Breakthrough Ideas (3-11-03 edition)
Don Mosher continues........
An Order of Magnitude Leap in Complexity.
Complexity favors the architectural design and engineering teams that have already mastered spread spectrum. Yet, Qualcomm's direct competitors are trying to jump into, but without leapfrogging over, what is for them a new spread-spectrum paradigm years after Qualcomm has already scaled its learning curve. Step-by-step learning from experience enables Qualcomm to continually and rapidly evolve a stable, but advancing, platform of chipsets that serve as the brains of multiband, multimode, and multimedia functionality, and to do so by using a simplified ZIF architecture that reduces board size, BOM, and power drain. In addition, Qualcomm has developed a data optimized solution, which was faithfully translated from advanced modulation theory, which remains well beyond these spread-spectrum-beginners' ability to duplicate, much less to incorporate into world phones.
Not only was Qualcomm better prepared when it began, but its task was far simpler. Returning to the 19-page summary of "The Rise of the 3G Empire," Deutsche Bank Securities indicated: "Back in the good old days of first-generation technology, DSPs operated at 5 volts and only had to handle 10 million instructions per second (MIPS), primarily because AMPS is an analog system. Then along came digital technology and the MIPS requirement quickly jumped by a factor of at least two to four. As wireless systems move from second-generation voice-circuit-switched networks to voice-and-data, packet-switched networks, the required processing jumps dramatically by as much as 1,000 times."
Thus, Qualcomm's direct competitors must undertake solving a set of interdependent problems that require orders of magnitude more MIPS because the complete set of requirements for advanced coding are currently so much more complex. And, instead of incrementally developing design rules that simplify the task by using engineering experience gained from solving similar problems in multiple prior generations of spread spectrum families, they must try to match Qualcomm's advanced performance in a first or second generation chip.
[Although some analysts, like Herschel Shosteck, are beginning to note that Asian experience with advanced features in handsets gives them a significant competitive advantage within European markets, no one appears to have yet grasped that the advantages come in large part from Qualcomm's advance along normal trajectories of performance within their families of chips. Each generation's advanced performance comes not only from the addition of more MIPs in keeping with Moore's law, but also from the engineering experience gained in using isomorphic Task Structure Matrices to solve problems within the spread spectrum Design Structure Matrices within the cumulating series of ever-advancing ASIC families. If Asian handset manufacturers are advantaged, so too is Qualcomm. Just examine the features in the MSM6250 to understand the difference in the intrinsic quality of Qualcomm's offer from chipset relying on first- or second-generation chips from Nokia. Then, compare, when networks build out, the form factors in the handsets, the interoperability solutions for moving from GSM to WCDMA to CDMA networks, and the clear advantage in gpsOne solutions.]
Whereas the GSM standard was optimized for simplicity in the RF system, CDMA systems require solving interdependencies without a continual recycling. Engineers must avoid solving one problem of interdependency by undoing the original solution, which now requires another new one. The requirements for power control and conservation, for additional memory, multimedia, color screens, connectivity, position location, and more also scale up complexity. Whether you originally selected the best scientific solution or settled for an inelegant one, these complexities still compound exponentially.
Your engineers must adhere to the design rules rather than creating a one-off solution. The solution to any set of interdependencies must be tested to verify that it balances the tradeoffs and that the integrated system works. Only then can the engineers begin to improve other aspects of system performance, like reducing the power requirements to the minimum, or reducing their coding so that it does not use up too much memory, or concentrating on reducing form factors of the mobile handset.
Each piece of the puzzle is hard enough, but assembling the whole becomes exponentially more complex, beyond the ability of any single expert, requiring a task structure developed through time in which teams of experts have learned to work and communicate together in order to eliminate interdependencies and to reduce fruitless recycling.
From the experience of producing nine generations of spread spectrum chip families, only Qualcomm has thoroughly mastered the power management and found the very best radio wave solutions for spread spectrum. Everyone licenses their seminal and necessary intellectual property to access the breakthrough ideas in their fundamental design rules, but the countless small refinements that produce the best synergies must be found anew by each engineering team. Each must move down the experience curve to discover each of the endless small improvements that ever more tightly integrate and improve system functioning.
On the one hand, Qualcomm built CDMA2000 technology atop its earlier successes with a set of simpler trial-feedback-retrial cycles from cdmaOne technology. The advanced set of CDMA2000 design rules was built upon two crucial foundation stones: the fundamental CDMA design architecture and the engineering experience with cdmaOne.
