Description

Overview

Applied Micro (APM) is a small cap fabless semiconductor company going through a major product and technology transition in which a fickle stock market has shortsightedly lost patience and finally confidence.  APM has made a huge bet developing X-Gene, an ARM architecture based System on a Chip for servers.  X-Gene will be the first ARM based Server on Chip to challenge the Intel x86 dominated server processor market.  Not only is X-Gene first to market with a 12-18 month lead, but also it has a technology advantage over competing ARM offerings.  X-Gene offers a “brawny custom core”, integrated switch fabric and integrated I/O (from the leading 100 GB Ethernet player) giving what we believe is a technology moat today that competitors will not match 12 months from now.

Perception of APM’s base business has been challenged for the past two years as its PowerPC architecture based embedded processor products have been in decline.  While this was been expected, it accelerated in the June quarter just reported and will culminate in a big revenue decline expected for FY2015 Q2 ending this September.  We believe the rest of the base business, however, is solid, will soon resume growth, and the low margin, cash cow, soon to be obsolete PowerPC business was never more than an immaterial addition to all the future cash flows this company will produce.

A massive R&D spend on X-Gene and the decline in PowerPC has masked a base business which is resilient (all products are sole source sockets with 3-7 years of life), high margin (60% gross margins) and which should grow rapidly from here as new products are announced and adopted.  Today you are able to buy APM for less than the value of the base business with huge optionality from the launch of the X-Gene SoC, which has already eliminated technology risk and is ready for commercialization.

Setup

In August, a more rapid than expected, decline in AMCC’s legacy PowerPC architecture based embedded chip business knocked the stock down 30% despite the products being known to be close to end of life, very low margin, and immaterial to the long term profitability of the company (probably $10-12 million in cash flow delta versus expectations prior to FY2015 Q1 earnings in July).  More recently, positive sentiment around the competiveness of Intel’s newest server processors has raised misplaced doubts among some investors about whether ARM based server processors will gain traction, knocking the stock down another 20%. 

Valuation

At a stock price of $7.22, APM trades at a market cap of around $596 million and enterprise value of $515 million (not including $100.1 million of federal and state R&D tax credits and federal and state NOLs of $1,137.3 million and $522 million respectively).  While, APM is expected to be roughly break-even in its current fiscal year (FY2015 ending March), this projection includes over $92 million of R&D spending.  Approximately two-thirds of this R&D spend is on a multitude of early revenue and pre-revenue products that if successful would significantly increase the size of the business.  Conservatively assuming R&D at 23% of sales instead of 46% would generate pro-forma operating income of $46 million and EPS of 54c (assuming no taxes due to the NOLs).  Comparable chip companies spend between 15% and 30% of revenues on R&D and we believe a low to mid 20s rate would be sufficient if APM was not trying to tackle a momentous task like an integrated server on a chip. 

Assigning a 14x multiple (as the base business should grow from here on its new lower base) + $1/share of cash would yield an $8.60 stock price or 20% above current prices.  So basically you’re getting the X-Gene and associated ARM product opportunities (X-Weave, Helix and likely others) for a discount when they could be worth multiples on the stock.  APM has spent approximately $530 million on R&D in the past 13 quarters (more than its current enterprise value) including $178.5 million acquiring Veloce (which has been the driver of the X-Gene product development).  The market is saying the entire ARM server on a chip opportunity, which has been technologically de-risked, has negative value.

Business Description

Applied Micro has two reporting segments: Connectivity, which you can think of as chips for

Telecommunications and data networking equipment and Computing which you can think of as embedded microcontrollers and processors for a multitude of devices.  APM will report revenues from its new X-Gene chips in its Connectivity Segment; however for clarity, we will discuss X-Gene here as a separate segment.

Connectivity

In its existing Connectivity segment AMCC supplies a broad range of physical layer, framer, and mapper solutions including high-speed mixed-signal PHY silicon solutions for networking and routing applications. Think telecommunications Wide Area Network equipment both copper and optical (hence the price spike to over $400/share in 2000).  This segment also serves the continued demand for higher speed data networking equipment such as the transition in datacenters from 10 GB Ethernet to 40GB and 100 GB Ethernet.  While this business does suffer from the ups and downs in telecom equipment capital spending, it should be a growth business with APM’s leadership position in 100GB Ethernet and the launch of the X-Weave products for datacenter. Importantly, APM’s products are sole sourced and long lived.  APM is not fighting for its life every year.  Sales in FY 2014 (Ending March) were $108.5MM growing from $90.7MM in FY2013, down from $110.3MM in FY2012 and $123.8MM in FY2011.

