"To the lights and towns below,
Faster than the speed of sound,
Faster than we thought we’d go…"
-Smashing Pumpkins, 1979
During the Internet bubble, John Doerr of Kleiner Perkins frequently decreed that the Internet was in fact "under-hyped." He argued that despite the wild speculation and runaway stock prices, people generally failed to appreciate the significance of the first globally interconnected communications network. I am now prepared to make the same bold proclamation regarding the rise of the 802.11 communications standard, also known as WiFi. Despite all the press and the hype, I believe that 802.11 is remarkably under-hyped relative to the massive impact this seemingly simple standard will eventually have on the entire wireless communications sector.
One clear lesson in the history of technology and business is that once an open standard gains critical mass, it is extremely hard to derail. The x86 computing architecture and the Ethernet networking standard are two salient examples of this truism. Once a single interoperable standard gains the acceptance of multiple vendors in a marketplace, a consumer bias toward compatibility and scale economics create an increasing returns phenomenon that is nearly unassailable.
Open standards obtain a high "stickiness" factor with customers as a result of compatibility. Once customers invest in a standard, they are likely to purchase more and more supporting infrastructure. As their supporting infrastructure grows, their switching costs rise dramatically with respect to competitive alternate architectures. Customers are no longer tied simply to the core technology, but also to the numerous peripherals and applications on which they are now dependent. All of these things make challenging an accepted open standard a very difficult exercise.
Favorable scale economics result from two key characteristics of open-standard architectures. First, when several companies all support the same standard, the architecture benefits directly from the collective R&D efforts of all players in the marketplace. This applies both to direct innovation with the standard, as well as an ever-growing community that builds peripherals and applications that reinforce the standard. The second reason that open standards have a powerful economic advantage is that high volumes lead to lower prices. Lower prices then increase the market for said technology, which in turn drives even higher volumes. The cycle then repeats, driving even lower prices and ever increasing usage.
These dynamics were unquestionably present in the last quarter century evolution of the computing industry. When IBM introduced the x86-based personal computer in 1981, most analysts had very little idea as to the usefulness of such an expensive and seemingly limited personal computer. Moreover, no one could have ever estimated that the processor inside that device would become a near universal standard throughout the entire computing industry. And while many technologists would tell you that the x86 architecture is anything but elegant and certainly far from optimal, Intel recently shipped its billionth x86 processor. Along the way, many proprietary computing architectures have resultantly faced obsolescence. There have also been many well-funded attempts to unseat the standard (remember PowerPC?) — all to date unsuccessful.
Founded in 1980 by Intel, DEC, and Xerox, Ethernet has enjoyed a similar success story in the networking industry. Prior to the 1980’s, most computer makers followed a vertically integrated strategy, and as a result each had their own networking standard. However, as more and more vendors jumped on board the Ethernet bandwagon, each of those networking architectures slowly faded away. While it started as a LAN technology, today Ethernet is being used in the WAN (wide-area network) and even the MAN (metro-area network). What’s more, it is now being used for voice and video, two applications that were never envisioned when the standard was introduced.
Simply put, 802.11 is to wireless communications what the x86 is to computing and what Ethernet is to networking. This "open-standard radio" is today supported by more than 115 vendors with more than 900 certified products. The collective R&D of Intel, Broadcom, Cisco, and Motorola as well the entire venture capital community will move this technology further and further along the price performance curve. Already in five short years, a backwardly compatible 802.11g chip offers about 25x the performance at about 1/20th of the price of the first generation radios in this market. As before, these low price points are leading to increased market opportunities and lower and lower prices. Currently 802.11 radios are a 50 million unit per year market, although history would suggest this is merely the beginning.
With some 802.11 radio chips approaching $5 price points, WiFi will likely be embedded in every electronic product under the sun. This pervasiveness will impact the communications market in two remarkable ways. First, vendors that build supporting infrastructure and applications will come to assume that WiFi is on board, further entrenching the standard. Perhaps more importantly, as a client technology, 802.11 will increasingly be considered "free." In the wireless communications world, the cost of client technology (sometimes referred to as CPE for consumer premises equipment) typically has a huge impact on overall system economics and therefore adoption. With "free" CPE, 802.11 will have a distinct competitive economic advantage.
There are three patterns that emerged in previous open-standard architectures that are likely to play out in the open-standard radio market as well. First, numerous vendors underestimated the importance of backward compatibility. Second, vendors were amazed at the performance evolution of the key interoperable standard. Lastly, and as a result of the first two points, everyone underestimated the scope and pervasiveness that the standard eventually encompassed.
Let’s start with compatibility. It is quite clear today that one of the key advantages of the "Wintel" architecture is the huge investment that customers have made in terms of software applications as well as training and education on those applications. Switching costs are massive, and therefore projects such as OS/2 and PowerPC failed to gain real traction, despite arguable technical superiority. The same is true of competing networking standards such as Token Ring and FDDI. These technologies, while also potentially technically superior, are simply too expensive to implement when you consider total cost of ownership. 802.11b is already being used in more than 50 million devices, and by next year will be in 50 million more. That is a massive installed base of "free" clients that will represent a significant hurdle for any challenger. Ironically, this hurdle can even be an impediment for other open standards that lack customer momentum. Vendors pushing 802.11a and 802.16 will find this to be a significant challenge.
While 802.11b already enjoys quite impressive performance at 11Mbs, many companies building competitive offerings are quick to highlight the relative shortcomings of the technology. Cellular equipment providers will tell you that 802.11 cannot support mobility or voice. UWB (ultra wide band) chip manufacturers will tell you that 802.11 has "too much" range and not enough channels. 802.16 chip vendors will tell you that 802.11 has "too little" range. All of these vendors are hazardously ignoring the potent impact on innovation of collective R&D investment. 802.11 will not sit still. Before you know it, the performance gap (especially on a value per dollar basis) will quickly narrow. The x86 processor has increased its MIPS performance by 2x every 18 months. Ethernet performance has increased 10x every three years. The same will happen with open-standard radio, and those that promote the weaknesses of the standard are merely writing the feature list for future innovation on top of the standard.
As a result of compatibility and superior price/performance, experts will continually be surprised by the increasing scope and application of open-standard radio. Many suggested that Bluetooth would occupy many of the sockets now owned by 802.11. Likewise, many vendors now believe that a new standard, UWB, is needed for wireless communications in consumer electronics. However, system companies such as Linksys and chip companies such as ViXS are showing that these same applications can be tackled without abandoning the open-standard radio that already has momentum. Likewise, while some vendors will argue that 802.11 doesn’t do enough to support roaming, mobility, voice, or range, engineers across the globe are hard at work improving the standard to do just those things. Vocera, TeleSym, and Meru have all provided value-add on top of open-standard radio and are now deploying compelling voice solutions.
Additionally, citywide networks deployed using Tropos’* open-standard WiFi cells already support roaming and mobility. From a performance standpoint, networks built using Tropos already support average download performance levels that are twice those of Verizon’s EVDO beta deployments. Upload speeds are twenty times higher. What’s more, these networks have significant advantages in cost of deployment and operation.
Make no mistake about it. 802.11, or one of its backwardly compatible descendants, will dominate the wireless communications sector over the next 10 years the same way that the x86 architecture dominates computing and that Ethernet dominates networking. There will be numerous doubters and numerous challengers, but they will all succumb to the inescapable power of the first true "open-standard radio." Resistance is futile. Moreover, 802.11 is indeed under-hyped.
*Benchmark Capital has an investment in Tropos Networks.