Summary of Testimony
My testimony is presented in six sections (plus a conclusion), briefly summarized here.
Section I: Introduction
Microsoft has played an important role in bringing the benefits of computers to millions of people. For more than twenty years, Microsoft has invested great effort in developing innovative new technology and broadly distributing that technology to the widest possible audience. Microsoft’s efforts to develop a common platform for software development—Windows—and to promote the development of a wide range of compatible products have helped to make computers more powerful, easier to use and much less expensive. Hundreds of millions of customers, and tens of thousands of companies, have benefited from Microsoft’s efforts.
Success in the software industry requires constant innovation and risk-taking. Rapidly changing technology and business circumstances present new challenges—and opportunities—all the time.
One such challenge and opportunity, for Microsoft and the entire computer industry, has been the explosive growth of the Internet in the 1990s. Though some of our competitors hoped Microsoft would be "asleep at the switch," we have mounted a tremendous effort to reorient our entire business toward the Internet, developing a range of new operating system and applications products to help make the Internet easier to use and more useful for customers and the industry at large.
Competition between Microsoft and Netscape has been vigorous. As that competition illustrates, the software industry is healthy and vibrant. Rare is the industry in which a start-up could immediately pose a major competitive challenge to a leading firm such as Microsoft, but Netscape has. Today Microsoft faces a new range of competitive challenges presented by Netscape’s acquisition by America Online ("AOL"), which provides Internet access to about half of U.S. home users, and their strategic alliance with Sun Microsystems, which provides a range of client and server technologies, including Java platform technologies.
Tomorrow will bring new challenges. Investment in the software industry is at record levels; new competitors are entering all the time; and customers have access to new and continually improved capabilities. In particular, tens of millions of people are now using software to browse the Web that is vastly superior to the software available just two or three years ago. People who a short time ago thought they would never use a computer are now using computers for purposes they never imagined. While good for Microsoft, that is also a positive development for the entire U.S. economy.
Section II: Why the Software Industry is So Intensely Competitive
The computer industry in general and the software industry in particular is intensely competitive and subject to rapid change. There are several important reasons for this. Perhaps most importantly, the underlying platform on which all software runs—semiconductor technology—has been improving exponentially for the past 20 years, and will continue to do so for the foreseeable future. Every year microprocessors get faster, memories get larger and prices fall. Advances in semiconductor technology mean that every year more ambitious things can be done in software. Couple this with the fact that the software industry is still in its infancy—with many opportunities to improve existing products or create entirely new products—and one quickly sees that there is constant opportunity to innovate and unseat existing competitors. In fact, most existing software products will be rendered essentially obsolete in three years, if not sooner. Microsoft is keenly aware that any software publisher that rests on its laurels will rapidly be rendered obsolete.
As a result of the relentless advance in semiconductor technology and software built upon it, the computer industry is inherently unstable, unpredictable and subject to dramatic shifts, often called "inflection points" (a term made popular by Dr. Andrew Grove, the chairman of Intel). These inflection points set the computer industry apart from more traditional industries in which change is much slower and more gradual. Existing computer companies can easily stumble when faced with such dramatic change, their prior success often hindering them. These inflection points also provide a regular opportunity for new firms to enter and grow rapidly, often with products and business models that differ greatly from existing firms.
Indeed, in high technology industries such as computer software, the greatest competitive threat to existing products frequently comes not from similar products (i.e., operating systems that compete directly with Windows today), but rather from wholly new products that render entire product categories or technology approaches obsolete. For example, the greatest competitive threat to Smith Corona typewriters came not from other typewriter companies, but rather from the development in the late 1970s of computers from Wang and others that were dedicated to a new function: "word processing." And the greatest competitive threat to Wang came not from other vendors of dedicated word processors, but rather from the development of an all-new device: the personal computer, which could perform word processing functions, and much, much more.
Microsoft itself was born as the result of an important "inflection point"—from mainframe computers to less expensive, more broadly distributed "personal computers" made possible by a new technology: the microprocessor. The established computer industry leader at the time, IBM, failed to fully appreciate the significance of this important shift. That permitted start-up companies like Compaq, Dell and Gateway 2000 to become powerful players in a newly-configured computer industry.
Microsoft is always on the look-out for the next inflection point. Indeed, Microsoft has survived several inflection points over the past 15 years that could have spelled the end of its operating system business had Microsoft not innovated rapidly. In the late 1980s, microprocessors became fast enough and memory cheap enough to offer customers an easier way to interact with computers—the "point and click" graphical user interface. At the time, Microsoft’s principal operating system product, MS-DOS, did not provide such an interface. If Microsoft had not responded with Windows 3.x, customers surely would have switched to competing software. Another inflection point occurred in the early 1990s, when microprocessors and memories shifted to a new level known as "32 bit." This shift opened the door to operating system competitors such as OS/2 2.0 and UNIX, which sought to capitalize on the fact that Windows 3.x could only run a few 32-bit applications, and did so poorly. Microsoft responded, however, with a risky and unprecedented strategy—simultaneously developing two complementary 32-bit operating systems, Windows 95 and Windows NT. While difficult to implement, this proved to be an effective strategy, enabling Microsoft and its business partners to survive and, in fact, benefit from this technological shift.
Another major inflection point is underway now—the Internet, and the certainty that network bandwidths (which affects the speed at which information can be transmitted) will soon increase ten-fold and then one-hundred fold. The build out of this network infrastructure will undoubtedly create new challenges and opportunities that few recognize today, just as few could have seen the significance of the Internet as recently as six or seven years ago. Microsoft knows that if it does not continue to rapidly improve all its products to provide the Internet-related capabilities demanded by the marketplace, those products will quickly lose value and, before long, be rendered obsolete. That is why Microsoft has been working so hard to provide Internet support across its product line.
Another major inflection point is widely-recognized to be underway now: the advent of so-called "information appliances" such as "Intelligent TVs" (televisions that incorporate a computer) and small, handheld devices. IDC, a market research firm, predicts that by the year 2001 Intelligent TVs will have greater processing power than today’s personal computer. Indeed, television "set top boxes" are available today that provide more computing power than the personal computer of just a few years ago—at a small fraction of the cost. Intelligent TVs and other forms of information appliances will perform many of the functions, such as Internet access and email, performed by personal computers today.
Just as low cost and widely-distributed personal computers rendered large mainframe computers less important, the new breed of even cheaper and smaller computing devices may render personal computers less important. As with prior inflection points, Microsoft must mount a major response if it is to remain a viable supplier of operating system software. The outcome is uncertain at best.
Aside from the destabilizing effect of constant technological change, competition in the software industry is unusually intense because, unlike many traditional products, software is relatively easy for talented software engineers to develop, produce and distribute. The tools of the trade—personal computers and software development programs—are readily available. No expensive manufacturing plants are required. Once created, software can be copied once or a million times at nominal cost and distributed quickly over computer networks (such as the Internet) and in many other ways.
Software is also different from many products because it never wears out and can be easily copied illegally. Thus, a software company introducing a new product must compete not only against the products of existing competitors and new competitors that spring up each year, but also against prior versions of its own products and against illegal copies of its own products. All of this makes for a highly competitive business, promotes innovation and keeps prices of software products low for customers.
Section III: Microsoft’s Business Model
In view of the competitive circumstances described above, Microsoft has adopted a "high volume/low price" business model for operating system software. Microsoft recognized long ago that the then-fledgling personal computer industry would not thrive if it were based upon a morass of conflicting and incompatible technologies (as it was in the early 1980s). To promote the development of products that are compatible with one another, Microsoft has developed a range of operating system products with openly published interfaces and actively "evangelized" those interfaces to software and hardware developers. Microsoft has worked hard to maintain the integrity of the Windows platform, both at the level of technical interfaces and the user interface, so that people can rely upon it. Importantly, Microsoft has widely distributed its operating system software to computer manufacturers at low prices (less than 5% of the price of a typical new computer) to attain high volumes.
Microsoft’s wide distribution of Windows makes it more desirable as a platform for software development, which in turn makes it more valuable not only to software developers, but also to computer manufacturers and customers. In other words, Microsoft’s continuing efforts to innovate in Windows and its high volume/low price business model provide important benefits to all three constituencies (and others). Software publishers can build applications that take advantage of innovative platform technologies that are built into Windows and are widely available on personal computers, freeing them from the need to continually "reinvent the wheel." Computer manufacturers can build new computers that leverage Microsoft’s very substantial research and development in Windows and that enable their customers to run a wide range of compatible applications. And customers can choose from literally thousands of computer models and tens of thousands of software programs that are compatible with one another.
By making Windows widely and inexpensively available, Microsoft has marshaled the resources of literally hundreds of computer manufacturers and thousands of software publishers in investing engineering resources to develop new products that work well with Windows. The collective contributions of so many talented and varied companies largely explains the phenomenal success of the personal computer industry, which has grown by leaps and bounds since its inception in the 1980s. And Microsoft’s widespread licensing of Windows has facilitated vigorous competition among computer manufacturers and software publishers who develop compatible products, keeping prices low.
Microsoft’s successful business model stands in stark contrast to that pursued by many of its operating system competitors, such as IBM, Apple and Sun Microsystems. These competitors focus primarily upon selling their operating system products as part of relatively expensive new computer systems, necessarily limiting their sales volume, and thereby reducing incentives for other software developers to create applications that run on their operating system products. Since their sales volumes are lower and fewer companies compete with directly compatible technology, prices for products such as the Apple Macintosh or Sun workstations (and associated applications) have generally been and remain considerably higher than comparable offerings from computer manufacturers and software publishers who build upon Microsoft’s Windows operating systems. And their prices would surely be higher still if they did not face competition from the many companies that build upon Windows.
Ironically, the very thing that makes Windows valuable to computer manufacturers, software publishers and customers—the compatibility it provides across a wide range of products and Microsoft’s commitment to deliver innovative new platform technologies rapidly—is now under attack in this lawsuit. The popularity of Windows, owing entirely to Microsoft’s efforts to innovate, evangelize and license the software cheaply to promote wide distribution, is derided as "monopoly." Microsoft’s low prices for Windows are called "high"—though competitors such as Apple have generally refused to license their operating system software at any price. The large base of Windows-compatible applications that Microsoft has nurtured for years is called a "barrier to entry"—though the benefit to consumers is plain and developers are free to (and do) develop applications for other platforms.
