POST OF THE DAY
New Paradigm Investing
Re: A Good Sweating

Format for Printing

Format for printing

Request Reprints

Reuse/Reprint

By Scollag
March 22, 2004

Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!

Most people on this board agree that ARM is the gorilla in wireless but disagree about its gorillahood in other areas such as set top boxes.

Include me out.

I think that view derives from an imperfect understanding of how ARM operates, and of the value proposition the company offers. I believe ARM's dominance is much greater than this sector-oriented view suggests, and covers the entire embedded RISC market generally.

ARM's licenses are sector-agnostic. When ARM sells a multi-use license (and multi-use licenses are the mainstay of ARM's license revenue, according to ARM), it can be used for whatever purposes the client conceives. The same license can be used to produce chips that are used in multiple sectors, and individual chip designs may themselves be used in more than one sector. Adoption of the ARM architecture is driven primarily by license sales, and licenses are sector-agnostic. SUDLs are exceptions, and usually the intended purpose is disclosed in the license announcement, precisely because there is a single chip product planned, which is usually intended for a specific nominated application.

Sector percentages of royalties and units are estimates. ARM places no requirement on clients to reveal the end use of any shipped cores. Indeed, in some cases the client may not know precisely to what use the chips are put. Of course, some chip designs are very sector-specific, which must help with the estimates, but my point is that perhaps we sometimes place too much significance on figures that ARM themselves do not necessarily regard as definitive (which is not to say that they are not as accurate as ARM can determine). This again strongly suggests that the sector in which the shipped core is eventually applied is less important than some here seem to think.

Current royalties are derived from licenses signed several years ago. Tracking of quarterly or annual trends in shipments and royalties ascribed to particular sectors reveals nothing about current licensing activity, but about a series of events, decisions, and trends that have occurred over the last decade or so. Shipments do not occur without design wins by the client. Design wins do not occur without design starts. If a particular design start is to include an ARM core, then it must be preceded by a license agreement. A single license agreement can generate multiple design starts, each of which can generate multiple design wins. Each design win can spawn various levels of production and shipments.

The wireless sector dominates the embedded RISC market. I believe that the rapid adoption and development of the cell phone, principally through GSM, over the last ten years or so has led to a concentration of semiconductor design activity in this sector that we may eventually regard as disproportionate. I think we can infer from the way the technology has evolved that there have been more opportunities for chip design wins in this sector, and probably more design starts given that the rapid sector growth has attracted fierce competition. I freely admit I have no source data here, but it seems intuitively obvious to me that the growth rates we have seen sustained, and the turnover of models we have seen, must have created more opportunities for design wins, which in turn must surely have created more design starts, than more slowly-growing sectors. Adoption of the ARM architecture depends on both license sales and design starts, and I suggest the reason that the wireless category features so prominently in ARM's shipment and royalty data has more to do with the number of design starts and design win opportunities in this sector over a number of years, compared to those in other sectors, than with any inherent difference in the ability of the ARM architecture to penetrate individual sectors. That is, ARM may already have achieved similar dominance in other sectors, but they have not matched the pace of innovation in the wireless sector (yet) and ARM depends on design starts and design wins for shipments - which apparently is all we watch.

The development project is at least as important as the product. I have been banging this drum for so long that perhaps it has become background noise, but in my view the advantages of the value proposition that ARM provides are not restricted to the price/power/performance qualities of individual cores. For a potential client planning a design project, the choice of a commercial, off-the-shelf (COTS) core rather than a home-grown one offers a faster, cheaper, and safer route to market. ARM is the only IP vendor able to provide COTS cores with these proven advantages, and it helps that in general they are in the sweet spot of the power/price/performance matrix. With all due respect to the achievements of ARM's design engineers, I think it is incorrect to ascribe the success of the architecture purely to the technical qualities of the cores, and to infer that those technical qualities disqualify the architecture from any sector.

ARM cores are already shipping in automotive, networking, secure, storage, imaging, consumer entertainment, and microcontroller sectors. The pace of adoption and penetration may lag wireless, but as design win opportunities arise, and as overall shipments ramp up, I am confident we will see increasing dominance of the ARM architecture in these sectors, reflected finally in shipment data.

Sikit, sikit, lama, lama, jadi bukit*, as they say in the kampongs of Kuala Terengganu.

Scollag.

*Little by little, slowly, slowly builds the hill.


Become a Complete Fool
Join the best community on the web! Becoming a full member of the Fool Community is easy, takes just a minute, and is very inexpensive.