.. e that allows the company to terminate the agreement at any time for any cause; in this event, the receiving party must return all copies of Intel specifications. The NDA also covers samples, or prototypes of unreleased Intel processors, which must be returned if the NDA is terminated. Without these prototypes, system makers cannot test their designs with the new Intel processors and therefore cannot ship their products in a timely fashion. In the workstation or server markets, most vendors design their own boards and ASICs.
Many PC makers simply buy third-party chip sets and motherboards that work with the latest Intel processors, but, to differentiate their products, the top PC vendors do their own design work. Thus, to be a major system vendor, an Intel NDA is an essential requirement. Clearly then, Intel’s termination of a customer’s NDA is a severe punishment. Intel admits to taking this step twice in the past year, first with Digital then with Intergraph. In the first case, Intel took action after Digital launched a hostile lawsuit without warning, accusing Intel of stealing intellectual property and then lying about it.
In this case, Intel was provoked into taking drastic action, but one can ask whether the punishment fit the crime. Within six months, Digital chose to settle the suit, at least in part because it needed a good relationship with Intel to remain competitive. The Intergraph situation is even more questionable. Intergraph had threatened patent-infringement lawsuits not against Intel but against some of Intel’s other customers. These customers asked Intel to indemnify them, shielding them from Intergraph’s patents. Instead, Intel asked Intergraph to stop asserting its patents; after Intergraph refused, Intel terminated its NDA. Intergraph has since filed suit against Intel for failing to give the system maker access to technical data and other anticompetitive behavior. According to contract law, Intel can terminate NDAs whenever it chooses and can partner with whomever it wants.
Antitrust law, however, applies a higher standard to a company that is the only supplier of a critical component, a category into which Intel clearly falls. Such a company must treat its customers fairly and equally.Intel claims that it does treat its customers fairly and equally, but that Digital and Intergraph acted unfairly and thus deserved to be singled out. By this interpretation, Intel was merely quelling bullies who were causing trouble in the neighborhood. But suppose Digital or Intergraph has valid patents that are being infringed? Shouldn’t these companies have the opportunity to prove their case in a court of law? Instead, Intel can easily bring these companies to their knees simply by cutting off their NDAs. I believe Intel’s dominant position gives it an added responsibility to support all of its large customers equally.
Any grievances it has with a system vendor should be settled through negotiation or in court, not by cutting off that vendor’s ability to deliver competitive products. Allowing Intel to terminate NDAs arbitrarily gives an already dominant vendor too much power over the entire computer-systems market. Intel Seeks Salvation on Internet A lack of compelling new applications is reducing demand for Intel’s more expensive processors, dampening the company’s revenue growth. Little progress has been made since then. Intel’s latest strategy is to use the Internet to build demand for Pentium III–a strategy that has as much to do with Microsoft’s intransigence as with the processor’s technical capabilities. Over the past year and a half, Intel has tried to build demand for CPU-hungry technologies such as digital photography, video-conferencing, 3D graphics, and voice recognition.
Although some have caught on to a modest degree, they remain niche applications without appeal to broad markets, particularly the businesses that still buy the majority of PCs. These technologies have not become ubiquitous for a variety of reasons, but perhaps the biggest is Microsoft’s refusal to embrace them. With Windows, Office, and other applications, Microsoft controls the core software used on most PCs. But with no strong competitor to spur its creativity, the Redmond giant sees no need to improve its user interfaces with 3D graphics or voice recognition. At the recent WinHEC, the company paid little heed to these areas other than a lame 3D-GUI demo that may not have any more impact than last year’s ill-fated Chromeffects demo. While waiting for Microsoft to upgrade its user interfaces, Intel has been unable to build widespread interest in 3D or voice recognition, the very tasks that Pentium III is best at. Intel’s new strategy revolves around the Internet, something that Microsoft doesn’t fully control.
By working with many plug-in vendors and Web-site owners, Intel is spurring the use of 3D, audio, and video on the Web in ways that make a high-end processor more attractive. Of course, these features are already available today via browser plug-ins. The vast majority of Web sites, however, consist of text and simple graphics, leading to the widespread belief that users who are just surfing the Web should buy the cheapest available PC. Intel’s fervent hope is to turn this paradigm around, such that anyone using the Web–which means just about anyone with a PC–will want a more powerful processor. To start, Intel has made sure that important plug-ins, such as Macromedia Shockwave and RealNetworks Real Player, are available with Pentium III enhancements.
Other plug-ins, such as Haptek’s Virtual Friend, have not been widely used in the past but deliver a more compelling experience with the new enhancements. To showcase these new features, Intel has put together a Pentium III owner’s site, www.intelweboutfitter.com, but this site is mainly an exclusive race track: it’s fun to drive around the site with your Pentium III (no other processors allowed), but after you finish, you haven’t accomplished much. To build demand for its high-end processors, Intel is working with more-mainstream sites such as www.disney.com, www.etoys.com, and www.sharperimage.com. These sites draw a lot of traffic, most of it from people with processors less powerful than a Pentium III. Such visitors may want a processor upgrade after seeing the new features on these sites.
For example, Sharper Image allows you to download a 3D image of certain products so you can examine them from multiple angles, just as you would in a physical store. This feature works with Pentium IIs and other slower processors, but with SSE, the images have more resolution and/or better lighting, making them look more realistic. Compute power, not bandwidth, is the issue here; unlike video, 3D images work well with merely a standard modem. Getting all these plug-in vendors and Web sites behind Pentium III before the product was even launched required plenty of legwork by Intel–and quite a bit of money. This funding is typically in the form of technical assistance to implement the new features but is sometimes a direct investment from Intel’s $400 million venture fund.
