According to CSCO's latest report, WIND-powered Cerent business is approaching a $ billion run rate.
Allen - Does that statement appear credible to you, and - if true - what kind of business should this be generating for WIND ?
As I recall, Cerent's optical switches sell in the $40K range. A $billion run rate implies about 25,000 units shipped worldwide annually. It doesn't seem impossible given Cisco's sales channel.
I honestly don't know how much of the $1 billion would go to WIND, but I can guide your through various ways you can use to estimate WIND's portion.
1. Think ports not switches.
Ports are a common unit of measure for determining royalties for switching devices. 25,000 switches at, say 20 ports each on average, would be 500,000 ports annually. In a competitive OS world, the high-volume OS royalty per port might be significantly less than $1. However, if the semiconductor company supplying the network processor has a single reference design based only on VxWorks, a higher royalty might be acceptable. To the extent that TMS is deployed, the royalty would be increased substantially, say to somewhere between $1 and $5 per port, say $2 for guesstimating. This approach would suggest WIND's revenues from Cerent would be running in the neighborhood of $1 million annually.
2. Think software contribution in the value-chain.
MSFT's software contribution in PC's has steadily increased over the years, to the point that it now approximates 10% of the retail ASP. In the embedded world, the software contribution is about .5% to 1%, and rarely greater than 2% of the Bill of Materials (BOM). The relevant BOM for a Cerent switch probably would be the network processor and chip set. If each processor costs, say $100, then the royalty per processor normally would be 50 cents to a dollar. However, once again, we know from last year's Q1 CC that TMS boosts the value-add to more like 5%, which would be in the $5 per processor range. If each switch has 4 processors on average, this guestimate would be in $500K range. If each switch has 8 processors on average, the $1 million becomes reasonable.
3. Think Switching Costs.
This is not well-known, but it is one of my favorite ways to estimate revenues from a design win, or even better, from a market niche. You should realize that this approach is the only appropriate way to establish product pricing.
The idea is this. What would it cost Cerent to junk all of WIND's products, replacing them with the most appropriate alternative. By cost, I mean the total economic cost, including not just the direct cost of conversion, and any fixed or variable costs (like product licenses and royalties), but also indirect costs of slower time to market, reduced reliability, reductions in functionality, etc. The most WIND should expect to charge, and retain Cerent as a customer, is the present value of all these economic costs, direct and indirect. The only rational for WIND charging significantly less would be strategic, such as, keeping the barriers to competition high, or to provide extra incentives to capture in-house developers.
The fact that Cisco paid $7 billion for Cerent says there is huge value there that needs to get to market as fast as possible. To the extent that the Tornado/VxWorks platform and TMS are woven into that success story tells me that Cisco wouldn't dare switch WIND at almost any price. Even if all software were ported to Cisco IOS for future product development only, I would think the risks of the less-powerful development environment slowing down product development would generate huge, indirect costs. If the opposite is true, and Tornado/VxWorks/TMS represents a minor add-on, tangential to the real work of product development, then an annual stream of as little as $500K or $1 million could justify a switch, thereby limiting the WIND's revenues to relatively small amounts. The size of the actual revenues depend almost entirely on whether Cerent deems the Tornado/VxWorks/TMS platform essential for getting to market quickly and reliably.
The main reason I like the switching-cost method of valuing design wins is because it takes into account competition, the main force limiting prices. Change the competitive landscape by consolidating the players, by adding value through software verticals, or through alliances with semiconductor companies, and the pricing picture changes, possibly dramatically.
The switching-cost method provides the best explanation as to why the embedded systems sector has been so slow to meet the revenue expectations of investors. During most of the nineties, competition centered on the OS. The diversity of applications prevented the emergence of a dominant player through economies of scale or network effects. Multiple "adequate" solutions kept switching costs relatively low, and therefore limited the opportunity for any vendor to hike prices much above the norm. As the constant leader in innovation, WIND was always able to out-revenue the competition, but not sufficiently to attract a sizeable following on the Street.
If there was any company that could challenge WIND in the late nineties, it was ISI. No one else was even close. By application of the switching-cost method, we now know that WIND's value-add just changed dramatically with the merger, from being the price ISI charges plus the additional economic value that Tornado/VxWorks brings to the table, to the cost of, say, in-house development plus the huge economic value added by Tornado/VxWorks. This is not chump change. We are talking order of magnitude stuff here.
But that's not all. Look at how Tornado/VxWorks/TMS are dominating the network processor space, including Intel and the exciting IXA, IBM and its counter-punch intelligent network processor, and virtually all the other network processor companies like PMCS, MMCN, GALT, etc. The PR released at the InterOP makes it clear that the focus has shifted to much more than the OS. Integrated protocol stacks and management functions of every stripe are requirements for an adequate software solution. The huge switching costs that would be required to build a solution on top of an ordinary kernel, which sort of runs with a third-party compiler on the processor of choice, give testimony to the increase in WIND's value-add.
With dominance comes network effects which accelerate the value-add even more. To switch to another solution in the face of this value proposition would be absurd except under extreme pricing pressure by WIND. It follows by the switching-cost revenue estimation method, WIND can look forward to the inflection point investors have always expected in this space. Perhaps this is what Curt Schacker, VP Marketing and Business Development, meant in a recent article when he stated he believed WIND was encountering an inflection point.
Allen |
