Description

Business Description:
Creates and sells electronic design automation (EDA) software used by designers (mostly electrical engineers) of integrated circuits (ICs). Synopsys has a virtual monopoly position in the logic synthesis segment of EDA software (39% revenues), and has a 50% share in the verification & test segment (32% revenues) for ASICs. (Cadence has the other 50%.)

Industry Note: the ASIC design process requires software to design, analyze, simulate, model, implement and verify electronic designs. The software is an essential tool – making engineers more productive, enabling designs not otherwise possible, improving accuracy of complex designs and shrinking time-to-market schedules. Generally speaking, the ASIC design industry is a duopoloy, with SNPS dominating the so-called “front end” of software design (logic design - the architectural equivalent of plotting the details of a house on graph paper) and CDN dominating the so-called “back end” of software design (place & route).

Investment Thesis:
I believe SNPS is a long because:
1) Software upgrade cycle: new advanced software versions (costing2-4x times the older versions) are required for deep sub-micron IC designs
2) Better industry pricing: new subscription license model adds greater revenue visibility and less pricing pressure
3) Better earnings quality: new subscription license model insures that ALL payments collected AHEAD of booked revenues
4) Growing R&D budgets at customers: software purchases funded out of R&D budgets, not capex budgets
5) High barriers to entry (salesforce): industry incumbents typically win more share of the market during upgrade cycles, especially when there is no change in software language
6) Historically low valuation: SNPS currently sells at 5-6x ’01 CFFO, as a result of a recent earnings announcement that lowered expected top-line growth from 20% to 17%
(show me a stock what can you buy for 6x CF that grows at 17%)
7) $500m share buyback program just finished; new $500m program just initiated


My research edge
Spoke w/ many designers at leading ASIC design firms
Spoke w/ management of SNPS and chief competitor, Cadence

Detail on Investment Thesis:
1) Software upgrade cycle: new software costs 2-4x times the older versions
Deep submicron chips (i.e. < 0.13) are nearly impossible to design with current software – it’s like trying to run your office building on a dial-up modem instead of a T-1 line. In logic design, where SNPS has >80% market share of a $325m market, the newest SNPS product (Physical Compiler) has just started rolling out and is expected to reach $110-$125m in orders by FYE (Oct ’01) – that’s a minimum of $30-$40m per year in revenues for the next 3 years. (Licenses must be renewed every 3 years.)

Assume Physical Compiler costs 3x the older product, Design Compiler; also assume at least 50% of the market must upgrade to the new software: the logic design market will grow to $600-$700m, which means SNPS will double its sales in this segment. (COGS associated with software licenses is minimal, and SG&A/R&D do not need to change – so all of this goes to the bottom line.)

2) Better industry pricing & earnings quality: the new subscription license model
Historically, under the perpetual license model, customers would wait to buy software on the last day of the quarter, knowing SNPS and CDN needed to book orders and revenues to meet Wall Street expectations on book-to-bill. SNPS and CDN were effectively held hostage on pricing. In addition, payments lagged revenues, though collection was good in this industry because software seats were easily shut off.

CDN adopted the subscription license model in June ’99. SNPS followed in Oct. ’00. Licenses are about 3 years in duration. Both companies have much better visibility into customer renewals, and book-to-bill is meaningless, so there is much less pressure on pricing. (In my conversations with customers price was barely mentioned as key to the purchase decision, particularly for newer - i.e. costlier – products.

In addition, the new model requires that all payments be received prior to revenue being booked, creating huge amounts of deferred revenue – and the major source in addition to net income for the estimated $400-$500m in CFFO expected for FY01. Think of it this way: 50% of next FY’s revenues are already “in the bag” and virtually no new orders are required for SNPS to make their goal of $900m in revenues for FY02.

3) Growing R&D budgets
Historically in downturns, R&D always grows at semi and system companies, though more slowly. The EDA budget is typically 1/3 of R&D tends to be less affected than other parts of R&D. Most importantly, design engineers are rarely fired – rather they are moved from non-core to more essential projects. A common industry saying is that “you have to design your way out of downturns.”