Most importantly, Microsoft’s efforts to provide the industry with new platform capabilities in Windows that are relevant to computing in the late 1990s and necessary to keep Windows competitive is attacked as illegal tying—even though the history of operating system development from the 1960s to the present has been marked by the steady addition of new features and functionality to operating system software to enable new uses for computers and reduce costs. Any outcome to this lawsuit that would require Microsoft to allow computer manufacturers to pick and choose which parts of Windows they would like to install would destroy its most valuable feature: the compatibility across a wide range of hardware and software products made possible by the cohesive, well-defined operating system services it provides. Absent such compatibility, the development of new products for use with Windows will be slowed, costs will rise, computers built on Windows will generally be less useful, and demand for such computers will decline.
Section IV: Windows Faces Intense Competition
Windows faces immediate and impending competition from a wide variety of sources. I believe that the testimony of the plaintiffs’ witnesses on this point, particularly their economists, fails to account for the highly dynamic nature of competition in the software business, or the ways in which software innovations can render entire product categories (such as "operating systems for Intel x86 chips") obsolete very quickly.
Plaintiffs’ economists adopt a static world view, focusing narrowly upon operating system choices that computer manufacturers are likely to make tomorrow or the next day. But competition in software development is driven by the important forces that will shape the future: Will the Internet render Windows irrelevant or less valuable, as many have suggested? Will Sun Microsystems and its many allies (such as AOL/Netscape, IBM and Oracle) develop a new and more valuable platform that runs across personal computers and many other kinds of computing devices? Will the "personal computer" itself continue to meet customer needs, or will it be replaced over time with new "information appliance" devices that leverage network and server resources? As network bandwidth increases, will operating system software come to be delivered across the Internet (diminishing or even eliminating the significance of installation of operating system software by computer manufacturers)? I believe that if Microsoft viewed its business as narrowly and in such static terms as the plaintiffs’ economists do, Microsoft would have been run out of business years ago.
Even viewed narrowly, Windows faces competition today from a number of competing operating systems such as Apple’s MacOS, Be’s BeOS, the various "flavors" of the UNIX operating system and the Linux operating system distributed by companies such as Caldera and Red Hat Software and IBM’s OS/2 Warp. The Linux operating system, in particular, is widely-perceived to be a very serious emerging competitor to Windows. Although few had even heard of Linux a few years ago, today millions of people run Linux, major computer manufacturers are offering computers with Linux preinstalled, and major applications are available for Linux. Windows also faces competition from older versions of Windows, from so-called Windows "clones," and from illegal "pirated" copies of Windows.
In addition to other full-fledged operating systems, Windows faces formidable competition from "middleware" products such as Netscape’s Web browsing software and Sun’s Java suite. The purveyors of such middleware products are well-funded companies that have every opportunity to compete with Windows. I believe that one of the key contradictions of the testimony of plaintiffs’ witnesses is how they trumpet the fact that middleware products like Netscape Navigator compete with Windows, while simultaneously asserting that such middleware products somehow exist in separate "markets."
Windows also faces impending competition from so-called "network computers" (which are being promoted by powerful companies like Sun, Oracle, Netscape, and IBM) and from "information appliances" such as the set-top boxes I referred to earlier. As network bandwidth grows dramatically, as it definitely will, new generations of computing devices that customers can simply plug in anyplace to tap into the power of the Internet will likely be developed. AOL, Sun Microsystems and others are said to be developing such devices now for mass consumer distribution, and they will make very little use of existing operating system software (and probably no operating system software from Microsoft).
AOL is, far and away, the leading provider of Internet access and content. AOL’s acquisition of Netscape and its associated strategic alliance with Sun significantly increases the opportunity and likelihood that AOL will combine its strong position on the Internet with technology from Netscape and Sun to offer an alternative non-Windows consumer platform both on personal computers and television. In fact, these new computing devices (and the software in them) may well be offered to customers "free of charge"—much like cell phones—to encourage customers to sign up for more lucrative monthly service plans and to engage in recurring electronic commerce transactions. Cable companies may well adopt a similar business model, presenting a major competitive challenge to manufacturers of personal computers and their suppliers, such as Microsoft.
Section V: Microsoft’s Decision Not to Charge Separately for Internet Explorer
The plaintiffs have placed great emphasis in this lawsuit on attacking Microsoft’s business decision not to charge separately for its Internet-related improvements to Windows, calling it "predatory." But the basis for Microsoft’s decision is simple and straightforward: Microsoft believed that it had to significantly improve Windows to meet the new challenges presented by the Internet, and that the company was in no position to charge customers a separate fee for those improvements. Rather, we believed that Microsoft needed to do what it has been doing for more than fifteen years—improve its operating system products with the addition of new functionality to respond to changing customer and developer needs.
Microsoft believes that the inclusion of Internet Explorer technologies in Windows was vitally important to support applications development: not only traditional applications like word processors, but—from both a customer perspective and a technical perspective—Web pages too. I share the view of many people, both within and outside of Microsoft, that Web pages are an important new category of software "application." Any operating system platform that wishes to provide value in the future must provide strong support for developers creating Web pages, as Windows now does.
Even more fundamentally, the very same technologies that are used on the Web, such as the HTML display format for Web pages, are becoming increasingly popular in creating general purpose software applications. For example, new applications are being created today that employ HTML-based user interfaces, and many applications enable customers to create documents in the HTML format, ready for posting to the Web or any other use. Recognizing this, Microsoft designed new versions of Windows (starting with the release of OSR 2.0 in 1996) so that any software program could take advantage of the HTML support provided by the Internet Explorer technologies, thereby making Windows a better platform for software development and helping to foster the use of this (and other) Internet-related technology.
Improving Windows obviously helps Microsoft to license more copies of Windows, generating additional revenue for Microsoft. In addition, Microsoft’s own applications developers benefit from the new Internet-related platform capabilities of Windows, just as third party software developers do, spurring sales of a wide variety of Microsoft applications products. For example, Microsoft’s business productivity suite, Microsoft Office, makes use of the Internet Explorer technologies in Windows, and Microsoft has developed new Web-based applications, such as the Expedia travel service. Even a slight increase in the sales volume of Windows or Office would enable Microsoft easily to earn a handsome return on its Internet Explorer development costs. (Of course, Microsoft’s decision to provide new capabilities in Windows in no way prevents other software publishers from developing their own software programs to run on Windows and provide similar capabilities, as frequently happens.)
Microsoft’s decision not to charge separately for Internet Explorer technologies is also consistent with the actions of its competitors. All major operating system vendors include Web browsing software with their operating system products at no extra charge. Even Web browsing products that were designed as an application, rather than an operating system upgrade, have historically been offered without charge—first by Mosaic, then by Netscape, and now by companies such as Encompass and Bigfoot. Companies that develop Web browsing software earn revenue, as Netscape has done, by selling related software such as Web servers, by earning recurring revenue from services (as AOL has done) or by generating advertising revenue by driving customer "traffic" to particular Web sites. Many businesses, such as network television and cell phones, follow this model, a model to which software is particularly well suited because the marginal cost to produce and distribute a software program is very, very low.
Section VI: Microsoft’s Relationship with Other Companies
Section VI describes certain aspects of Microsoft’s relationship with three specific companies: Intel, Apple and Netscape. I provide this testimony to respond to some of the specific allegations made by plaintiffs’ witnesses. In all three cases, I believe that the plaintiffs have attempted to portray instances of routine interaction between companies in the computer industry, who inevitably must deal with one another, as something nefarious.
For many years Microsoft has worked closely with Intel to develop a range of new technologies. Such cooperation continues today. From time to time, disagreements may arise, as one would expect. In the summer of 1995, for example, Microsoft believed that Intel’s NSP software threatened to hinder, rather than promote, the advance of the personal computer architecture. As Microsoft explained to Intel through technical briefing papers and in meetings, NSP was not compatible with Windows 95. That was important because Intel and Microsoft both hoped Windows 95, which was about to be released, would spur sales of new computers (and new Intel microprocessors). Microsoft worked cooperatively with Intel, however, so that many of the capabilities envisioned for NSP were provided to the industry in ways that are compatible with Microsoft’s operating systems.
Intel’s Steven McGeady testified that Microsoft threatened to delay support for Intel’s new MMX instruction set in 1995. In this case, Microsoft had some concerns about the extent to which other microprocessor vendors and their customers would be able to develop compatible MMX implementations, and we expressed those concerns to Intel. Mr. McGeady’s suggestion that any Microsoft concern about MMX could pose a "threat" to the new instruction set is greatly exaggerated, at best, because Intel designed MMX so that it does not require support in Windows or other operating systems in order to function. In any event, Microsoft has fully supported MMX in its software products, and MMX is a great success today.
The plaintiffs have tried to portray the broad set of agreements entered into between Apple and Microsoft in August 1997 as relating primarily to distribution of Internet Explorer. This is simply untrue. In those agreements, which plainly benefited both companies and customers who use Apple computers, Microsoft and Apple granted broad patent cross-licenses to each other and Microsoft agreed to keep developing innovative new versions of the Macintosh version of Office for five years. From Microsoft’s perspective, the most important part of the August 1997 agreements by far was the patent protection it received from Apple, because Apple has a history of aggressively pursuing intellectual property claims against Microsoft, even though Microsoft is a key developer for the Apple Macintosh platform. Internet Explorer was a relatively unimportant part of the transaction.
Finally, I respond to Jim Barksdale’s allegations that Microsoft did not provide good support for Netscape’s software development efforts on Windows. In fact, Microsoft’s record of support for Netscape has been exemplary, and Netscape’s Web browsing software for Windows 95 was a huge commercial success.
I.
INTRODUCTION
For more than twenty years Microsoft has innovated rapidly and competed vigorously to bring the benefits of computing to millions of people. This section of my testimony provides a very brief overview of these efforts, including Microsoft’s efforts to develop software products that are designed to work well with the Internet.
A. Microsoft’s Vision and Role in Enabling Millions to Use Computers
Microsoft was formed in 1975 as a partnership between then nineteen- year-old Bill Gates and his friend Paul Allen. Microsoft’s goal, then as now, was to develop great microprocessor software (hence the name). Microsoft’s first product was a version of the BASIC programming language for a small computing device called the Altair. The Altair didn’t do much. With the addition of Microsoft’s software, and a little luck, a hobbyist might get the Altair to flash a couple of lights. Few would have thought that anyone could actually make a business out of developing software for such rudimentary devices. At the time, the business of computing was synonymous with one company: IBM. A computer generally meant a "mainframe," a large machine costing $1 million or more, maintained behind glass walls. Few had access to such machines, and few thought about the software that ran on them.