This campaign is well beyond what other CPU makers can afford. Conceivably, a K6 III processor with 3DNow could offer an experience similar to that of Pentium III, but most of the plug-ins don’t support 3DNow. Intel says its agreements with vendors are nonexclusive and don’t bar them from supporting non-Intel processors, but AMD can’t match Intel’s ante to these vendors. External competition is not Intel’s primary concern, however. The company must keep PC buyers from focusing exclusively on Celeron to the detriment of Intel’s more profitable high-end products. Given Microsoft’s delays, Intel’s Internet strategy could be the quickest solution to this problem.
Success requires that more sites adopt 3D and other high-performance features; if PC buyers appreciate these additions, their affinity for the cheapest PCs could be broken. Socket Strategy Challenging for AMD Although AMD has struggled to earn a profit over the past several quarters, some good signs have emerged recently. We estimate the company’s share of the PC processor market rose to 12% in March of 98, and it is particularly strong in the U.S. retail market, capturing 35% in June (according to ZD Market Intelligence). While AMD may be profitable for the rest of the year, maintaining growth and profitability in 1999 will be challenging. The company’s growing market share is due in part to its internal improvements.
The K6-2 is a strong product that matches all of Intel’s high-volume speed grades and even includes 3D acceleration features that Intel’s chips lack. Perhaps more important, the company appears to have overcome the manufacturing problems that kept it from shipping enough parts in 1997 and early 1998. After some initial problems, the new 0.25-micron process is ramping well and has completely replaced the older process. AMD’s gains, however, should also be viewed within the context of PC market cycles. When Intel began pulling out of the 486 market (some say prematurely) in 1995, AMD’s market share surged to 17% as it became the number-one supplier of 486 processors. During 1996, however, the company was unable to keep PC makers interested in 486 motherboards.
As a result, AMD’s market share plummeted to 9% by April 96. This crash was particularly severe due to the vendor’s lateness in delivering a Pentium-class product. If history repeats itself, AMD should be nearing the top of its cycle. Intel’s withdrawal from the Pentium market is nearly complete, and PC makers are turning to the K6 to fill their Socket 7 motherboards. But in the past, PC makers have tended to follow Intel to new sockets, most lagging Intel by no more than a year. This window set a limit on AMD’s ability sell into the older socket.
This cycle, however, could be different. The PC market is much larger than three years ago; even at 10% to 15% of the market, AMD today represents enough units to keep chip-set makers and motherboard vendors interested. Intel is also helping in this regard: by short-sightedly walling off Slot 1, it has given the alternative chip-set makers little choice but to support AMD’s plans. PC makers have also changed. In the past, the market was so small and uniform that they tended to use a single motherboard for all systems.
This discouraged support of alternative sockets. Today, many PC makers have separate motherboards for business and consumer systems, and for mainstream and low-cost systems. The new model allows the use of Socket 7 for low-cost consumer systems, for example, without changing other platforms. Intel itself is recommending that OEMs use Slot 1 for mainstream systems and the new Mendocino socket in low-cost systems next year. These changes will keep the Socket 7 market from disappearing as quickly as in previous processor transitions.
We project continued sales of Socket 7 processors well into the year 2000, although the total shipments of these chips is likely to gradually decline during that time. The shape of this decline poses challenges for AMD. That vendor now supplies about 75% of all Socket 7 processors, with the remaining share held by cutthroats like Cyrix and IDT, which have recently been playing limbo with Socket 7 prices. (The current winner is IDT with its $30 WinChip-180.) AMD has mainly stayed above this fray, but to increase its Socket 7 share, it would have to cut prices (and margins) to compete with these smaller vendors. The alternative is to try to reverse the downward trend of Socket 7 sales by getting OEMs that have already converted to Slot 1 to go back to Socket 7.
Most of these systems are entrenched Intel design wins; to make headway in this difficult area, AMD needs both a strong product offering and some weakness on Intel’s part. Intel isn’t planning to cooperate. As PC prices dropped, AMD benefited from Intel’s slow response time. Recent indications, however, are that PC prices may be firming, and Intel has finally responded with a strategy. Celeron brought Intel into sub-$800 PCs, and we expect the combination of the socketed Mendocino and the integrated Whitney chip set will enable Intel-branded sub-$600 PCs by mid-1999.
This combination will not leave significant market segments for AMD to dominate. As long as OEMs want a competitive CPU market, AMD has a good shot. Although its reliance on Socket 7 poses challenges, the company’s success will ultimately be determined by the performance of its processors. With a strong product portfolio, AMD should be able to maintain its share in 1999 or even see some growth. Sizable gains, however, will be more difficult to come by without aggressive pricing that would jeopardize profitability. In review, the myths about the sub-$1000 computer raises many questions.
Even though they may represent 40% of the PC market, two-thirds of all PCs are still bought by businesses. With all the new and undecided technology, businesses are not convinced that the low end computer is the best deal because there is another myth not mention in the review, you may get what you pay for. Bibliography Miller, Michael. Cybersex Shock. PC Magazine 10 Oct. 1995.
Shifting Into The Fast Lane. U.S. News & World Report. 23 Jan. 1995. Wilson, David.
The Internet goes Crackers. Education Digest. May 1995. Computers and Internet Essays.