Description
THIS IS MY ANNUAL UPDATE ON TRANSWITCH, WHICH REMAINS MY LARGEST PERSONAL HOLDING.
MY APOLOGIES IN ADVANCE IF SOME OF THE TABLES DO NOT COME OUT CORRECTLY BECAUSE I HAVE LARGELY CUT AND PASTED FROM A REPORT THAT I RECENTLY WROTE. THE STOCK HAS MOVED A LITTLE IN RECENT DAYS SO THE PRICES IN THE REPORT ARE VERY SLIGHTLY STALE. I HAVE NOT BOTHERED TO REVISED THE PRICES AND I BELIEVE THE STOCK IS BOTH ATTRACTIVE AND TIMELY HERE.
ALSO, BECAUSE THIS IDEA CONSIDERABLY EXCEEDS THE MAXIMUM WORD COUNT, IT WILL BE CONTINUED IN THE Q&A SECTION.
I. INTRODUCTION: THE COMPANY AND THE INDUSTRY
THE COMPANY
The name of the company is TranSwitch and the ticker symbol is TXCC. The company was founded in 1988 by its current head, and has been public since 1995. There are currently about 195 employees. The company conducts business globally, but corporate headquarters are in Shelton, Connecticut.
Currently, there are about 95 million shares and the recent price is about $1.08, which is essentially the 52 week low. Over the past year, the stock has traded as high as $4.
As can be seen below, during the heady days of the telecomm bubble, this stock was an institutional favorite and at one point traded as high as $75. In 2000, the company merited the number six spot on Fortune’s list of fastest growing companies.
Despite its low share price and lack of meaningful sell-side research coverage, the stock retains a substantial institutional investor base. At June 30 of this year, large shareholders included Capital Guardian (12%), Firsthand (4%), Fidelity (4%), Pequot (3%) and Oppenheimer (3%).
THE BUSINESS
The company designs very highly specialized semiconductor chips for use in advanced telecomm networks throughout the world. Important customers include all the large global manufacturers of telecomm equipment, including Cisco, Nortel, Fujitsu and Samsung. The ultimate end customers are the telecomm service providers—the Verizon’s and Bellsouths of the world, both large and small.
In many cases, the company’s chips will enable the critical functionality of the next generation of telecomm equipment. Carrier requirements here are particularly rigorous, but product cycles are extremely long (5-10 years), and once the company’s chips are accepted into the network, they are not easily replaced.
COMPETITION
The markets in which the company competes are highly specialized and require a great deal of cumulative intellectual property, as witnessed by the company’s large patent portfolio. In general, it is this accumulation of industry knowledge that probably constitutes the greatest barrier to entry. For the most part, the company’s products address niche markets which are unlikely to attract the interest of the larger semiconductor companies. Furthermore, service provider requirements are extremely rigorous, and the competition is limited to the relatively select number of players who can meet these requirements.
Competition has historically come from three fronts:
1. The larger, public companies including PMC Sierra, Vitesse Semiconductor, and Applied Micro Circuits, as well as divisions of larger firms such as Agere and Infineon.
2. Smaller venture capital funded startups such as Galazar and West Bay.
3. The internal divisions of the large equipment vendors such as Lucent and Nortel, who often implemented customized chip solutions.
In recent years, competition on all three fronts has markedly diminished, owing primarily to the severe and protracted downturn that has gripped the industry.
During the bust, the larger competitors not only reduced aggregate R&D investment spending, but also substantially redirected their focus towards other markets which were expected to recover faster than their core telecomm market. Each of the three large “pure play” competitors, diversified into storage and enterprise applications, which now represent almost half their revenue stream. While these diversification efforts allowed for a faster initial recovery in revenues, each of these companies has ironically reported disappointing recent financial results, owing largely to shortfalls in these newly developed or acquired product lines.
Venture capital funded firms, often a source of innovation, have been constrained by the lack of venture funding over the past several years. The few that have been funded have far less ambitious business plans, and must operate on a smaller scale.