Bill Gates and Paul Allen had a different vision. They believed that rapid advances in a new invention, the microprocessor, combined with innovative software, would make possible new generations of computing devices that would bring the benefits of computing to millions of individuals. From the beginning, Microsoft directed its software development efforts toward the goal of making computers useful enough that there would be "a computer on every desktop and in every home"—certainly an ambitious goal. Microsoft knew, however, that no company, acting alone, could hope to develop all the necessary technology. Nor could any one company make all the necessary technology available broadly—and inexpensively—throughout society. This would require the combined efforts of an entire industry.
In the early 1980s, Microsoft adopted a business and technology model intended to foster development of the nascent "personal computer" industry. Microsoft recognized that the growth of the industry would be greatly accelerated if common operating systems were developed that would provide good platforms for software development and that were widely licensed to computer manufacturers. In this way, the engineering talent of dozens or even hundreds of hardware and software companies could be leveraged to contribute to the development of a wide range of compatible personal computer products. By constantly improving its operating system products, Microsoft would create new opportunity for software developers and help to make personal computers more functional.
Today, thanks in part to the common platforms provided by Microsoft’s operating systems, customers can choose from among literally thousands of computer models and tens of thousands of software applications that are compatible with one another. A huge number of companies employing hundreds of thousands of people contribute technology to make this phenomenon possible.
A lot has changed since the 1980s. One thing that has not changed is Microsoft’s intense focus on making computers easier to use, more powerful, more widely available and less expensive. This requires innovation and risk taking in the face of constantly changing circumstances—new technologies, new competitors and new customer demands. It also requires a never-ending effort to discern and satisfy the needs of software and hardware developers that are also racing to exploit new opportunities.
Because software is ultimately just "ideas," Microsoft must recruit and retain the "best and the brightest" technical talent and develop a management team with a deep understanding of software product development. Microsoft also must maintain the internal discipline needed actually to deliver large, complex pieces of software created by hundreds of software developers. In addition, Microsoft must constantly work closely with other companies in the industry to encourage the development of complementary software and hardware products and generate customer interest and excitement. Finally, Microsoft must invest for the long term. This is the real foundation of Microsoft’s success.
B. The Advent of the Internet
In the mid-1990s, customer and developer attention turned to the Internet, a vast network of widely separated computers. The Internet arose from research funded by the federal government that began in the 1960s. By 1990, the Internet was in widespread use in the academic community all over the world. Many researchers experimented with applying a well-understood computer concept ("hypertext") for linking pages of information together via the Internet.
In particular, Tim Berners-Lee, working at a physics research facility in Switzerland called CERN, developed the concept of the "World Wide Web" and the associated software used with it, called a "browser." He is acknowledged as the creator of the Web. Mr. Berners-Lee writes in his "The World Wide Web: A very short personal history":
One of the things computers have not done for an organization is to be able to store random associations between disparate things, although this is something the brain has always done relatively well. In 1980 I played with programs to store information with random links, and in 1989, while working at the European Particle Physics Laboratory, I proposed that a global hypertext space be created in which any network-accessible information could be referred to by a single "Universal Document Identifier". Given the go-ahead to experiment by my boss, Mike Sendall, I wrote in 1990 a program called "WorlDwidEweb", a point and click hypertext editor which ran on the "NeXT" machine. This, together with the first Web server, I released to the High Energy Physics community at first, and to the hypertext and NeXT communities in the summer of 1991.
(See DX 2243.)
Mr. Berners-Lee has since been awarded a MacArthur Foundation Genius Grant and a Distinguished Fellowship of the British Computer Society for his innovations. Much of the software he wrote, such as LibWWW, was later integrated into browser software by Netscape, Microsoft and others. He continues to drive the Web and the broader Internet forward through his work in the W3C Consortium. Neither Netscape nor Marc Andreessen and his colleagues at the National Center for Supercomputing Applications at the University of Illinois (NCSA) invented the Internet, the World Wide Web, or the concept of browsers that use hypertext navigation to traverse the Web.
By the mid-1990s, a combination of factors led to the explosive growth of the Internet’s popularity: (i) the speed of computer modems had increased while their cost had decreased; (ii) improved, high speed data switches called routers became much cheaper; (iii) fiber optics had dramatically improved the cost and capacity of long distance communications; and (iv) millions of computer users now had a graphical environment (Windows) on their computers. In view of these factors, many companies recognized that conditions were ripe for the Internet and its uses, including e-mail and Web browsing, to spread from the academic world to the general populace.
Dozens of companies, including Netscape, sprang up in 1994 and 1995 to develop new technologies that would make the Internet accessible first to millions, then tens of millions, of people. Hundreds of existing companies, including Microsoft, Apple, IBM, Sun Microsystems, Novell, Oracle, SAP and Intuit began to add Internet capabilities to their products.
C. Microsoft’s Response to the Internet
Bill Gates and others at Microsoft called the Internet phenomenon a "sea change" and, later, a "tidal wave." For Microsoft, the Internet was both a challenge and an opportunity, a giant step on the way to a vision the company had long held—that computers would enable users to access information anywhere, anytime, quickly and easily. Since 1990, Microsoft had promoted that vision as "Information at Your Fingertips." At the same time, the Internet created the very real possibility that Microsoft and its products would quickly be rendered irrelevant if Microsoft did not rapidly enhance its software products to promote and take advantage of the benefits of the Internet.
We could have been complacent, letting the Internet tidal wave crash down upon us. Or we could seek to ride the wave. Some in the industry hoped that Microsoft would fail to appreciate and respond to the significance of the Internet—that just as the mainframe and minicomputer industry had inadequately responded to the phenomenon of the personal computer, Microsoft too would fail to innovate, leaving the field clear for others.
Microsoft knew it had no choice but to try to ride the Internet wave—indeed, to reach for the forefront of the wave. As a result, although we already had our hands full developing Windows 95 and Windows NT, Microsoft set out to develop a range of new software that would contribute to the development and widespread use of the Internet.
There was much to do. Simply connecting a personal computer to the Internet was a challenge for most people. The types of information that could be presented to customers over the Internet was limited to static text and graphics; security was rudimentary; and publishing information across the Internet was difficult—just to name a few challenges.
Every group in Microsoft was charged with focusing on how to make its products work well with the Internet. The first step was providing strong support for existing Internet standards in all our products. We also worked hard to present the Internet-related capabilities of our products in a way that was easy-to-use and attractive for our customers.
Thanks to the built-in Internet support in Windows, developers today can quickly and efficiently create great software products that exploit all facets of the Internet. Customers can access information on the Internet just as easily as they access information located on their own computer or on a network at the office. Customers have the best features of the Internet and the best features of the Windows environment carefully designed together for a simpler and more valuable experience.
We also have enhanced our software development tools with strong support for the Internet, developed a range of server software for setting up and running Web sites and revamped our business productivity software, Microsoft Office, to make it easy to view and create information in Internet standard formats like HTML. We have heavily promoted the value of developing corporate information systems that take advantage of the ubiquity of the Internet, hoping to spur new rounds of investment in hardware and software products.
D. Vigorous Competition Between Microsoft and Netscape
Netscape did a great job of associating itself with the Internet wave—and contributing to the wave by promoting the benefits of the Internet and delivering Web browsing technology. I do not believe, however, that Netscape should have expected that the 80%-90% share of browser usage that it obtained almost overnight in late 1994 would last forever. The Internet was too important a factor in the computer industry for Netscape to have the field to itself for very long. Other companies such as Microsoft, Sun Microsystems, IBM, Real Networks and many others had their own ideas on how to take advantage of the connectivity that the Internet provides, and they have been pursuing those ideas vigorously.
Netscape nevertheless set its sights very high, declaring publicly and repeatedly in 1995 and 1996 that its products would render not only Windows, but also leading groupware products such as IBM’s Lotus Notes, obsolete. Netscape said it would rapidly integrate new features and functionality into its Web browsing software to support a wide range of customer activities and transform it into a platform for software development. Netscape in fact integrated features from previously separate applications, such as e-mail and calendaring, into its Web browsing software, and courted software developers to write for the new Netscape platform. In an October 1996 Netscape press release, Marc Andreessen said:
In 1995, Netscape led the first wave of the Internet with the explosion of the Web. In 1996, we led the second wave as Web technology penetrated the corporation with rapid adoption of Intranets. . . . In 1997, Netscape will once again lead the industry with Internet technology that combines the richness of Web content with enterprise applications, producing the next killer application—Web-rich email and groupware.
(See DX 2244.)
These announcements were not lost on Wall Street securities analysts, who took note of Netscape’s strategy to replace the Windows platform with a Netscape platform in their research reports, and believed that such an outcome was a distinct possibility. On the basis of those reports, the price of Netscape stock soared, providing Netscape with financial strength to compete with Microsoft.
Of course, Microsoft was already in the platform business, and we too saw the benefits of providing Internet capabilities in platform products. Indeed, the point is obvious: if computing is turning to the Internet, then developers will need Internet support in platform products. This is why Microsoft "exposed" many functions performed by the Web browsing capabilities in Windows via "application programming interfaces" (APIs) so that third party software developers could use those capabilities in their own products, which many developers are doing today. Netscape has not kept up with the fast pace of Microsoft’s platform development efforts, and today its Web browsing software provides only a portion of the platform functionality provided by the Internet Explorer components in Windows.
The upshot of these events has been spirited competition between Netscape and Microsoft. Tens of millions of people are on the Web today, using Web browsing software that is much better than was offered just a few years ago. Web browsing technology is also widely available, having been built seamlessly not only into operating systems, but also into applications of many kinds, as well as entirely new types of computing devices such as Internet-enabled televisions and telephones. Like many other technologies, consumers usually need not pay separately for Web browsing software because it is just a component of other products (such as a personal computer, Internet-enabled television, or a handheld device) or is supported by revenues from associated software (such as server software) or services (such as advertising). From the consumer perspective, it is hard to imagine that any technology has offered better price-performance gains than Web browsing software has over the past few years.
I believe that Microsoft’s efforts have contributed significantly to the growing popularity of the Internet. By providing enabling technologies, we have helped unleash new rounds of innovation, spurred greater investment by the industry in the infrastructure needed to support better (including higher-bandwidth) connections to the Internet and enabled the rapid development of the Internet as a truly useful medium for exchanging information and conducting business.