One the other hand, the European learning curve becomes sharply steeper because they not only must find acceptable tradeoffs within a set of multiple interacting factors that they have not worked with before, trying to discover which constituents of what interdependencies degrade the overall system performance. This requires the iteration of many cycles of trial-test-feedback followed by retrial-retest-feedback. And, newer solutions cannot be built on top of either older GSM or cdmaOne solutions.
Solving the original set of CDMA problems was still rather complex. After all, Ericsson said it could not be done. But, it was, if you will, the beginner's set of spread spectrum problems. Moreover, it was simpler because Qualcomm's intended commercial solution was based on an ideal theoretical solution. Solving each set of problems, in turn, became the foundation of an integrated learning base in spread spectrum technology that enabled Qualcomm's discontinuous leaps forward.
Qualcomm, as an organization, developed a set of technical capabilities and an immeasurable treasure trove of knowledge, some specifying ideal architectural design rules, some based on accumulated experience in disregarding failed routes in favor of tried-and-true engineering strategies that permitted solutions based on small and incremental refinements. All acquired, bit by bit, through excellent training, hard work, shared experience, communicated across feedback loops between the different teams involved, feedback from developed measurement techniques, and volumes of practical spread-spectrum problem solving experience.
So far, neither Nokia nor NTT DoCoMo has solved many of these complex problems. Indeed, many do not appear close to being solved. This is why Qualcomm must help them with interoperability testing using Qualcomm MSMs. The only commercialized 3G networks, in Chandler's sense of "commercial," are the 18 [28?] CDMA2000 networks.
The comparison to placing a man on Mars is somewhat less wild than it seemed when you remember how few MIPS it took the U.S. to send a man to the moon, less than in your PC. [Please note that this was written lightheartedly months before the recent Columbia reentry disaster with it sad loss of seven astronaut's lives.]
Without vast experience in developing, designing, and installing cdmaOne systems, the order-of-magnitude increase in complexity to leap a generation from 2G GSM to 3G WCDMA leaves GSMs' TDMA engineers trying, but failing, to solve unfamiliar WCDMA problems that are simply too time-consuming and complex without guidance from the expert masters of spread spectrum. Because it is necessarily far easier to solve simpler than more complex problems and much easier to accumulate a knowledge base from experience with previously solved problems, Qualcomm leads the technology race in multimode solutions for GSM/GPRS/UMTS. In addition, only its ZIF architectural breakthrough permits tight integration of multiple modes in an MSM.
A Stranded Installed Base.
To call the GSM installed base "stranded" would be hyperbole if, and only if, GPRS to EDGE to UMTS offered a viable evolutionary path. But, even were its longer path eventually proved viable in an economic sense, its technology still would be not evolutionarily compatible. Instead, it remains a rival revolution without the usual revolutionary benefits of greatly advanced performance.
Hence, the bottom line in the 3G standards war is that the battle is defined by Qualcomm's evolutionary CDMA2000 and a rival revolution in UMTS, which is not and cannot be made compatible with GSM's air interface. This provides Qualcomm with a significant competitive advantage both economically from its first-mover to commercialization and technically from its integrated learning base.
The proposed GSM=GPRS=EDGE=UMTS migration path is long and torturous, prolonging the life of the installed GSM base at a 300 billion-Euro expense. Because it is impossible to ever evolve GSM to UMTS, the proffered migration path adds unnecessary expense by adding far from optimal technology intermediaries, GPRS/EDGE. These "migratory" data-technologies remain GSM-based technologies using time-slots.
After this so-called migration, the carriers must still swap out their TDMA airlink for a WCDMA spread spectrum airlink. This expensive and pointless GSM/GPRS/EDGE migration robs the Peter of Voice to pay the Paul of Data, when Peter is already petered out. Europe and Cingular desperately need increased voice capacity, yet they consume it in order to claim they have a second-best data solution.
European vendors and 3GPP commit a fraud on the public by consistently obfuscating the total incompatibility of the GSM-Second-Generation RAN with the WCDMA-Third-Generation RAN. This so-called migration is a futile exercise in expensive but inadequate GSM-upgrades. It buys time in the hope of developing a mature UMTS, but it does not eliminate the swap-out. Only a stable platform that evolves by using Moore's Law's plunge off the cliff of cost per MIPS can sustain and expand the advance of third-generation networks. UMTS has yet to cross the faux-migratory foothills to climb the mountain to reach the cliff.
A stranded installed base of 2G GSM cannot effectively compete with the fitness bestowed by a naturally evolving and expanding platform of third-generation architectures.
Missing Capabilities.
