Replies to post #1732 on Qualcomm Inc (QCOM)

[Data_Rox]

Breakthrough Ideas (3-7-03 edition)

Residual Uncertainties About Qualcomm's Potential.

A key question for the investor to ask himself is: What's the level of residual uncertainty around any potential investment? Answering that question requires answering questions about residual uncertainties in a company's potential, profitability, and competitive position.

My analysis has focused on competitive position, emphasizing Qualcomm's strategic architectural control, resulting from its breakthrough ideas that permitted a 2G-performance play strategy and a 3G-evolutionary strategy for commercializing a spread spectrum platform that opened a window to strong and sustainable profitability.

I introduced a nascent theory of dynamic paths that permits the advantaged use of information by filtering it through the lens of theory, specifically to search for key events that signal significant, consequential, or momentous change, including sudden reverses or dramatic changes in scale in the form of tipping points or inflection points.

In Part II, I claimed that the ETSI decision in January 1998 marked a threshold of significance, which was confirmed in March 1999 by Ericsson taking a license from Qualcomm and purchasing its infrastructure division. These momentous events, first, reduced the residual uncertainty as to which architecture would dominate third-generation mobile wireless by committing the industry to spread spectrum. Second, Ericsson acknowledged by contractual licensing Qualcomm's ownership of CDMA's proprietary architecture. If you will, this event changed the Level of residual uncertainty from Level 2, a competition between two standards, to 1, a forecastable stream of 3G royalties for Qualcomm.

Moreover, these seminal events marked a technological tipping point, forever altering the balance in the competition for third-generation architectural dominance because of QUALCOMM's superior integrated learning base in spread spectrum. As first-mover to commercialize spread spectrum, its base of profits from royalties alone for all 3G handsets, terminals, cards, chips, and infrastructure could be used to continually advance its evolving and expanding CDMA platform, leaving its direct competitors to play catch-up as Qualcomm shapes the future.

A noteworthy exception to the general rule of residual uncertainties in high tech is the fact that Qualcomm will receive royalties on all manufacturers' ASPs for 3G wireless mobile products. That is, there is unusual certainty that Qualcomm will receive royalties on a large market of third generation handsets, ASICS, and CDMA infrastructure using licensed CDMA IP.

It is indisputable that over 100 telecommunication companies have licensed Qualcomm's intellectual property and each will pay royalties. Although forecasts are rife with residual uncertainty in any high technology marketplace, let's concretize this analysis of Qualcomm's position by tentatively accepting the ITU's forecast of 2 billion 3G users by 2010. Back-of-the-envelope calculations on, say, 4% royalties of, say, at a $75.00 manufacturer's ASP per handset yield royalties of $6 billion dollars for Qualcomm. This is not a forecast but more like a minimum WAG estimate to set a margin of safety.

The usual residual uncertainties about market share and size remain for their MSMs and CSMS. Size and share of demand for products and services possess Level 3 uncertainties, a range of possible outcomes. That is, there is a range of possible outcomes for third-generation mobile wireless because it is difficult to estimate customer demand for new 3G data services, more expensive 3G mobile terminals containing richer features but fewer subsidies, pops penetration, replacement/churn rates, and relative competitive advantages years in advance. When compared to both direct and indirect wireless or broadband competitors, uncertainties remain about customers' preferences for new products and service plans, replacement rates of handsets, a possible gray market in used handsets, the continuity versus the disruption of GSM by GSM1x, personal income growth and distribution of wealth in many markets from India to Russia to Latin America, and the like.

Nonetheless, although we cannot predict the exact size, at bare minimum, Qualcomm will get its royalty share of it all, and some uncertain portion that is bounded, say, by 20% and 80%, if you dare extend the trend lines of the present installed base of GSM and CDMA, or, say, 60% to 90% of CDMA ASIC sales only, as a conservative WAG if you choose to reduce the 90% market share of its 2G CDMA MSMs by one third. Bottom line is, you analyze the stocks qualitatively, look at the estimates of market size skeptically, and allow plenty of room by making sure the lower end estimate is worthwhile and the high-end estimate is deeply rewarding.

Aside from the difficulties of forecasting inherent in a world of residual uncertainties, the logic of this approach remains unusually valuable as a conceptual tool.

First, specifying the level of residual uncertainty clarifies strategic analysis and its implications for investors.

Second, this conceptual framework categorizes each means of value capture by its level of residual uncertainty (RUs).

Third, the level of residual uncertainty interacts with a company's strategic choices. Specifically, a shaping strategy successfully implemented in actuality sharply reduces residual uncertainty about the outcomes that its succeeds in shaping.