I am proud of all the innovation and plain hard work represented by Microsoft’s development of literally hundreds of new software products over the past fifteen years. This hard work has helped to enable Windows, and the many, many firms and customers who benefit from it, to meet the opportunities and challenges that the rapid pace of technological change has imposed.
Although some of plaintiffs’ witnesses have sought to characterize Windows as a "gateway" or a "chokepoint," the success of Windows has always depended on just the opposite being true. Microsoft broke with industry tradition by licensing its operating system software broadly and cheaply, separate from expensive hardware (unlike the approach of Apple and others). As a result, any computer manufacturer can leverage Microsoft’s investment in Windows to build personal computers that are compatible with products from hundreds of other manufacturers. Any software developer can create applications that run on Windows based on widely-published information, and literally millions have done so, leading to sales of billions of dollars of software that runs on Windows. And thanks to Microsoft’s enhancements to Windows, anyone can use Windows to connect to any of the millions of sites on the Internet—located around the corner or around the world—with ease.
E. The Software Industry Is Working Well
Innovation and integration—by many companies, in software and in personal computer hardware—is what gave us personal computers that are as powerful as the mainframes of a generation ago, and that year by year deliver more functionality for the same or lower price to the consumer, enabling millions of people of all ages and walks of life to do things that they never imagined.
Given the nature of the software industry, with its continuous change and almost limitless opportunity for product improvement, we can expect the rate of innovation to continue at its rapid pace, especially in the area of platform software. Application vendors, knowing that there is often an advantage in establishing early leadership on a platform, will switch very rapidly to any new platform that appears to offer benefits over existing platforms. If a new platform is closely associated with a technology "inflection point," it can gather support so quickly that it "runs away" from existing platforms, and support for older platforms falls off rapidly.
The investment community is keenly aware of the large business opportunities associated with developing popular platforms, and thus there is no shortage of investment funds for developers to bet on new platforms. Netscape is a great example of this. In 1995, Netscape had no difficulty tapping hundreds of millions dollars from the capital markets to compete with Microsoft. Another example is the large investment being provided for software developers to write for the Sun Java platform. Sun and its partners (such as the Kleiner Perkins venture capital firm) have established a $100 million "Java Fund" to fund such developers. They are betting on the establishment and run-away success of a new platform, running across mainframes, personal computers, network computers, and new generations of "intelligent appliances," that could rival or effectively eliminate Windows.
Indeed, the past three years have seen record levels of investment by the venture capital community in software companies, and soaring stock market valuations based on investor expectations concerning the business opportunities that lie ahead. In addition to creating enormous amounts of wealth, the increase in stock prices has allowed these companies to finance further technological investments, pursue acquisitions, and speed the pace of technological development. In short, the software industry has enjoyed virtually unprecedented economic success. John Doerr, a senior partner at Kleiner Perkins and a member of the Netscape Board of Directors, has called this era "the greatest legal creation of wealth this planet has ever seen." (Institutional Investor, June 1996.)
Consumers have benefited from the rapid improvements in and broad distribution of computer technology. Not only is the personal computer of today dramatically improved in every dimension compared to ten years ago, but literally hundreds of millions of people are using computers in ways they never would have imagined. Simply stated, plaintiffs seek to "fix" an industry that is not broken. I believe it would set a very bad precedent for the high-technology industry if a rule were established, ostensibly with the objective of promoting competition, that actually impeded efforts to compete through product improvement.
II.
WHY THE SOFTWARE INDUSTRY IS SO INTENSELY COMPETITIVE
The software industry, and the broader computer industry of which it is part, is intensely competitive. Rapid technological change constantly creates new business opportunities and threatens existing products and business models. Rivals recognize the huge potential rewards of successfully identifying new technological trends and delivering products that capitalize on those trends. The software industry is characterized by constant ferment, as competitors new and old jockey for position with new ideas, new technology initiatives, new business strategies and ever-changing business alliances.
The intensity of competition in the software industry has been a fact of life since the advent of the personal computer in the late 1970s. When Microsoft got its start the computer industry was led by IBM, known for its mainframe computers, Digital Equipment Corporation, known for its minicomputers, and their rivals such as Burroughs, Sperry Rand, Data General and Wang. These companies had revenues and other resources that vastly exceeded those of Microsoft or any personal computer company. By contrast, most of the leaders in the industry today—such as Microsoft, Compaq, Dell, Sun Microsystems, Novell and Oracle—did not even exist fifteen or twenty years ago, or were in their infancy.
A range of factors combine to make competition in the software industry intense and fluid. Microsoft grapples with these realities of the software industry every day.
These factors are important to understand because they mean competition is inherently dynamic. Yet the testimony of the plaintiffs’ witnesses treats the world in which Microsoft competes as largely static. For example, plaintiffs limit their competitive analysis to products that are directly substitutable to Windows today. While a range of direct alternatives to Windows (discussed in Section III of my testimony below) in fact exist and are in use, plaintiffs overlook the critically important facts that (a) existing software products will quickly be rendered obsolete if they are not continually improved and (b) even with steady improvements, software products that are successful one day may quickly be displaced by entirely different technologies that provide greater value for customers. I believe that any company that fails to appreciate and respond to the dynamic nature of competition in the software industry is unlikely to succeed.
In Sections II.A through II.F below, I identify and briefly describe the factors that make competition in the software industry so dynamic.
A. Rapid Technological Change and Inflection Points
A principal reason for the unrelenting change that drives competition in the software industry is the fact that semiconductor technology, the underlying engine that creates the opportunity for more capable software, continues to progress rapidly. The semiconductor revolution, which began in the late 1960s and has gathered force since then, ensures that every year more ambitious things can be done in software. In other words, the software industry can never be stable because the platform on which software runs—the microprocessor—is always becoming more capable.
Gordon Moore, one of the co-founders of Intel, is well known for having postulated in 1965 that microprocessor capabilities would double every 18 to 24 months—an extraordinary rate of growth. For example, in 26 years the number of transistors on a microprocessor chip has increased more than 3,200 times, from 2,300 on the Intel 4004 microprocessor in 1971 to 7.5 million on the Pentium II microprocessor of today. In addition, the speed at which the chips operate has increased by a factor of over a thousand-fold. As a result, a typical personal computer now provides users with more raw computing power than an IBM mainframe of a generation ago, and the software that runs on a personal computer differs completely from the software that used to run on a mainframe. Moore’s Law has held true for over 20 years, and there is no let-up in sight.
The rapid rate of growth in microprocessor power makes possible—and creates—market demand for a steady succession of dramatic improvements in software. Capabilities that are totally unfeasible one year can become commonplace just a few years later. Software vendors must continually innovate to support new uses and functions, such as graphical user interfaces, multi-tasking, networking and multimedia. Any software vendor that fails to keep pace with the demand for such new functions will soon be left with obsolete and thus nearly worthless products. Microsoft knows that all of its existing products will probably be rendered obsolete within three years.
In fact, constant improvement in the underlying microprocessor technology regularly creates changes that alter the entire competitive landscape. These changes can rapidly diminish or even eliminate the value of existing products. The phenomenon is often referred to as an "inflection point."
Inflection points are destabilizing because they occur when some set of technological factors accumulate to the point that they suddenly allow the broad introduction of new capabilities with high value to consumers. In the case of major inflection points, users rapidly switch to new products with the new capabilities and are willing to leave behind advantages of older products, including compatibility with complementary products (such as phonograph records played on turntables, or software applications that ran on operating systems like CP/M, an early, but now obsolete, leader in operating systems for Intel microprocessors).
Such destabilizing change is very difficult for companies with existing products to deal with. Past success often induces complacency. Prudent managers become reluctant to risk investing in nascent technologies, fearing that they may turn out to represent dead ends, rather than inflection points. Further, even in well-managed companies that foresee the urgency, resources are often tied up responding to legitimate user requests for improvements to older products.
The difficulties that established firms face in responding to technology inflection points are described in Dr. Grove’s book Only the Paranoid Survive: How to Exploit the Crisis Points That Challenge Every Company (1996) and in Clayton M. Christensen’s book The Innovator’s Dilemma: When New Technologies Cause Great Firms to Fail (1997), books that I read with interest in the normal course of my work at Microsoft.
Fear and anticipation of the next inflection point runs deep at Microsoft because Microsoft is both the beneficiary of, and a contributor to, the inflection point that marked the shift from expensive mainframes and minicomputers to low cost personal computers. None of the then-existing mainframe and minicomputer companies weathered this shift well.
For example, in the early 1980s, industry visionary Ken Olsen, who founded Digital Equipment Corporation (maker of the famed line of VAX minicomputers), is said to have remarked that there would never be any real use for computers in the home. Indeed, although Microsoft was already working toward its vision of "a computer on every desktop and in every home," many industry observers regarded personal computers to be little more than toys. Digital Equipment has since been acquired by Compaq, a company that built its business on personal computers running Microsoft operating systems.
Microsoft was challenged by several important inflection points in the 1980s and 1990s, each of which could have devastated its operating system business. In the 1980s, Microsoft’s operating system revenues derived almost entirely from MS-DOS, an operating system that required customers to learn and type arcane commands. If Microsoft had failed to develop a graphical user interface, its operating system business inevitably would have been eliminated by competitive forces such as the Apple Macintosh.
Microsoft had to make three expensive attempts (Windows 1.0, Windows 2.0 and OS/2 version 1.21, jointly developed with IBM) at developing an operating system with a graphical user interface before achieving success with Windows 3.0 in 1990. Had competitors such as Apple, IBM and the various publishers of UNIX-based operating systems made different product and marketing decisions in the late 1980s, they, rather than Microsoft, would likely be the operating system leader today.
Microsoft next had to make the transition from the "16-bit" generation of software such as Microsoft’s Windows 3.0 and 3.1 to new "32-bit" software made possible by a new microprocessor architecture from Intel. Existing "16-bit" software had limited capability to run large programs or to run multiple programs at the same time. Such 16-bit software also had limited robustness, security, server capability and portability to other microprocessor architectures. Once 32-bit microprocessors became ubiquitous and cheap, failure to respond and offer operating systems that took advantage of such new microprocessor features would almost certainly have spelled the end of Microsoft’s operating system business.