Finally, the internal chip design divisions of the large equipment vendors have all been sharply cut back during the massive industry-wide retrenchment. Importantly, as chip designs increase exponentially in complexity, it will be increasingly impractical to design custom chips. PMC, for example, has estimated that the opportunity for standard merchant silicon, relative to custom silicon, will increase from 30% in 2000 to 83% in 2005, an almost threefold increase.
THE EVOLVING INDUSTRY
There are two important types of traffic: voice and data. Voice traffic involves a direct circuit connection between two endpoints. In a traditional telephone call, for example, all traffic takes one specific route from end to end, not unlike two soup cans connected by a string. Data traffic, on the other hand, is “connectionless,” as each packet of data can take any of a number of divergent routes, so long as it ends up at the correct endpoint. By way of example, imagine driving from uptown to downtown: depending on traffic patterns you might take any of a number of different routes to end up in the same place.
An important conceit during the boom days was that the internet would drive virtually limitless growth in data traffic. What’s more, a new technique to transport telephone calls over the internet had been developed, and forward looking technologists anticipated the day when even voice calls would be transmogrified and transported as data traffic. Traditional phone networks were designed to carry voice traffic and could at best be crudely adapted to carry data traffic, and so the tried and true public telephone network was increasingly viewed as a thing of the past. New, data-centric networks would be required to fulfill the transport requirements for this burgeoning data traffic, and so a new breed of carrier arose to build out the networks that would meet this demand.
In retrospect, several important truths have emerged:
• While data has become the dominant form of traffic, continuing to almost double every year, and voice traffic is at best stagnant, the data-centric business model was flawed. Simply put, voice traffic is still where the money is.
• The traditional voice telephone network is the result of hundreds of billions of dollars of investment. In addition to being paid for, it was also well understood and extraordinarily reliable. Carriers were just not about to dismantle these networks so quickly.
• Traffic patterns were importantly miscalculated, as the importance of super high speed, long haul traffic (e.g. LA to NYC) was overestimated, but the importance of slower, but more granularly provisioned metropolitan traffic was underestimated.
THE INDUSTRY OPPORTUNITY
As the industry begins to move forward once again, the fundamental question facing the carriers remains how to transport, and profit from, this exponential growth in data traffic. Budgets are still constrained and the notion of all new, data-centric networks remains a vision that, for the most part, continues to lie well in the future. For now, the central issue facing service providers is, how to efficiently transport ever increasing amounts of data traffic over networks designed for voice traffic? The answer lies in a set of recently devised protocols, collectively known as Ethernet-over-SONET/SDH or Data-over-SONET/SDH.
While the discussion here can quickly become highly technical, I will emphasize several important concepts:
• Ethernet is the protocol for almost all local area networks, such as those found in an office or campus. SONET (Synchronous Optical NETwork), or its European equivalent SDH (Synchronous Digital Hierarchy), is the underlying telecomm transport protocol that links buildings and cities with each other. As the importance of data increases exponentially, so does the importance of transporting this data from one building to another, or one city to another. In other words, the inter-office network is spreading out to become the intra-office network.
• These new protocols dramatically increase the efficiency with which carriers can transport data over their existing voice-optimized infrastructure.
• Implementation of these new technologies requires only relatively modest incremental capital costs. The vast infrastructure core remains essentially unchanged, and only the edges of the network need to be upgraded. .
• Newly upgraded systems will allow service providers to greatly enhance their data transport offerings, enabling a wide range of new and profitable services, including broadband internet access and virtual private networks.
Because these new technologies solve a number of important problems that carriers face, they have been rapidly embraced by the industry and indeed are spurring a generational upgrade in telecomm equipment. The critical mapping and switching capabilities of this new generation of equipment lie almost entirely in chips such as those made by company and its competitors.