This downturn has proved slightly different because of the heavy debt loads at Lucent, Nortel, etc. Nonetheless, SNPS continues to grow orders at 17% yoy (20% had been the target). The slower growth has come from customers upgrading, say, only 10 of the 20 software seats. (This of course is only a “deferral” of revenue, hardly a “lost” revenue.) Most customers are renewing old licenses on schedule, and the upgrades are occurring at high premiums. (One customer noted that they have an internal library that automatically renews all software seats as necessary, while upgrades are typically approved by the division manager.)

4) High barriers to entry
Before Synopsys, the first step in chip design was done on graph paper – and usually in-house. SNPS essentially created the logic synthesis segment about 10 years ago and has maintained > 80% share ever since. No new entrant has managed to break SNPS’s stronghold on logic design, though a few have tried – including Cadence. (This is consistent with other areas of EDA software, as well.)

The sources of SNPS (and for that matter CDN’s) advantage are captive customers and economies of scale related to salesforce and distribution. For customers, switching costs are extremely high – chip designers never want to learn a new software language or a new product (it’s 10x the pain of migrating from Lotus to Excel), and only do it when absolutely necessary. However, upgrade cycles do put customers up for grabs and tend to invite competition from upstarts. The last line of defense is a significant investment in sales personnel, making the cost of customer acquisition very high. Lastly, in downturns, customers tend to consolidate their suppliers and rely on a few healthy ones, as opposed to start-ups, etc.

5) Current valuation is historically low
On 8/31/01, SNPS’s stock price was $46.14, valuing SNPS’s equity at $2.81b.
Net Cash is $391m, so Enterprise Value = $2.42b.

FY01 CFFO $450-500m - P/CF ’01: 5x
For value investors, the best way to value SNPS is on CFFO. SNPS currently trades at 5x FY’01 CFFO - historically the lowest it’s ever been (and certainly the lowest it’s been at the beginning of an upgrade cycle). I challenge fellow value investors to find a $1bn+ company with top-line growth >15% that sells for 5x CFFO.

FY01 revenue: $680m-$690 - EV/’01 revenue: 3.5x
FY02 revenue: $890m-$910 - EV/’02 revenue: 2.7x
The market and most tech investors value SNPS on sales. (If anyone is interested I’m happy to discuss why in the Q&A.) Admittedly as a result of the subscription model, comparisons are difficult; but current revenues are MORE stable and earnings are of HIGHER quality, so higher multiples are warranted. On the other hand, the underlying growth of engineers entering the business is slowing (although still at 8-10%+ yoy). But in this next upcycle, new & more expensive products will allow yoy growth of 15-25%.

Historically over the past 10 years, SNPS has been valued as low as 2..25x sales and as high as 7x or 8x sales, depending on when/where in the upgrade cycle and what’s happening in the underlying semiconductor business. Despite current weakness in the semiconductor market, current valuations at 3x revenues are very low, compared to past upgrade cycles (and this upgrade cycle is the largest yet in terms of price premiums).

I believe SNPS deserves a 5-6x revenue multiple – this normalizes the trough semiconductor industry valuations and adjusts for the upgrade cycle. At such multiples, SNPS is worth $70 - $85 per share in under 12 months and over $100 per share into FY ’03.

FY01 EPS: $1.10-$1.15 - P/E ’01: 40x
FY02 EPS: $2.15-$2.25 - P/E ’01: 20x
EPS multiples are almost meaningless given the change in license model, though the actual eps numbers are obviously closely watched. The numbers are presented for completeness.

6) Share buyback program
$500m buyback program just completed, $500m new program just started. SNPS has traditionally been aggressive at $44-49/share. On the recent conf. call, they bought back over 1m shares @ $52.
Admittedly, some is offset by option dilution, but definitely not all of it.

Catalyst

1) aggressive share repurchase program
2) ramp up of software upgrade cycle (occuring in spite of tech downturn)
3) affirmation of 17% top line order growth for this year and next
4) very high earnings quality, as expressed in CFFO as % of net income