Microsoft was thus faced with the simultaneous challenges of developing the necessary operating system innovations to exploit the 32-bit world, providing a smooth migration path to the 32-bit world, and improving things for customers in doing so. Rather than shrink from the challenge, Microsoft elected to view this as a decisive opportunity, and responded by simultaneously developing not one, but two complementary 32-bit operating systems—Windows 95 and Windows NT. These new operating systems would share certain innovations, such as a new user interface and a new set of APIs that would enable developers to create applications that would run on both of them. In other respects, however, these new operating systems would be radically different. Windows 95 was designed with greater emphasis upon providing compatibility with existing Windows (and MS-DOS) software and hardware. Windows NT was designed with an emphasis upon laying down new technical foundations for the future.
Windows NT had to be written completely from the ground up, effectively rendering obsolete every line of operating system software that Microsoft had written to date. This was the only way to ensure that the new operating system would be secure, robust and portable to a variety of microprocessor architectures (not just Intel x86), and that it would offer high-performance client as well as server functions. Windows 95 also had to be written largely from the ground up to take advantage of 32-bit microprocessors while continuing to work well with a broad range of older 16-bit software. Microsoft knew that this major investment of effort was absolutely essential to stay competitive. Yet because so much of this software was new and unproven, there was no guarantee of success.
Microsoft’s simultaneous development of these operating systems enabled it to satisfy a wide variety of user needs. None of Microsoft’s competitors attempted such an ambitious development effort, even though other firms (such as IBM and Sun) clearly had the resources to do so. It is this type of radical innovation and willingness to invest for the long term that allowed Windows to leverage the "graphical user interface" and "32-bit" inflection points and continue to compete effectively.
We were not able to rest, however. In the mid 1990s, a new inflection point appeared, arising from the rapid advance of digital communications technology, which found form as the "Internet."
The Internet revolution was, and will continue to be, as far-reaching and profound as the semiconductor revolution that gave rise to the microprocessor. In a May 1995 memorandum, Bill Gates called this inflection point "The Internet Tidal Wave." He began the memorandum this way:
Our vision for the last 20 years can be summarized in a succinct way. We saw that exponential improvements in computer capabilities would make great software quite valuable. Our response was to build an organization to deliver the best software products. In the next 20 years the improvement in computer power will be outpaced by the exponential improvements in communications networks. The combination of these elements will have a fundamental impact on work, learning and play.
. . .
. . . Perhaps you have already seen memos from me or others here about the importance of the Internet. I have gone through several stages of increasing my views of its importance. Now I assign the Internet the highest level of importance. In this memo I want to make clear that our focus on the Internet is critical to every part of our business.
(GX 20.) I have no doubt that Microsoft would be failing today if it had not rallied to meet the challenge of the Internet (which is ongoing) with the same kind of commitment and innovation that Microsoft mustered in response to earlier inflection points.
The speed with which inflection points can unseat established leaders is illustrated by the reaction of many in the software industry to Microsoft’s position in Internet software in 1995. Many believed that Microsoft had missed an all-important inflection point, and thus was condemned to the same sort of long-term decline that the mainframe industry has suffered. By the summer of 1995, competitors of Microsoft such as Netscape and Sun were openly declaring that Microsoft was washed up and that Windows was little more than "legacy" code (a term of derision in the software industry that means "obsolete"). Indeed, many industry observers began predicting the demise of Microsoft, saying that Microsoft "didn’t get" the Internet. Rick Sherlund, a leading financial analyst for the software industry, downgraded his rating of Microsoft stock for the first time ever on the stated grounds that the Internet constituted an inflection point that would devalue Windows. Later, a banner headline across the top of PC Week, a leading industry publication, summed up the situation well: "Windows Everywhere or Nowhere." (PC Week, November 18, 1996.)
Rather than settle into a period of decline, Microsoft worked hard to incorporate support for the Internet into all of our important products, including Windows.
Microsoft’s vision for the Internet went beyond providing basic and high quality support for existing uses. Cheap and universal connectivity provided by the Internet opened up new and exciting opportunities. Microsoft has been a leader in promoting the use of the Internet for audio and video conferencing (NetMeeting and Netshow), for conducting commerce and collaboration (Site Server), for enabling non-technical people to set up their own Web sites (FrontPage) and for enabling people to make use of inexpensive televisions to gain access to the Internet (WebTV).
In addition to developing technologies to support these emerging ways to use the Web, Microsoft set about the difficult task of taking the best of what customers liked about Windows and the best of what customers like about Web pages, and combining them into a new synthesis—one that would make access to information, of whatever type, wherever located, easy and consistent.
Just as the marketplace voted with its feet in the rapid transition to the graphical user interface (despite resistance from some), we believe the marketplace is in the midst of a transition to an emerging "Web-oriented Interface" (based upon "links" and use of "back" and "forward" buttons, among other things). Microsoft was one of the first to see the importance of developing software that would enable customers to interact with information and applications on their computers the same way they do with information and applications on the Web. The improvements to the Windows 98 user interface are a step in this direction, although we have a long way to go to further simplify and unify user interface software. I believe that any software publisher who ignores technology and customer demand shifts like these does so at its peril.
As with previous inflection points, these efforts have involved significant investments, disruption to existing development efforts and lots of risk with no assurance of success.
The Internet inflection point is not over yet, for exploitation of global connectivity by software is just beginning. We are competing in the second inning of a long game. We nevertheless are pleased that our efforts are starting to bear fruit.
Even as we are still grappling with the implications of the Internet, we are witnessing the emergence of a new inflection point. In particular, we are seeing the emergence of new classes of microprocessor-based computing devices that are different from existing personal computers—devices such as intelligent televisions (or "set-top boxes"), hand-held computers and game consoles. These devices, often called "information appliances," are rapidly becoming cheap and powerful enough to participate in the broader flow of information between people and to perform a wide range of computing functions. Moreover, the advent of information appliances is being reinforced by, and in turn will reinforce, the semiconductor and Internet revolutions.
Already, information appliances are providing consumers with alternatives to the communications capabilities and software applications provided by personal computers. For example, within the next few years, televisions will be turned into powerful computing platforms. For software vendors and computer manufacturers, the emergence of information appliances presents yet another inflection point—a major opportunity to innovate or be left behind.
B. Significant Room for Improvements in Software Products
A second important reason why the software industry is so competitive is that it is still in its infancy. Opportunity exists to improve products in almost every dimension. Despite enormous advances, today’s software products are too complicated, too difficult to learn how to use and too hard to fix when things go wrong. This creates a big business opportunity to apply computing resources to make things simpler, without making them simplistic. Integration is one of the primary ways to achieve this goal. If different aspects of software can be made to behave in the same way and follow the same conventions, there is less for the user to learn, and more opportunities to apply existing knowledge.
There are also whole frontiers for software that have yet to be crossed. For example, speech is an obvious way for people to interact with a computer, but the ability of software today to deal in a meaningful way with speech is very limited. Enabling the populace at large to interact usefully with computers via speech will require software that not only can discern different sounds, but also understands the meaning of words. An opportunity also exists to make software better able to anticipate a customer’s needs and to support groups of people working together, sharing their learning and knowledge. There is an opportunity to make software that is able to understand other types of information, such as photographs, and to help customers organize and find such information.
Other examples of the limitations and deficiencies of today’s software abound. The relentless advance of semiconductor technology continually opens new possibilities to apply techniques that would have been too slow or expensive just a short time before. This means that there will be major opportunities in the coming years for companies that can supply simpler, higher quality software.
C. Ease of Creating Software Products
Unlike many traditional industries, which require heavy investments in capital infrastructure or access to natural resources, software is based entirely on ideas that are implemented as written instructions (software code) for execution by a microprocessor.
Like other forms of written expression, software can be created by anyone skilled in the art—and literally millions of people are skilled in the art. Since the late 1970s, computer programming has been widely taught in academic institutions. Governments fund the development of software through academic and research grants, which provide a rich source of innovation and new ideas. Indeed, the precursor of the Internet, called DARPANET, was funded by the Defense Advanced Research Projects Agency, a unit of the U.S. Department of the Navy. Tim Berners-Lee’s creation of the World Wide Web was financed by government money for physics, and Marc Andreessen’s first Web browser (Mosaic) was funded by government research at NCSA.
Because the basic tools of the trade (personal computers) are relatively cheap and widely available and the Internet now spans the globe, companies developing software products have access to talent all over the world. Geographically separated people can now cooperate as never before. The emergence of communities of people who have never met each other and yet work as part of broader teams has led to the development of such sophisticated, public domain software products as Linux and Apache, which are discussed below.
These factors do not mean that it is a simple undertaking to develop a full-featured operating system such as Windows. They do mean, however, that software products can often be developed quickly and relatively cheaply, and such products may reduce the value of Windows and possibly displace it over time—particularly if they exploit a technology inflection point.
Netscape Navigator is a perfect example of the ease with which software products can be developed that can pose a competitive challenge to products that have been under development for years. Version 1.0 of Netscape’s Web browsing software was commercially released in December 1994, just eight months after Netscape was founded. Based largely on the success of that product, Netscape completed one of the most successful initial public offerings in history in August 1995, and soon began to position its Web browsing software as a competitor to Windows. Indeed, Microsoft and others recognized that with further work Netscape’s simple program could in fact greatly reduce the value of Windows, if Windows were not improved to respond to the new uses and burgeoning market demand created by the Internet.
Another example of a small company creating a major and important software product was the creation of the popular Lotus Notes program by Iris Associates. A group of less than thirty programmers created the early versions of this product, for which IBM later effectively paid approximately $3 billion in connection with its acquisition of Lotus. Yet another example is Linux, an emerging competitor to Windows originally created by a student in Finland with volunteer help.
D. Ease of Producing and Distributing Software Products
Unlike products in many traditional industries, software products can be produced and distributed in vast quantities very rapidly. Once a software product is created, the cost to copy it is near zero, and the product can be quickly distributed around the block or around the world via the Internet or other networks.
Once again, Netscape Navigator is an early example of this phenomenon. Netscape’s officials were clear from the outset that the company’s ability to distribute its products via the Internet was a big advantage not previously enjoyed by most start-ups. As James Barksdale, Netscape’s Chief Executive Officer, said:
You can download browsers at the speed of light and you can change them at the speed of light. It’s not a big deal.
. . .
The fact that I now have a distribution channel to millions of clients who can download my software with one-button access, that’s a great advantage for a very small company. We can distribute globally at the blink of an eye. This is a huge advantage that has not appeared on this planet before.