Dr. Alfred Chandler (2002) based his concept of an organization's integrated learning base on careful analysis of the histories of the evolving learning paths in the consumer electronic and computer industries. He argued that first-movers created their industries by creating an integrated learning base that embodied the organization's technical and functional capabilities. Chandler (p. 238) recognized, "Those learning bases provided the economies of scale and scope, the head start in cumulative proprietary learning, and an assured income that together created powerful barriers to entry."
Although this historical account contains important lessons for investors in high technology, my tale must soon end or I will grow too old and perish. But, these paragraphs from Chandler (pp. 248-249) give you his argument's essence:
"One way to emphasize the uniqueness of the evolution of the consumer electronics and computer industries is to depict their story in the manner of a classic epic. In Greek drama, the gods set the stage for human action but the Fates often intervene to alter the course of events. In this epic drama there are two major protagonists, both U.S. companies that shaped their industries worldwide. One protagonist, IBM, follows the virtuous path of learning for maintaining the long-term profitability and growth of the enterprise. The other protagonist, RCA, after succeeding by taking the virtuous path, is lured away and falls from grace. By the 1960s each is challenged by smaller, more agile Japanese enterprises, which acquire IBM and RCA learning. IBM comes through the ensuing competitive battle with historically unprecedented success. RCA's deviation leads to its own self-destruction and that of its national industry."
IBM developed an integrated learning base that enabled it to define and shape the architecture of the mainframe computer, to sustain control of that industry, and to ignite the PC revolution. IBM followed the virtuous learning path, using its income and learning base, to introduce world-dominating new products like the 360 and 370, the world's first modular computer systems. However, its Future System project failed because of its immense complexity. According to Chandler (pp. 248-251):
"The Greek term is hubris. IBM tried to do too much. …Then, the fates intervened. First, one of IBM's own senior managers is instrumental in making the Japanese industry a viable competitor. Next, the Japanese industry makes a crushing attack on the U.S. memory chip business. But at that same moment IBM unintentionally rescues the U.S. industry by mass-producing and mass-marketing a product first commercialized by youthful hobbyists. It does so almost at the cost of its own life. In the process, again unintentionally, it turns two of its suppliers, selected by middle-level managers, into the industry's path definers in the U.S. PC sector, which quickly dominates the industry worldwide. Nevertheless IBM survives and in 1966 remains the industry giant.
On the other hand, by 1996 RCA, the protagonist in the U.S. consumer electronic industry, pays the ultimate price for leaving the virtuous path; it dies, and so does the U.S. industry. In the 1960s RCA achieves world dominance comparable to IBM's and does so by following the same virtuous path. But then it turns to a false path where it lives on hope more than on learning. First, it challenges IBM despite having little in the way of learned capabilities - some in product development, but certainly none in production or, particularly, in marketing. Second, it turns to acquiring enterprises in which its managers have no learning at all. In terms of the Greek epic, RCA is lured by the Sirens from the business press, the academy, and Wall Street to the rocks of corporate disaster."
Chandler continued by describing how this permitted the Japanese to use the virtual paths of learning to conquer world markets in consumer electronics and the European and Asian markets in computing.
My point is that an integrated learning base of organizational capabilities, which is demonstrated by successful commercialization in national and international markets, is the vital and indispensable foundation for dominance of world markets.
Following the virtuous path allows the first-mover Qualcomm to improve and expand the architectural platform that shapes the future of its value web and its industry. Companies can founder reaching for complexity that is beyond their grasp; whole industries be lost by managerial hubris. If you trod beyond the boundaries of your integrated learning base, beware because dragons, Sirens, and the Fates lurk in that terra incognita. Europe's hubris, together with spread spectrum's complexity, dooms it to lose its dominance of mobile wireless communication. Qualcomm stands by, a dominant friend offering its own Marshall Plan to rebuild a devastated Europe.
Some spread spectrum technical capabilities in the integrated base of organizational learning in Nokia, NTT, and 3GPP are missing. So to is strategic leadership. Failing to understand the significance of a complete standardized solution, 3GPP employs a flawed political process that fractionates potential standards-based solutions into competing, incompatible, and company-proprietary products. 3GPP, including NTT, is missing "strategic leadership" because no one sees or intervenes to halt fragmentation's looming threat to roaming, the putative basis of its 2G GSM strategic success.
The 3G-first-mover Qualcomm's integrated learning base now shapes and controls this industry's future. Asia contains the new markets and a growing horde of fast-followers on the virtuous path of 3G CDMA, a path already worn smooth by footprints from the earlier well-trodden Asian dynamic paths in computers and consumer electronics. Superior fitness of third-generation CDMA will vanquish the installed-base of the second-generation GSM-leader, just as the sun rises in the east. How many suns will rise first, nobody knows.
Market Data 
Markets 
AI