Strategic action impacts the future, and if effective, strategic actions transform the range of potential outcomes over time from unbounded uncertainties into bounded uncertainties into mutually exclusive probabilities, and finally into decisive actualized outcomes. Each of these shaping transformations that further reduce residual uncertainties impact bottom line profits.

Fourth, we can monitor the dynamically shifting tides of levels of residual uncertainty to assess a company's effort to shape its future through both its strategic actions and its subsequent economic outcomes. We watch our companies march through time to monitor its dynamic path along the stepping-stones of competitive advantage that lead to success. Each step along a dynamic path, once achieved, reduces residual uncertainties.

Let's turn to a consideration of RUs and value capture and then on to RUs and shaping strategies.

Each element in the value capture entries of the business design may have residual uncertainties that not only differ dynamically across time but also that distinctively echo the specific dynamics of their profit zone. Because both drivers and strategic control points can differ across products and profit zones, then it is the actual interaction of competitive and complementary forces with industry drivers that generate outcomes. Ideal outcomes occur when the features in products and services are tailored to match customer needs and priorities. Whole products and complete standardized end-to-end solutions are ideas that describe optimal degrees of fit.

Turning specifically to Qualcomm's value capture: (1) Segmented Licensing of CDMA IP; (2) Royalties on Manufacturers ASP; (3) Sales of CSMs; (4) Sales of MSMs; and (5) Brew Transactions.

At this point in time, licensing of CDMA IP is necessary for any manufacturer who hopes to compete in the 3G markets. This is true regardless of the form chosen because the key elements of spread spectrum itself, its architectural skeleton, are incorporated into Qualcomm's essential patents. WCDMA may require additional patents for anyone other than Qualcomm. I assume Qualcomm would pass through its cross licensing of WCDMA for all Qualcomm multimode products, but not for products specific to WCDMA, which would require additional licensing from over 100 manufacturers who might make claims.

What might qualify as a key WCDMA patent is not established. And, it is not clear that even NTT, which has relied on its own research, is truly at a commercial level.

Of course, licensing is always far cheaper than doing the R&D, and it ensures an earlier time to market. Only a discontinuous innovation can threaten this patent position by replacing the 3G spread spectrum paradigm.

As the size of the worldwide market expands and as new niches develop, new manufacturers will be drawn into the value web. These manufacturers must license CDMA.

The issue becomes how large will the modular cluster of interrelated manufacturing industries become and which new companies will require patents for their products. For instance, will Wi-Fi incorporate an interface with CDMA or WCDMA, as CDMA2000 will do with Wi-Fi in its 7xxx chipsets?

Wi-Fi, as a nonproprietary standard, can be easily incorporated into Qualcomm's platform, but will manufacturers who start with a Wi-Fi product add other connectivity? Probably not, but an answer depends on still unknown business models and unknowable conditions and circumstances. It is not possible to place boundaries on these uncertainties. Hence, Wi-Fi dwells in the land of still unknown and unknowable Level 4 RUs.

However, there are two exceptions the problem of bounding residual uncertainties: all networks increase in value as they increase their number of nodes, and larger networks attract smaller networks. Notice that these exceptions are based on theoretical principles not on knowledge of future conditions and circumstances. There is nothing as practical as a good theory based on sound principles for predicting the future. Such principles reduce unbounded residual uncertainty by predicting powerful features in how networks scale. Such principles shape the future of all networks.

Royalties on manufacturer's ASPs follow directly from the licensing agreements. The issue here becomes the size of the 3G market itself because Qualcomm is sure to receive the royalties. The category of Qualcomm's market is customarily defined as mobile wireless. However, CDMA2000 1X includes wireless local loops, as in India, or in wireless area networks (WLAN), and the prevalence of browsing the Internet in a mobile wireless mode.

CDMA1xEV-DO may include wireless Internet where there are no better options available from indirect competitors with competing technologies. Again, this potential wireless Internet market has yet to form, remaining at Level 4 RU. Also, this market seems likely to segment by regions. For instance, one the one hand, in the U.S., telephone modems, DSL, TV cable modems, fiber to the curve, and various satellite satellites or fixed terrestrial wireless platforms form an unstable and open set of possibilities that remain unbounded Level 4 RUs, too early to call given a number of competing technologies, for the market shares of each, or for the range of possible shares among a set of competitors.

In India, on the other hand, the first Internet experiences may come from WLL for people without PCs or any other forms of Internet access.