(Wired, March 1996.) Mr. Barksdale has reiterated the great distribution enjoyed by Netscape many times. Less than a year ago Mr. Barksdale said that "[t]he story of Netscape is also verification of the power of the Internet as a marketing and distribution agent." (See DX 2186 at 6.)
Sun’s Java software is another good example of the ease with which software can be distributed over the Internet and in other ways. Java technology was commercially released in May 1995, just three and a half years ago, yet Java technology now is installed on at least tens of millions of computer desktops. In short, there are no barriers to distribution of software in large quantities over relatively short periods of time through an existing infrastructure.
The Internet will become an even more significant form of software distribution over the next few years, as electronic distribution becomes faster, easier, and more efficient. "Bandwidth"—the rate at which information can be transmitted via a telephone line, television cable or other "wires"—is steadily increasing. To start, the speed of modems has increased over the last 18 months by nearly 100% from "28.8 Kbps" speed to "56 Kbps," significantly reducing download times. Moreover, there are major impending changes in transmission technology. The United States cable television industry is in the process of a massive change from analog to digital transmission. At the same time, telephone companies are introducing a technology called "DSL," which will allow very high speed digital transmission of information over existing telephone lines without disrupting existing voice transmissions. Both digital cable and DSL are already available as a residential service in some cities, and both services are being rapidly deployed in other cities. These two competing technologies will mean that the majority of homes and businesses in the United States will in short order be able to download software at a rate of millions of bits per second—a greater than ten-fold increase over today.
Greatly increasing bandwidth will make it easier and more convenient to download additional software. More fundamentally, Microsoft expects that increasing bandwidth will change the very nature of how software is obtained by customers in the first place. Customers may purchase a personal computer with very little software preinstalled on it, take the computer home, plug it into a telephone line or cable TV outlet and have the bulk of the very latest operating system and application software downloaded onto it after purchase—and then updated frequently and automatically. This scenario is not a question of "if," but only of "when." When it comes to pass we may find that the company that supplies the network connector—the phone company, cable company or perhaps AOL, is in a strong position to determine which software is available to consumers.
E. Low Marginal Cost of Software Products and Software Piracy
The ease of creating, reproducing, and distributing software means that there is a great deal of free or low-cost software available, which can have an obvious effect on price competition. Sometimes software is free because vendors expect to make money on related software or services. Sometimes software is "free" in the sense that it is a feature of a broader software or hardware product. For example, most Web browsing software, graphics and window management software, user interface (or "shell") software, memory management software, client-side local area networking software and file server software are "free" for both of these reasons.
Sometimes software is free because programmers love to program, and delight in developing software that is widely used. For example, the most widely used software for running Web sites is not from Netscape or Microsoft. Rather, it is free software called Apache, developed and maintained by programmers working together via the Internet. Despite this fact, Apache’s Web server software is not just a hobbyist’s project; it is used by major corporations and supported by IBM and other leading software suppliers. Another example that recently has been getting a great deal of attention (and support from applications developers) is the Linux operating system created through the collaborative efforts of developers all over the world.
Finally, software "pirates" illegally copy legitimate software products, selling them for a tiny fraction of software vendors’ usual prices. The Business Software Alliance, an industry trade organization, estimates that about 40% of all software in use in the world today was copied illegally. Competition from software pirates limits the price that software vendors can charge for their products.
F. The Durability of Software Products
Software never wears out. As a result, every software vendor must compete with its own installed base of existing products. If Microsoft did not continuously improve Windows, for example, customers would have little reason to upgrade to newer versions of the product, whether by purchasing a new copy of Windows for an existing computer or by purchasing a new computer preinstalled with a new version of the product. This is not merely a hypothetical possibility; for example, many customers did not regard MS-DOS 4.01, largely the product of IBM’s development efforts, to represent a significant improvement over MS-DOS 3.3, and therefore declined to upgrade until Microsoft developed and released MS-DOS 5.0. New versions of Windows therefore must offer significant new capabilities and other advantages in order to be accepted by customers and drive sales of new computers, which benefits the entire industry.
III.
MICROSOFT’S OPERATING SYSTEM BUSINESS MODEL AND THE
BENEFITS OF THAT MODEL TO VARIOUS CONSTITUENCIES
At the outset of the personal computer era, Microsoft adopted a business model for developing and licensing operating systems that took into account the factors set forth above that make the industry so highly competitive. That model is based on investing for the long term to develop new operating system software products that anticipate (and create) customer demand and distributing those products at low prices to attain high volumes.
Many of Microsoft’s competitors pursued (and continue to pursue) a very different approach, relying on relatively modest sales of expensive computer systems that include their own operating system software. In particular, companies such as IBM, Sun and Apple primarily develop and market complete computer systems—hardware and software all in one box for one price—based on technology available only from that vendor. The "single vendor" approach to computer production has at least one clear benefit: restricting design work to a single company simplifies the engineering process of making sure that all of the various components work well together. There is a big drawback, however: such proprietary computer systems are generally incompatible with one another. Incompatibility means that the potential customer base for such products is necessarily limited, leading to higher prices (thanks to lower volumes and less competition), reduced customer choice and less innovation—both in the computer itself and in related products such as hardware peripherals and software applications.
When personal computers were first introduced in the late 1970s and early 1980s, manufacturers generally followed the single vendor model of traditional mainframe and minicomputer manufacturers. Leading offerings included the Tandy TRS-80, the Xerox Star, the Apple II, the Commodore PET, the Osborne I, the Kaypro II and others. These computers ran on different operating systems and were generally incompatible with one another in terms of the customer interface they offered and the application programs they supported.
Consumers value compatibility. Compatibility makes computers easier to learn and leads to a plethora of applications. Microsoft understood in the early 1980s that the potential of personal computers would not be unleashed until the industry provided a broad set of compatible hardware and software products. To help bring that about, Microsoft developed and promoted an important part of the standard personal computer architecture—a set of operating systems.
In Section III.A, I provide a brief overview of Microsoft’s business model and then focus on three specific elements of that model: (i) Microsoft’s support for software developers, (ii) Microsoft’s efforts to improve the Windows platform and (iii) Microsoft’s efforts to protect the integrity of that platform so that its benefits for computer manufacturers, software developers, and customers can be realized. These benefits are briefly described in Section III.B.
A. Microsoft’s Business Model
1. Overview
Microsoft’s operating system business model consists of four elements.
Firstly, Microsoft develops operating system products that serve as a stable, well-defined development platform to provide compatibility across a broad range of hardware and software products. In other words, Windows serves as a common layer in the personal computer architecture that masks differences in the underlying hardware and that provides services for application programs. This means that software developers can efficiently create applications that utilize these services (accessed via APIs), knowing that their applications will run across computers offered by a wide range of manufacturers. This also means that customers can choose from among a wide range of compatible hardware and software products and know that those products will work correctly together.
Secondly, Microsoft makes sure that its operating system products provide an easy-to-use and complete set of services that allow customers to manage their computer through a consistent user interface. Providing such a consistent user interface gives customers a uniform way to access their applications, data files, documents and peripheral devices (such as printers)—whether those resources are located on the customer’s computer or on remote computers. Consistency also allows customers to transfer their learning to other machines, to learn from one another and to draw upon support from the resources of the broader industry (including books, videotapes, classes, consultants and so forth).
Thirdly, Microsoft broadly licenses operating system products to computer manufacturers at attractive prices (typically less than 5% of the price of a new computer). Such broad licensing promotes the adoption and use of Microsoft’s operating system products, which in turn promotes the development of a wide range of useful complementary hardware and software products that are compatible with Windows and thus with other Windows-related products. Microsoft can afford to price Windows low because its strategy is to license it in large volumes, unlike competitors such as Sun, Apple and IBM that have generally sought to earn revenues from smaller sales of more expensive hardware products.
Fourthly, Microsoft works closely with hardware manufacturers, software developers, and customers to understand their needs and to promote their use of Windows technologies. For example, Microsoft meets regularly with hardware and software companies to brief them on Microsoft’s operating system development plans, obtain feedback on defining and improving the Win 32 APIs and actively "evangelize" their use of the platform capabilities of Microsoft’s operating system products. Microsoft also invests large sums of money in gathering customer feedback and in "usability testing." Microsoft uses all the feedback it obtains from the industry and customers to improve Windows so that it offers software developers and customers the opportunity to exploit the growing power of microprocessors.
In short, Microsoft has adopted a "high volume/low price" business model that is predicated on providing customers with the benefits of compatibility. This business model has been particularly attractive to Microsoft because of the exceedingly low marginal cost of software. By licensing its operating system software broadly at low prices, Microsoft has helped to foster the development of an industry in which consumers today can choose from among thousands of personal computer models and tens of thousands of applications that are generally compatible with one another.
Microsoft believes that its business model is a winning strategy in the software industry because it provides benefits for all key constituencies, as outlined below in Section III.B.
2. Support for Windows Developers
A key aspect of Microsoft’s successful business model—from 1975 to the present—has been the company’s focus on providing other software developers with great tools, training and technical support to help them build great software products for Windows. This fiscal year Microsoft will devote about $630 million to efforts aimed at helping software developers write great applications for Microsoft’s Windows family of operating system products. Approximately 2,000 of Microsoft’s 27,000 employees work full-time helping developers.
To my knowledge, no company has ever devoted greater resources to helping software developers than Microsoft. These efforts entail (i) the development of innovative, award-winning software development tools and programming languages (such as Microsoft’s Visual Studio, Visual Basic, Visual C++ and Visual J++), (ii) the extensive and early dissemination of technical information about Microsoft technologies and training programs and (iii) the provision of marketing assistance and product support.
Microsoft works with key software developers from the very outset of important new technology initiatives. Through a series of technical briefings called the "Open Process," Microsoft solicits industry input on new technologies at the stage when the product’s specifications are being drawn up. Microsoft wants to know from the outset that it is building operating systems that software developers want, and the Open Process helps us achieve that objective.
As the Open Process advances, specifications are shared with increasing numbers of developers and further feedback is solicited. The process benefits all concerned because it enables Microsoft to develop and deliver products that are suited to developers’ needs and it enables developers to understand the likely future direction of technology initiatives from Microsoft early in the development process—assisting them in their development of complementary products on a timely basis.