Over the past three years, at a time when its primary competitors had substantially refocused on other opportunities, TranSwitch redoubled its efforts in it core telecommunications markets. Whether through foresight or circumstance, the company has very adeptly anticipated the evolution of the industry. As a result, the company has been early to market with a number of key products, and is now poised for substantial and enduring gains in market share.
* * *
II. FINANCES AND VALUATION
THE BALANCE SHEET
The balance sheet is adequate. At September 30, 2004, cash and debt balances were as follows:
Cash & Marketable Securities $143 million
Less: 2005 Debt $ 25 million
2007 Debt $ 92 million
Net Cash $ 26 million
Operating cash burn was approximately $10 million last quarter. The company has guided towards operating breakeven by mid-2005, but my latest sense is that this is more likely to happen in late-2005. This breakeven will be achieved through a combination of revenue increases, as new products begin to contribute, and a dramatic decrease in expenses as R&D expense returns to a more normal level.
Anticipated cash burn between now and breakeven is about $25 million, which in the absence of any other financing would leave the company with about a zero net cash balance. Gross cash balances would still be significant, but the company would ultimately have to refinance its 2007 debt at some point.
The 2007 debt is a complex but clever security. Under certain conditions, including a stock price greater than $5.47, the company can force the conversion of any or all of this debt into shares. This would be an ideal outcome, leaving the company with about 25% more shares but a debt-free balance sheet.
Management is financially conservative, and has expressed a desire for greater equity financing. Most recently, they exchanged about $6 million of the 2007 debt for equity at an all-in price of about $1.62. I have advised them that I believed that such a conversion was both dilutive and unnecessary.
INCOME STATEMENT
The normalized business model for the company is appealing, with a generally favorable margin structure. Because the company outsources manufacturing, capital expenditures are modest and the threshold investment is in research and development.
The table below shows the company’s revenue progression during the favorable but relatively placid years prior to the telecomm boom:
Before the Boom: Good Growth, Good Margins
INCOME STATEMENT Q3:97 Q4:97 Q1:98 Q2:98 Q3:98 Q4:98 Q1:99
Revenues 7,438 8,334 8,606 9,607 12,264 13,692 14,461
Cost of Goods Sold 2,828 3,249 3,354 3,728 4,537 5,026 5,222
Gross Profit 4,610 5,085 5,252 5,879 7,727 8,666 9,239
Research & Development 2,227 2,353 2,543 2,537 2,887 3,086 3,131
Marketing & Sales 1,796 1,711 1,818 1,925 2,420 2,536 2,637
General & Administrative 523 680 564 589 672 691 719
Total Operating Expenses 4,546 4,744 4,925 5,051 5,979 6,313 6,487
Operating Income 64 341 327 828 1,748 2,353 2,752
As can be seen, during these relatively normal times the company was able to generate substantial increases in revenues and earnings owing to its ability to create and sell appealing niche products.
The table below shows the company’s performance in the seven quarters leading up to its peak revenue quarter at the end of 2000. As can be seen, during favorable times for the industry, the company can enjoy extraordinarily high profit margins owing to high gross margins and the relatively fixed nature of its other expenses.
The Boom Years: Hyper Growth, Hyper Margins
INCOME STATEMENT Q2:99 Q3:99 Q4:99 Q1:00 Q2:00 Q3:00 Q4:00
Revenues 16,119 18,537 22,290 26,747 34,072 42,600 51,664
Cost of Goods Sold 5,633 6,305 7,283 8,542 10,412 12,506 14,687
Gross Profit 10,486 12,232 15,007 18,205 23,660 30,094 36,977
Research & Development 3,203 3,527 4,153 4,424 5,408 6,011 8,378
Marketing & Sales 2,826 3,153 3,363 3,868 4,437 5,818 6,395
General & Administrative 800 894 961 1,038 1,537 1,205 1,854
Total Operating Expenses 6,829 7,574 8,477 9,330 11,382 13,034 16,627
Operating Income 3,657 4,658 6,530 8,875 12,278 17,060 20,350
As has been well documented, the telecomm industry went into a freefall during 2001 and 2002 as new investment in the communications infrastructure dried up virtually overnight. After two extraordinarily difficult years, which saw service provider spending decline by more than 60%, the company’s revenues finally bottomed in the fourth quarter of 2002 at a mere $3.6 million. The results since then have been reflective of a slow and unspectacular recovery, as can be seen in the table below.