Sales of Qualcomm CSMs and MSMs are based on the size of the 3G market and Qualcomm's share of it. In developed markets, these issues are classified as Level 3 RUs because one can foresee a range of possible futures. It is still so early in the developing 3G market that CDMA versus GPRS/EDGE/WCDMA have no meaningful trend lines, unless you accept the utter dominance of CDMA. But, time will tell.

As I frame it, the battle for market share is an issue of the power of the worldwide installed base of GSM versus the powers of being a first-mover, possessing intrinsic technology quality, and having established architectural control.

Also, anticipating possible futures here is dependent on a decision about whether the market can or will tip. It seems clear that most analysts are not structuring the problem in the way that Courtney suggests, because they continue to ignore residual uncertainties and project 2G trends into a "clear enough" 3G future.

Whether BREW will receive transaction revenues faces considerable residual uncertainty. The market for advanced cell phone applications is still forming.

But, even here analysts partition the debate into a false dichotomy, J2EE versus BREW. The issue is better framed in terms of who offers a whole product, where BREW serves as a complement to JAVA that augments its offer. This is so because JAVA is a programming language and BREW is a telephony-based binary run-time environment with a complete business solution. No matter how much or how many programmers favor Java for programming, carriers, who are ultimately held responsible by customers for quality control, will insist either that Java programmers establish a complete distribution, management, and billing system like BREW, or run their applications atop BREW to ensure control of quality and billing, a MECE Level 2 RU. Manufacturers will fall in line with carriers' decisions.

Some carriers, like Sprint, may believe that in-house development that yields proprietary applications is the best strategy. They will run directly into strong network effects created by the popularity and rapid growth of BREW applications that can run across all handsets and networks. That is, the principle of having many available applications is likely to trump owning a few proprietary applications that are limited to a single carrier. This is an instance where a larger network becomes more valuable and large networks attract small networks.

Initially, Qualcomm has been willing to yield some of its potential 10% transaction fees to augment the 10% of the carrier, using the principle of stimulating uptake of its system in hopes of generating network effects before relying on this revenue as a source of income. If BREW is successful, then Qualcomm will change its strategy to harvest a fair share of financial benefits.

When Thornley described BREW's service revenue as the third leg on the income stool, with royalties and ASICs income represented the other two legs, he clearly signaled Qualcomm's assessment of the immense potential of this market.

In his chapter entitled Shape or Adapt?, Courtney chose Qualcomm as his case-example of an enterprise shaping the wireless telephone industry toward its CDMA standard. If you will, the choice for 3G architectures was between TDMA/GSM and CDMA access modes. Before this choice was made, it represented a Level 2 uncertainty: two architectures competing to become the standard.

The industry, including all competitors, however, chose spread spectrum as the 3G architecture. Hence, given time, the industry reduced the level of residual uncertainty about the use of spread spectrum itself from Level 2 to 1.

However, some contend there is now a choice between WCDMA and CDMA2000 as two forms of spread spectrum. After assuming the market does not tip strongly nor become winner-take-all, notice that this construction of the issue as a choice between two spread spectrum standards creates Level 3 uncertainty, a range of possible outcomes in the market share of each competing standard.

However, I contend that Qualcomm's multimode architectural solution in reality shapes the probable future of this choice, just as its first-mover advantage in innovating and commercializing CDMA shaped the choice of spread spectrum as the fundamental 3G architecture. A seamless network will bring most of the chipset sales' share to Qualcomm because it both creates a more valuable market with more nodes and attracts everyone into a harmonized worldwide market.

Although it might even transform the market into a winner-take-all monopoly, the strategy of harmonization suggests that Qualcomm is adapting its WCDMA chipsets to a less than ideal set of specifications to grow the market now, rather than trying to force a tipping outcome to its preferred and more spectrally efficient CDMA2000 approaches. Qualcomm, once again, seeks a friendly win-win resolution rather than dominating power-over UMTS competitors already within its value chain.

According to Thornley at the September SBB Tech Conference in response to a Q&A request for an update on the MSM6200:

"We brought forward the introduction of the 6200 by one quarter. That is the chip we built into the prototype phones that are being used in testing WCDMA networks currently by most of the operators, most of the manufacturers in Europe. In fact, we have been producing a significant number of these phones as test equipment - and we are actually selling those phones for the test equipment. So we are actually generating some revenue from that. But more importantly, we are using them to accelerate the introduction of WCDMA in terms of interoperability. It is positioning us very well in Europe. And, the European operators are beginning to prescribe to their manufacturers that they should be using our chip which, of course, is very good for our long-term share."

