantially challenged, when in fact there are reasonable explanations that have little to do with the business being impaired.  Because software revenue is at such high gross margins, shortfalls in revenue can have outsized impacts on margins.

Prior to report Q3 results, Mentor had made over “6 years” worth of quarters in a row.  The stock was trading roughly in-line with EDA peers, and the business had posted several years of mid single digits revenue growth and expanding margins.  When the company reported Q3, they hit their guidance but guided Q4 revenues substantially below those implied by their full year guidance.  Q4 revenues are guided to be down 24% year over year, vs. prior expectations of roughly flat year over year.  The stock sold off from mid 27s to the 17s.  Management’s explanation for the reduction was confusing, citing 3 main factors – M&A within customers, foreign exchange pressure, and weakness in their emulation business.  To understand why this is not a big concern you have to first understand MENT’s rev recognition policy.

The majority of mentor’s license sales are term licenses, which entitle the customer to a certain number of seats (or in some cases, unlimited seats) for a 3yr term.  The company also receives maintenance and support revenues.  At the end of the term, the customer must renew the term license or no longer use the software.  In any given quarter, Mentor’s revenues are composed of term license renewals, new license sales, and Veloce hardware sales, with the bulk of revenue being attributable to license renewals.  With an average term of 3 years, that means that on average 1/12^th of the customer base is coming up for renewal in any period.  Results in any given quarter can be greatly impacted by the renewal rate & terms of that cohort of customers, and to a degree new license sales and veloche shipments.  So at best, revenue in a quarter is basically a reflection on an 1/12^th of the business.  In short, we believe a combination of optimistic Q4 forecasting and lack of early renewals (more on this later) are to explain.

The one issue that clearly impacted mentor in the quarter was Cadence, Mentor’s primary competitor in emulation hardware, releasing a new version of their Palladium offering which we believe is pressuring demand for Veloce.  These are also big budget purchases that are more capex driven and economically sensitive, which could also be contributing to recent weakness here.  We believe this factor should explain about 20M of the revenue shortfall, but we’d note again that this is a lower margin segment that we don’t ascribe as much value to, and with very different dynamics than mentor’s other businesses.

The company also blamed a spate of US semiconductor acquisitions negatively impacting the business, though subsequent commentary on their Q4 call and that of competitors has suggested this is not the case.  Further, when MENT actually reported Q4 results this seemed like less of a factor.  That said, we think the impact from these mergers should generally be minimal for both MENT and the industry.

So if you want to assume that all companies hit their synergy targets, these acquisitions potentially reduce total semiconductor R&D spend by about 3%.  If we assume that EDA spending decline is proportionate, that would be 3% of total EDA industry spend.  Considering Mentor is only half IC companies that would be a 1.5% hit for them assuming they lose their share.  You can make more aggressive assumptions on share shift, pricing pressure, etc and probably arrive to higher numbers, but bottom line these mergers should not drastically impact the business over a longer period of time, especially if you generally believe others will step in where the big players cut, which is generally what has happened historically.

If you focus on US based Semiconductor companies ex-Intel, you start to get more meaningful numbers: a 1.8B decline in R&D on a base closer to 15B of R&D spend, which would get you closer to 12% of total US R&D spend.  Further, we know that about 70% of renewals this year for Mentor are IC, so translated on their base of business that could be a hit of closer to 8.5% of this years revs.  When you further take into account a Q4 renewal cycle this year weighted disproportionately towards North American customers, and further factor in the fact that we have no idea which customers are coming up for renewal in Q4, you start to see how this can have an outsized impact in a given quarter.   There are other areas as it relates to contracting that make this even more complex, but rather than continue to rant we can follow up in Q&A on other factors influencing impact in a given quarter.

The Early Renewal Conundrum:

I wrote the majority of this write-up prior to the Q4 earnings release.  Although I can’t prove it and the company denies it, I believe a big part of Mentor’s weak Q4 and guidance for a weak H1 has to do with changes in historical practices associated with renewing large customers, renewing them at their natural renewal point vs. renewing them a quarter or two early, While this is positive development in the long-term that should improve cash flow and profitability, in the short-term is has the effect of creating an “air pocket” as we are currently witnessing and calling into question the  health of the business.  I can answer questions on this in the Q&A, but prefer not to post specifics in the write-up.

Ultimately, we believe the business is healthy and that recent weakness is not a sign of customer losses or other competitive issues.  Deferred revenue was basically flat year over year, and service & support revs (mostly software maintenance revs) were up 6.7% in Q4 despite the sharp decline in license revenues.  It’s also noting that Cash from Ops, which had lagged historically, more closely approximated adjusted ebitda for the full year, further suggesting cash collections are more closely approximating reported GAAP and non-GAAP profitability.

Activism Opportunity

We think Mentor has excellent technology, and from a technology leadership standpoint, the company is run very well.  The CEO, Wally Reinhold, is very well respected in the industry and has excellent customer relationships, which is where his focus lies.  That said, we believe is not effectively run from a financial or sales standpoint – we think there are several areas of the business where mentor is subscale and losing money, where they should not be competing (most notably in design software).  We also believe the company continues to be run like a “country club” and that there are meaningful opportunities for the business to be run more efficiently and with better incentives to drive shareholder value.  We’d also point to a culture of “early renewals” and general disdain for investors as inconsistent with maximizing profitability and share price.  Although there has been some activist involvement in the past, we think the complexity of the software and multiple disparate segments they operate in have been an impediment to someone more sophisticated getting involved and unlocking the true value that exists here.

There is tremendous value in the pieces of the business and a wide discount between where Mentor trades vs. its peers and similar companies.  We think this would be a very attractive acquisition for private equity in particular, who could sell off certain assets to strategics and focus on the areas where mentor dominates (physical verification).  We have heard industry rumors that multiple companies have tried to acquire Mentor over the years, but that management and the board have not been receptive.  Below is an indicative sum of the parts valuation on the business:

It’s worth noting that Carl Icahn acquired a stake in 2011 and ended up winning 3 board seats.  Although we have not talked to them directly, our impression is that he was very focused on improving the operating margins and lost interest in his efforts to push them for a sale at some point in his involvement.  He deserves credit, however, for modifying management incentive structure whereby management needed to hit 20% op margins to get a substantial portion of their bonus (which they have done every year since this was put in place).  As Icahn appeared to lose interest in the investment, his board members slowly moved on (with the last one resigning earlier last year.  Icahn recently began selling his stake (including half of which was repurchased by the company at what we believe to be attractive prices).  He had the opportunity to sell much higher in (the high 20s), so we believe this selling has little to do any unique view on Mentor specifically, and more likely a source of funds decision.

Valuation Comps:

MENT is exceptionally cheap relative to EDA & Mech CAD peers just about anyway you slice it.  We would argue the vast majority of this discount is not due to business quality or growth prospects but instead due to the other factors discussed.

Path from here:

Ultimately, we think Mentor has dominant positions in several niches, and that the hangover of early renewals should fade later this year, setting up solid mid-single digit growth for the foreseeable future.  At closer to peer multiples, the stock should have about 50% upside.  There is also significant optionality from a constructive activist getting involved and pushing the company to shutter money losing businesses, adopt a more shareholder friendly communication policy, and explore strategic alternatives for the business.