Computing

In its Computing segment AMCC embedded processors (historically largely PACKETpro and other PowerPC-based) into a multitude of product categories such as wireless LAN, residential, data center, and enterprise applications where they typically handle overall system maintenance and management functions.  The conventional wisdom is that AMCC has underinvested in new products for this segment as PowerPC architecture has fallen out of favor and AMCC has focused R&D dollars on ARM architecture products in its Connectivity segment (X-Gene and X-Weave).  The reality is that there is a small gap between the availability of their new ARM based embedded processors and the decline in their PowerPC business. AMCC has actually developed and won significant product design wins for its Computing segment based on the ARM architecture, which should ramp in late CY2015.  Furthermore, APM will announce a new family of embedded ARM processors called Helix next month at ARM TechCom (http://www.computerworld.com/article/2684311/components-processors/arm-64bit-processors-could-come-to-multifunction-printers-too.html).  Computing Sales were $107.7MM in FY 2014 up from $105MM in FY 2013 down from $120.6MM in FY 2012 and $124MM in FY 2011.

X-Gene, the ARM Server Market Opportunity.

APM’s X-Gene family of products consists of X-Gene, the first production SoC (Server on a Chip) and the X-Gene based X-C1 Server Development Platform and their follow on products.  APM has built the X-Gene SoC combining 10/40G mixed signal I/O with custom ARM v8 64-bit “Brawny” cores running at up to 2.4 GHz.  Importantly, APM purposefully built the X-Gene SoC for datacenter applications with custom  “brawny cores” (which competitors using the Standard A57 cores cannot match) providing 80-90% of the  power as an Intel Xeon chip without Xeon’s legacy baggage from being developing for a multitude of applications.

X-Gene sampled to customers in early 2013 and recorded almost $1 million of revenue in the June quarter (1-2 quarters before most expectations).  Management continues to expect meaningful revenue from X-Gene to begin in the December and March quarters.  Early shipments should be seen as the “tip of the iceberg” as datacenter operators will spend maybe $2 million to make small initial deployments to be followed the next 2-4 quarters with deployments 5-10x that size if everything works as planned.     That said, X-Gene has essentially been de-risked from a technology perspective as early customers and OEMs have been putting it through the “meat-grinder” for the past two years.

X-Gene 2, the next generation silicon on 28nm sampled to customers in spring 2014 and should begin shipping in late CY 2015.  The Gen 2 device at 28nm is partly a process shrink but APM has also made some optimizations to the CPU cores, while taking advantage of the higher frequency and lower power consumption of the 28nm process.  APM expects an incremental 15-20% improvement in performance.  X-Gene 3 is also under development and is planned for 16nm, fabricated using FinFET process technology.   

Market Opportunity

Today, the estimated $12.3 billion (CY 2013) server processor market is dominated by Intel’s x86 architecture and its Xeon processors.  These processors are powerful, expensive (thousands of dollars), draw a lot of power and create a significant amount of heat.  However, a significant portion of the server market does not need the computing power offered by these processors.  Applications where input/output functions are more important than computing power potentially can be served (pun intended) by lower power consuming, less heat generating, lower price point chips designed specifically for the datacenter using the ARM architecture, and integrated switch and I/O functionality onto the chip.  ARM architecture today dominates tablets and cell phones for precisely these reasons.  While enterprise datacenters running ERP and HR applications don’t likely fit this bill, cloud datacenters serving web traffic and storage applications most likely would be better served from a lower Total Cost of Ownership on the system level compounded by potentially even greater savings as Intel’s monopoly pricing in server chips could be challenged as x86 processors would not be the only solution. 

Our bullish thesis is based on APM capturing 2% of 50% of the server processor market (excluding enterprise servers) or $200 million of this growing market place.  Servers for datacenters is growing 20%.  At 60% gross margins, $120 million of operating income should fall straight to the bottom line as the expense base has already been supporting X-Gene R&D for three years.  Approximately $1.42 in EPS at a 20x multiple (due to growth profile) would result in a $28 stock price or 4x the current market price.  Given the moat around X-Gene’s technology moat relative to the other ARM competitors (brawny core, switch and I/O integration), its time advantage and the potential size of the market, its not impossible that earnings power could be as much $3.00 and at 15x you have a $45 stock.

Intel

The most recent 20% stock decline has been blamed on the most recent Xeon product release announcement.  However, while I can find many people who can tell me that that is why the stock is down, industry experts tell me there is nothing new or unexpected in Intel’s new 18 core Haswell chip.  Furthermore, the fact that these new chips do not integrate I/O continues to leave the door wide open for ARM to capture meaningful share of certain submarkets.  Furthermore, the desire to have an alternative to Intel will leave the door open for alternatives to Intel.  While some might insist that no one has managed to fight intel in the PC, the server market, particularly the massive datacenters employed cloud players such as Facebook, Google, Amazon and Yahoo have the ability and desire to use other than shrink wrapped software, break the Intel monopoly, and improve their cost structures. 

Risks

Competition from Intel

Competition from other ARM server processor vendors

ARMv8 server software ecosystem does not continue to develop

Decline in telecom equipment demand could impact base business

Decline in use of the Internet or Wireless connectivity would impact both new products and the base business.

Catalyst

Turnover in shareholder base (which we believe is basically done this week).

ARM Techcon on October 2^nd where new products and design wins will (Helix)/may (others) be announced.

Announcements from HP regarding their Moonshot Server Products and deployments expected in Q4.

Earnings