The Open Process is well illustrated by Microsoft’s development of the Win32 API set, which is the set of APIs to which software developers write in creating applications for Windows NT or Windows 95 (and later versions). Microsoft sent the initial Win32 API specifications to 25 leading independent software vendors in November 1990, nearly three years before Windows NT was released and nearly five years before Windows 95 was released. During 1991, Microsoft sent updates of the Win32 API specifications to more than 300 independent software vendors and obtained useful feedback, which was reflected in later versions of the specifications. By January 1992, the specifications were posted free of charge on America Online, Compuserve and the Internet. By the time Windows NT was released in July 1993, the Win 32 API set was the most thoroughly reviewed API set in history.
Microsoft has continued to improve the Win32 API set since its initial public release in 1993, providing information to developers and soliciting valuable feedback as the Win 32 API set evolves. (One example of Microsoft’s improvements to the Win32 API set, of course, are the hundreds of new APIs that Microsoft has added over the past few years to enable software developers to take advantage of Internet technologies, such as the HTML display format, in their products. These APIs are provided by the Internet Explorer technologies in Windows, a subject about which my colleague Jim Allchin will testify in greater detail.)
Microsoft also holds an annual event called the "System Design Review" to which nearly 200 of the industry’s top software developers are invited to learn about Microsoft operating system initiatives and to provide feedback. On a much broader scale, Microsoft "evangelizes" its products to other software developers through a wide variety of methods, such as developer conferences and the regular dissemination of software and support via the Internet (and online services before the Internet became popular) and CD-ROMs.
In addition to the above events, Microsoft regularly holds "Professional Developers Conferences" to provide professional developers with an understanding of Microsoft’s technical initiatives to improve the Windows platform. At Microsoft’s most recent Professional Developers Conference, attended by over 5000 people, Microsoft held more than 200 technical seminars over four days, and distributed the most recent beta of Windows NT 5.0 (recently renamed Windows 2000), which is now under development. Microsoft also holds other developer conferences throughout the year all over the country and abroad. "Tech Ed" is a multi-day event that draws 10,000 developers for education and training on current technology offerings from Microsoft. "Developer Days" are annual, one-day events hosted by local technology experts in more than 120 cities in 51 countries. "Microsoft at the Movies" is a one-day event where education and training presentations are beamed by satellite into movie theatres all over the country to give developers easy, convenient access to Microsoft information.
In total, Microsoft provided information to more than 450,000 software developers last year at conferences and other events. Employees of Microsoft’s primary competitors, including Sun, Netscape, Oracle and IBM routinely attend these events and benefit from them.
To provide information and software to the broadest reach of software developers, Microsoft offers a variety of "self-help" programs under the Microsoft Developer Network ("MSDN") umbrella. For example, MSDN Online is a free program that offers online training, online access to the MSDN Library (which contains a vast store of technical information, including sample software code), and other benefits to developers. As of October 1998, about one million software developers belonged to MSDN Online. Similarly, the MSDN Independent Software Vendor Program provides software developers, free of charge, with opportunities to participate in joint marketing activities with Microsoft. All in all, Microsoft has shipped nearly 100 million CD-ROMs to software developers over the past five years through the MSDN Subscriptions program.
Again, Microsoft provides this valuable information to its direct competitors. For example, between 1995 and 1998 nearly 5,000 CD-ROMs with information about Microsoft operating systems and sample software code were sent to Netscape alone.
Microsoft also makes ongoing training courses available to professional software developers in the form of the "Microsoft Mastering Series" courseware titles. These courses are available in both an inexpensive, self-study CD-ROM form, and through traditional instructor-led classes offered at local Microsoft Certified Training and Education Centers ("CTECs"), which are independent training companies that offer training services to local developers. Through these programs, Microsoft trains over 200,000 developers per year in a variety of tools, technologies, and programming techniques to help them be more successful writing software for Windows.
Microsoft must continually work hard to win developer support because developers have the choice of writing for a variety of different platforms, and they do so. Software developers are actively courted by various platform vendors, and thus are a demanding group.
To better understand the extent to which developers create software for multiple platforms, Microsoft contracted with Market Decisions Corp, an experienced independent research firm, in May 1998 to conduct an unbiased survey of 760 professional software developers. The survey was undertaken in the regular course of Microsoft’s business, not for use in any lawsuit. The developers chosen for the study were randomly selected from among independent software vendors, corporate MIS departments and third party software consulting firms, and are believed to be broadly representative of the software developer community at large. The study shows that the vast majority of developers create software applications for multiple platforms. While 91% of developers report targeting some version of Microsoft Windows, a full 82% of developers reported targeting a variety of non-Windows platforms as well. In particular, 41% of developers create software for various flavors of UNIX, 36% target Novell Netware, 22% write for Sun’s Java, 19% develop for IBM’s AS/400 and about 10% of developers target each of Apple’s Macintosh, Digital Equipment Corporation’s VMS, IBM’s MVS and IBM’s OS/2 operating systems. Many developers create particular applications so that they can be adapted relatively easily to work on multiple platforms.
Although we are always evaluating how we can do a better job, Microsoft’s efforts have paid off in developer satisfaction. In the May 1998 study described above, developers were asked which company comes to mind first on several attributes that affect developer satisfaction. In response to question after question, Microsoft ranked significantly higher in developer satisfaction than IBM, which generally ranked second. For example,
- When asked which company "provides high quality support and support programs for developers," 3.75 times as many developers answered "Microsoft" than the next highest platform competitor, IBM.
- When asked which company "treats developers like valuable customers," 4.4 times as many developers answered "Microsoft" than the next highest platform competitor, again IBM.
- When asked which company "keeps you informed, providing information in a timely fashion," 5 times as many developers answered "Microsoft" than the next highest platform competitor, IBM.
- Finally, when asked which company "offers a broad range of development tools that work well together," 11.4 times as many developers answered "Microsoft" than answered the next highest platform competitor, IBM.
Netscape and Sun ranked far lower on all developer satisfaction scores than Microsoft.
There is still a lot of room for improvement when it comes to making software development easier, and we intend to do better in the future. Nevertheless, I believe that Microsoft has done a very good job providing developers with the resources needed to write great Windows applications, and this hard work and long term investment has contributed to the success of the Windows platform. Unique among computer platform vendors in this regard, Microsoft was literally founded as a company focused on helping developers create great software, and that focus continues to play a central role in Microsoft's mission today.
Microsoft also provides strong support for hardware developers (microprocessor vendors, computer manufacturers and other hardware vendors) so that their products work well with Windows. For example, Microsoft offers "Device Development Toolkits" (similar to Software Development Toolkits), porting labs, an annual technology conference called "WinHEC" (for "Windows Hardware Engineering Conference") that focuses on hardware and the like.
3. Innovation in the Windows Platform
Another key aspect of Microsoft’s successful business model is that the company works tirelessly to advance the Windows platform. There is no doubt that Windows would quickly be displaced if the market believed that Microsoft had lost its commitment to improving the Windows platform to meet new computing needs, such as advanced graphics, networking and support for Internet-related technologies like innovative HTML support.
Microsoft has always invested heavily in research and development ("R&D") to improve its products. Over the past few years, Microsoft’s investment in R&D has grown rapidly as new opportunities and new threats have emerged. Whereas Microsoft spent about $350 million on R&D in 1992, Microsoft will devote over $2 billion to R&D this fiscal year, of which about $1 billion will be devoted to the Windows family of operating systems.
Recognizing that the company’s future is wholly dependent upon advancing the technologies we offer to consumers, Microsoft devotes an unusually large portion of its revenues—about 17%—to R&D. The comparable figure for Intel, Oracle and Sun is about 10%. IBM invests just 6% of its revenues in R&D.
Microsoft’s expenditures on R&D include a commitment to long-term, basic research. Microsoft’s Research Group, which consists of over 300 researchers, is one of the largest commitments to basic research in the industry. Microsoft hopes that the basic research now underway will lead to the development of breakthrough software technologies for the next century, such as technologies that will make speech recognition software truly usable on a mass consumer basis or software that will enable computers to recognize people and anticipate their needs.
Microsoft makes a focused and large investment in R&D to ensure that Windows embodies the latest technical advances in ways that are simple and effective for customers and developers to use. Among other things, we have developed high quality implementations of such important features as the "point and click" graphical user interface, networking, security, object technology, Internet-related technologies, multimedia, three-dimensional graphics and so forth.
We have also worked hard to integrate these new technologies with one another so that they are easier to use in combination with each other. Integration often makes it possible to do new things. For example, the integration of the underlying three-dimensional graphics support, multimedia support and HTML support in Windows 98 makes it possible to develop Web pages that present information in ways that simply were not possible before, using animated 3D models, movie images and synchronized sound.
Another example of the benefits of integration is provided by Microsoft’s efforts to steadily update the Win32 APIs to conform to a "common object model." A common object model is a set of specifications describing how software program modules (or "objects") can advertise their functions so that other software programs can access them. This integration, which spans all components of the system and constitutes innovation in and of itself, benefits software developers (both inside and outside of Microsoft) by making it easier to develop new applications. Software developers do not have to constantly "reinvent the wheel" if they are able easily to call upon existing "objects" that perform functions required by their applications. Microsoft’s development of a sophisticated common object model is important for Microsoft strategically because its platform competitors are developing competing object technologies.
It is important to note that Microsoft’s development of new features and functionality in Windows does not prevent other software publishers from offering similar technologies to run on Windows. For example, Microsoft has integrated networking support into Windows, but Novell can and does offer an alternative networking scheme. Other vendors offer alternative three dimensional graphics systems for use with Windows. And, of course, Netscape’s Web browsing software and Sun’s Java software are widely used with Windows.
In many cases, Microsoft takes the lead in promoting new technologies years before they become widely utilized by software developers. For example, Microsoft began promoting CD-ROM based technology in 1986, eight years before such technology became widely deployed in the industry. Microsoft was a pioneer in using object-oriented technology to allow different kinds of information (e.g., a spreadsheet in a word-processing document) to be placed in one document and edited together. It took several years for this concept of "object linking and embedding" to be accepted, but now it is critical to the way in which customers use their software. Microsoft began trials on "Intelligent TVs" in 1995; yet this technology is only now gathering momentum. Unlike certain other companies in the industry, Microsoft invests for the long term, persevering even when good ideas take several years to produce sizable returns.
Since Windows faces a wide range of competition, adding new capabilities to Windows does not necessarily enable Microsoft to charge more for the product. In fact, Microsoft believed that it could not raise the price for Windows 98 significantly above that charged for Windows 95, even though Microsoft had invested hundreds of millions of dollars over three years in creating an improved version of the operating system. If Microsoft had the "monopoly power" posited by the plaintiffs, we surely would have sought to obtain a significantly higher price for Windows 98.