Clearly, Qualcomm strives to shape the market in Europe, to drive it into a significant 3G market by the end of 2004 on the back of its MSM6200 [and MSM6250] as a means to solve the interoperability problems of its fractionating UMTS standard and to create a single worldwide global 3G market using roaming on its MSM6xxx platform.

However, if the market were to tip in a few years, after the competitive advantages of Qualcomm's complete standardized solutions for multimodal harmonization become widely apparent, then either this prediction of the probable reduction in residual uncertainty or its eventual actual occurrence must radically alter all forecasts for market share. Uncertainty collapses from Level 3's range of possible outcomes, through Level 2's mutually exclusive and collectively exhaustive choice of CDMA2000 or WCDMA, to Level 1, a clear enough future dominated by Qualcomm's multimode chipsets.

Hence, if you had confidence in the market size, the collapse of shares into winner-take-all would create a clear enough future to attempt point-forecasts for the temporary monopolistic winner who was now, given the actual occurrence of the predicted winning events, be safely settled on Main Street.

For instance, consider, say, forecasting Microsoft's share and margins for Windows, given a predictable rate of growth in PC market size as it grows worldwide. Given a stable competitive environment and under normal macroeconomic conditions, these point-forecasts assume either a linear or accelerating trends of growth that can be extended with close to Level 1 uncertainty. Because of Microsoft's monopolistic hold on its market, such forecasts might be clear enough. (Unless Linux running on PCs disrupts its dominance.)

What becomes significantly harder to predict is how successfully Microsoft can leverage that Windows advantage into new endeavors. But, even then, there is little doubt about Microsoft's power to leverage its architectural control of Windows to more or less shape new competitions in packaged software. Over time, we have watched Microsoft grow its share in new markets, like X-Box, constricting the range of potential outcomes in Level 3, until we begin to wonder whether it or its competitor, say, SONY will own this new market, a Level 2 issue. This is not to say that the "evil empire" will spread or even continue forever or that there are no limits to generalizing an integrated learning base to adjacent niches. The bowling pin analogy seems well suited to describe the necessary degree of overlap, the pins need to be aligned or the bowling ball's force is wasted.

To know whether Microsoft or others will probably demonstrate monopolistic dominance requires making this crucial distinction: Is this a product-competition or a standards-based-competition? In a standards-based competition, a winner-take-all outcome becomes more probable, given the reciprocal dynamic of increasing and decreasing returns. X-Box appears to be a product competition, but Microsoft.Net may have enough proprietary hooks to become a standards-based competition around two forms of "standard" XML.

The probability of product-based, compared to standards-based, monopoly is less. Also, remember that market contagion extends even to fads and fashions; tornados can spawn exponential growth that cannot be sustained. As we shall see below, growth alone does not account for lasting network externalities. Other things being equal, to know with absolute certainty that a market has tipped decisively requires waiting for clear evidence of an inflection point of pragmatist adoption and a subsequent growth trajectory called a tornado for a decisive answer about a whether a new de facto standard has been established.

In any event, both technology and strategy matter. Taking charge of our present action with an eye to the future that we want to create is a powerful means of achieving human desires.

Shaping strategies are designed to get ahead of the curve of uncertainty, to influence key uncertainties, to shape the evolution of the industry itself by changing its present game (competitive structure) to the game only one company best knows how to play and win.

Courtney offered five categories of shaping strategies: (1) creating an industry standard; (2) introducing product, service, or business system innovations, (3) restructuring the industry; (4) replicating existing business systems in new markets; and (5) influencing competitor conduct.

Let us apply Courtney's five categories of shaping to Qualcomm. Qualcomm shaped the industry by: (1) creating CDMA 1X and 1xEV-DO as industry standards; (2) introducing gpsOne, the BREW platform and BREW Distribution System as both product, and business system innovations, with its virtual marketplace and digital signature serving as new services; (3) restructuring the industry toward a harmonized multimode, multiband, multi-network of interoperable networks; (4) replicated existing business system in new markets by playing a series of Asia cards; and (5) influencing competitors' conduct by forcing them into a longer and more expensive migration path and to continually try to play catch-up to its rapidly evolving and expanding platform.

Hence, Qualcomm changed the third-generation game to the game uniquely suited to its own integrated learning base that enables both rapid evolution and augmentation of advanced performance and also complete standardized solutions in mobile wireless. Although residual uncertainties remain in its future, Qualcomm shapes 3G as only a company with strategic control of a proprietary open architecture can.