After the Bust: A Slow Recovery
INCOME STATEMENT Q1:03 Q2:03 Q3:03 Q4:03 Q1:04 Q2:04 Q3:04
Revenues 4,098 6,607 5,685 7,430 8,205 8,687 8,348
Cost of Goods Sold 1,340 1,772 1,612 1,695 2,146 2,404 2,500
Gross Profit 2,758 4,835 4,073 5,735 6,059 6,283 5,848
Research & Development 10,230 10,468 10,656 10,680 13,669 11,972 11,230
Marketing & Sales 2,682 2,710 2,784 2,788 3,331 3,212 2,789
General & Administrative 1,472 1,248 1,374 1,535 1,871 1,792 1,506
Total Operating Expenses 14,384 14,426 14,814 15,003 18,871 16,976 15,525
Operating Income (11,626) (9,591) (10,741) (9,268) (12,812) (10,693) (9,677)
Several line items here should be noted with considerable care. First, even in the worst of times, gross profit margins are healthy, attesting to the company’s highly specialized products. (There is a modest benefit in the above numbers from the sale of inventory that had previously been written off.)
Second, and more importantly, despite the plunging revenue line, the company dramatically stepped up investment into research and development, leading to a period of disproportionately large operating losses and cash burn. In fact, over the past two years, the company spent almost $95 million on R&D, well more than double the $38 million it spent during the two years leading up to its peak revenue quarter.
As can be seen, total operating expenses peaked in the first quarter at $18.9 million, as the company spent heavily to complete its product line. Operating expenses declined to $15 million in the most recent quarter and are forecast to decline to about $11 million over the three quarters, even as revenues climb sharply. The company has targeted an income statement that might look something like this by mid-2005:
INCOME STATEMENT Q2:05
Revenues 17,000
Cost of Goods Sold 6,000
Gross Profit 11,000
Research & Development 7,000
Marketing & Sales 2,750
General & Administrative 1,250
Total Operating Expenses 11,000
Operating Income 0
As can be seen above, under the currently targeted expense structure, operating breakeven occurs at about $17 million in revenue. The company has stated that it hopes to achieve this revenue level by mid-2005, although at the moment it looks like this goal will not be reached until later in the year.
Importantly, to the extent revenues do not ramp as anticipated, the revenues required for breakeven can be reduced further, to as low as $13 million or so. In this case, R&D expense would be reduced to about $5-6 million per quarter, a level which would still enable the company to both support existing products and introduce new products, albeit fewer than would otherwise be the case.
Because the company has spent so heavily on R&D in recent years, a sharp reduction now is unlikely to affect growth prospects until 2008 or so.
VALUATION
The current valuation of the company is as follows:
Market Capitalization, equity $102 million
Market Capitalization, debt $ 99 million
Less: Cash $146 million
Enterprise Value $ 55 million
Another way to put the enterprise value in perspective is to consider that over just the last three years, a span of time that approximates the investment period for the coming generation of products, the company has invested about $147 million in research and development.
In other words, for the relatively modest sum of $55 million, the investor can buy the following:
• A legacy business that, even at today’s depressed levels, provides a gross profit contribution of $20 million per year with no real expenses attached thereto.
• An important, and relevant, body of intellectual property, including a patent portfolio of 135 patents issued and 126 patents pending.