As noted above, one advantage enjoyed by companies such as Apple is that company engineers can closely coordinate the design and development of all the components of the computer system, since they supply all the pieces—hardware and software. Unlike these companies using the "single vendor" approach, Microsoft cannot advance the Windows platform merely by devoting its own developers to the task. Designing and developing a well-engineered computer system running Windows is a considerably more daunting task, as it requires ongoing discussions and close cooperation with a very wide range of hardware manufacturers and software publishers. It is precisely this effort that makes it possible for Windows to work on so many systems and to give customers so much choice among vendors.
For instance, Microsoft worked closely with many hardware manufacturers and software publishers to develop and promote the "Plug and Play" standard. Plug and Play allows any computer running Windows to recognize and quickly configure any Plug and Play compatible hardware device, such as a modem, a video card or a printer. Apple was able to provide a similar capability for the Apple Macintosh earlier because Apple controls both the operating system software and the hardware in Apple computers.
In seeking to promote the development of a broad range of competitive and compatible products, Microsoft often collaborates in various ways with companies that are, or may be, developing technologies that duplicate functionality that is or will be provided by Windows. Such collaboration benefits the entire industry. Companies developing 3D graphics-intensive games, for example, need to understand Microsoft’s development plans for providing graphics capabilities in Windows so that they do not unnecessarily duplicate that engineering work on their own (though of course they are free to do so).
Even in cases where software developers determine to develop technology that duplicates technology provided by Windows, there may be opportunities to obtain other technical design efficiencies relating to the interaction of each company’s products, or at least to avoid designs that are directly incompatible with one another and can lead to system malfunctions, installation problems or ease-of-use shortcomings. Customers benefit from discussions between Microsoft and other companies concerning ways to avoid such incompatibilities. For example, Microsoft helped software vendors such as Netscape use the Internet Shortcuts technology in Windows 95 to enable any Web browser to become the "default browser" on Windows. My colleague, Dan Rosen, covers this in greater detail in his testimony.
4. The Integrity of the Windows Platform
Microsoft’s business model is also dependent on its ability to maintain the integrity of Windows as a stable, cohesive, well-defined platform that provides compatibility across a broad range of products and meets ever-evolving market demand for new system services (made available through APIs).
In general, software developers will be disinclined to use system services unless they know that those services will be widely installed on users’ computers. Software developers would lack that assurance if computer manufacturers were permitted to dismember Windows 98, installing some parts of the operating system and not others. In such circumstances, software developers would not know whether particular system services relied on by their applications would be present on any given brand of personal computers.
If the ability to define the contours of the operating system were transferred from its creator, Microsoft, to computer manufacturers acting individually—so that each computer manufacturer could create its own idiosyncratic version of a "Windows-like" operating system—there is little doubt that Windows would become highly fragmented. A "tragedy of the commons" problem would likely result as each manufacturer, seemingly acting rationally in its own self interest, would modify Windows in various ways to claim some differentiation from competitors’ versions of Windows. Before long, the very thing that made Windows useful in the first place—the compatibility it provides across hardware and software—would be destroyed, slowing the development of new products, reducing the utility of personal computers built on Windows and, in the end, reducing demand for such computers generally.
If computer manufacturers were allowed to alter Windows as they please, software developers would likely be forced to develop separate versions of their products for each unique "flavor" of Windows. Software development costs—and prices—would undoubtedly increase, and developers’ incentives to invest in new product development would be reduced.
Another possibility is that in an effort to mitigate the operating system fragmentation, software developers would redistribute with their products the Microsoft operating system software code that provided the missing system services their products needed to function. Occasionally, Microsoft has to allow software developers to distribute operating system software in this manner in order to fix bugs in existing Windows subsystems or to add an important new feature to a previously installed version of Windows. There are a number of important drawbacks to this approach, however: (i) the matrix of possible configurations of customer systems that has to be tested by Microsoft, other software developers and computer manufacturers becomes very large; (ii) different software publishers’ products can install incompatible versions of Microsoft operating system software; (iii) costs for software media increase; (iv) support calls from users go up, and so forth. This phenomenon, derisively labeled "DLL hell," benefits no one. As a result, licensing software publishers to redistribute Microsoft operating system software is an option that Microsoft has learned to use only with a lot of caution. In a situation where every computer manufacturer had a different version of the operating system, it would be a complete nightmare.
The UNIX world has suffered from such operating system fragmentation, which is one of the principal reasons why UNIX has higher associated prices and lower numbers of users. UNIX is a sophisticated operating system originally developed by AT&T Bell Laboratories in the 1970s and licensed widely to various hardware manufacturers. Each manufacturer was free to modify UNIX as it saw fit, and they did so. Sun’s version of UNIX is called Solaris, Hewlett Packard’s version is called HP-UX, IBM’s version is called AIX, Data General’s version is called DG-UX, and so forth. Today there are more than 15 leading "flavors" of UNIX, all of which are generally incompatible with one another to varying extents, despite all being based on the same "core" code or services.
Each of these vendors sells high-end personal computers (often called workstations) running their own flavor of UNIX. This means that software publishers have to specifically adapt and test their software application programs on each combination of hardware and software. Furthermore, slightly different development tools are required for each platform. As a result, these publishers’ customers face significantly higher prices and a smaller range of software applications than customers that purchase workstations built on Microsoft’s Windows NT. This is one of the main reasons why unit shipments of Windows NT have grown far faster than unit shipments of UNIX systems since Windows NT was first released five years ago.
Similarly, it is important that customers be able to rely on Windows being delivered to them in a form that meets their expectations. They need to know that (i) the application software they purchase will function as expected (because no part of Windows has been removed), (ii) the training investment they have made will be applicable, (iii) their internal support personnel know what they are dealing with, and (iv) they can accurately describe their computer systems when they call Microsoft or other organizations for external support.
While Microsoft allows customers themselves to tailor Windows in any way they see fit, Microsoft needs to be able to offer Windows to customers, through any channel of a distribution, in a consistent way. Providing Windows in this consistent form allows customers to have a secure foundation from which to reap the benefits described above.
B. Benefits of Microsoft’s Business Model
Microsoft’s business model provides substantial benefits to many constituencies, described briefly below.
1. Benefits to Computer Manufacturers
Windows enables any manufacturer to develop personal computers that leverage Microsoft’s substantial investments in developing operating system technology. And any manufacturer can produce personal computers that provide customers with the benefit of the huge base of applications software and customer support and learning that has been built up by customers and the industry.
If every computer manufacturer were required to design, develop and test its own operating system (as Commodore, Tandy and others used to do), far fewer personal computer manufacturers would exist, and the computers they sold would be more expensive and have fewer capabilities, which would suppress demand for computers overall.
The resources of any single company are inherently limited, making it hard for computer manufacturers acting alone to match the rapid pace of innovation in hardware products designed to work with Windows. For example, although Apple has always been a great engineering organization, its repeated failure to license its platform technology broadly left Apple to compete as a single company against the combined resources of literally hundreds of hardware manufacturers investing to improve the Windows architecture. Similarly, fewer software developers create products for the Apple Macintosh because there are fewer Apple Macintosh customers to buy such products.
For these reasons, Bill Gates urged Apple in 1985 to license its Macintosh operating system software broadly to other computer manufacturers, and even offered to help Apple do so. (A copy of Mr. Gates’ July 8, 1985 letter to John Sculley and an accompanying memorandum dated June 25, 1985 is Defendant’s Exhibit 2245.) On July 29, 1985, Mr. Gates sent a follow-up letter to Mr. Sculley, noting that he "remain[ed] enthusiastic about the benefits of licensing Mac technology" and offering "to help in any way I can with the licensing." (See DX 2246.) In an article in the Harvard Business Review, Andrew S. Rappaport and Shmuel Halevi said the following of Apple: "The missed opportunity was vast. Put simply, Apple could have been Microsoft." (See DX 2247 at 72.)
2. Benefits to Software Developers
For software developers, the breadth of services built into Windows means that their software products can (i) leverage "best-of-breed" software technologies in Windows that are developed and tested by Microsoft and (ii) be written once to run across the broad range of computer hardware running Windows.
For example, thanks to recent innovations in Windows, the operating system now provides support for high performance three-dimensional graphics that any software developer can use. Such "3D" support is important not only to developers of computer games, but also to developers of computer aided design software and developers exploring innovative new ways of presenting business information. In the past, software developers had to incur the expense and time of developing 3D software themselves, or license it from third parties. Once the 3D code was written, the developers faced the difficulty of trying to persuade hardware manufacturers to provide compatible graphics hardware, or else trying to make their own software compatible with whatever graphics hardware they anticipated customers might use, which varied from machine to machine. As a result, 3D software tended to be expensive and provided limited hardware support, which led the broader software development community to shy away from it. By including 3D system services (called "Direct 3D") in Windows and encouraging hardware manufacturers to both support it and develop compatible hardware, Microsoft has made it easier for software developers to create great 3D applications to run on Windows. I could provide many additional examples.
3. Benefits to Customers
For customers, Microsoft’s business model provides a tremendous range of choice in hardware and software and a broad set of learning and resources to call upon. Importantly, Microsoft’s business model also promotes vigorous innovation and price competition in the computer industry.
The contrast for customers between computer offerings from manufacturers such as Compaq, Dell, and Micron, which have built their businesses on low-cost operating system software from Microsoft, and manufacturers such as Apple and Sun, which generally distribute their operating system software only bundled with expensive hardware, is clear. Manufacturers such as Apple and Sun are necessarily dependent upon charging their customers high prices so they can earn a profit on relatively modest hardware sales. Such manufacturers also generally do not benefit from the economies of scale enjoyed by computer manufacturers that assemble hardware components manufactured in large quantities for computers running Windows. In addition, these manufacturers generally seek to obtain "premium" pricing for their products because they face less competition than the hundreds of manufacturers that offer personal computers running Windows.
For these reasons, prices of Apple’s Macintosh computers have historically been a few hundred dollars higher than prices of comparable personal computers running Windows. The same is true of the price of Sun’s Solaris workstations in comparison to comparable offerings running Windows NT from Compaq or Hewlett Packard. Similarly, software applications written to run on Windows tend to be less expensive than software written for competing operating systems because software dev