• An R&D investment in next-generation products on the order of $150 million. This R&D investment has already resulted in both significant new relationships, such as Cisco, as well as significant new products, including the following:
o Ethermap-3, Ethermap-12 and Ethermap-48
o Envoy 8FE, Envoy 2GE, Envoy CE2 and Envoy CE4
o PHAST-3N, PHAST-12N, PHAST-48V and others
o VTXP-6, VTXP-12 and VTXP-48
o TEPro-FX28
o PacketTrunk-4
The design, fabrication and testing of these products is largely complete, and R&D investment has already begun to decline materially.
As it relates to the company’s valuation, the threshold questions are,
1. What will be the timing and pace of the industry’s eventual recovery?
2. What role will the company’s product play in this recovery?
If the company’s R&D has been even remotely on target, then the likelihood of a meaningful increase in revenues over the next few years is very high. Although current revenues are low, and visibility still quite limited, it is my belief that the company’s efforts will be well rewarded, as the real growth in telecomm spending is likely to be directed towards the company’s areas of focus.
A COMPARISON WITH PMC SIERRA
PMC Sierra (PMCS) would generally be perceived as the largest and most important telecomm chip company. In recent years, the company has significantly diluted its telecomm focus by expanding into other areas such as enterprise and storage applications. These businesses are generally much shorter cycle and often carry slightly lower operating margins. In the past, pure telecomm chip companies have typically sold at a higher multiple of revenues and earnings.
Below is a valuation comparison between TranSwitch and PMC. The best valuation metric is probably a comparison of enterprise value to annualized quarterly revenues. Because I view the company’s recent R&D investment as critical, I have also included a comparison of that metric.
PMCS TXCC
Market Capitalization (MM) $1,953 $ 103
Enterprise Value (EV, MM) $1,629 $ 54
EV / Revenues 5.7x 1.6x
EV / R&D (Current Cycle) 4.2x 0.4x
Over the past year, PMC has been profitable, although in recent months its businesses have slowed significantly and it is now operating near breakeven, with a loss projected for the fourth quarter. Since June, the December quarter revenue expectations have declined by almost 40%, consistent with the sharp decline in the share price. Even at these depressed prices, however, PMC stock carries a healthy valuation, consistent with the low end of its historical valuation range; in contrast, the company’s shares are trading well below their historical valuation range. PMC has a comfortable balance sheet, and TranSwitch does not; however, this does not come close to explaining the glaring disparity in valuation.
In the past, the company’s shares have typically traded in an EV/Revenue range of 4x to 10x, with a median level of about 7.5x. If the industry recovers, and the company is once again viewed as a rapidly growing provider of high margined semiconductor devices, a reasonable and “correct” valuation would be in the range of 5x to 7x revenues, equating to a P/E multiple of 25x to 35x under a normalized margin structure. If this revaluation were combined with a much improved revenue outlook, the resultant share price could be very substantially higher.
RISK VERSUS OPPORTUNITY
The major risk, at this point, is that revenue development lags and the company continues to burn cash. It is difficult to envision a scenario where the telecomm industry does not resume some level of growth, and the company’s products do not play some reasonable role. The timing of the recovery continues to progress at a measured pace and a meaningful upward inflection in growth is not quite yet at hand.
Regardless of how revenues develop, expenses are set to come down quite significantly, substantially reducing the rate of cash burn. Management is acutely aware of their net cash position, and has formulated a backup plan should revenue recovery be further delayed.
Management has expressed a desire for greater equity financing, and in the past has exchanged modest amounts of debt for equity in open market transactions, effectively increasing the company’s net cash position by reducing debt. Management will likely continue to effect these exchanges in a modest and incremental way and, to the extent that the recovery continues to lag, the cumulative effect of those exchanges may be significant. In such a case, worst case potential dilution might be as high as 25-35% which, in my mind, is palatable considering the upside once business turns.
While finances are tight here, I believe that management is focused and the probability of true financial distress here is quite low. The potential for creeping dilution is real, but the cumulative effect should be acceptable.
* * *
Catalyst
Rising revenues as a new telecom investment cycle begins.