To the casual observer, it might seem that Intel is stuck in a business model that requires its customer base to buy ever-faster chips in perpetuity even as the market for faster chips is fading.
But, in reality, when you look more closely at the interplay among the company's many divisions, new products, research efforts, and investments - and the new Internet economy - there is a business strategy shift in motion. Intel's business model requires it to expand its customer base, which will buy ever faster and more various chips.
The company's recent activities challenge the impression that Intel is a chip company alone. Certainly its new Intel Online Services (IOS), as well as its newly formed Communications Product Group, point to a company that is focusing on finding solutions for emerging e-business problems.
True to its new mission statement "to be the pre-eminent building block supplier to the worldwide Internet economy", Intel is now involved in everything from flash memory and set-top boxes to desktops, servers, routers, network interface cards, and switches. The company is building a framework into which all of its far-flung efforts will fit, and that framework will now capitalise on the Internet rather than on the PC revolution. It will also expand Intel's role in IT purchases to include broad infrastructure components.
Yet both insiders and outsiders are quick to point out that most of Intel's revenue still comes from making microprocessors for the computer industry.
"CPUs, chip sets, [and] motherboards are the $US25 billion piece of the $30 billion pie," says Richard Dracott, director of e-business marketing for Intel's architecture division.
"I'm not sure competition from any company in the desktop market is Intel's biggest problem. A bigger problem is generating demand for its faster, more profitable desktop chips," says Keith Diefendorff, a senior analyst at MicroDesign Resources, a research company in the Silicon Valley.
But the chip-making giant appears to get it, according to one economist at the University of Texas. "One could say that no company can survive in this new Internet economy without constantly reinventing itself," says Andrew Whinston at the Center for Research in Electronic Commerce. "The economy is under major reconstruction."
Reviving the fundamentals
No one argues that faster servers will be unnecessary some day, but another oversimplification is the idea that, at some point, IT executives will no longer need to follow the traditional Intel road map for at least desktop and mobile systems.
Although MicroDesign's Diefendorff says he sees the Internet as a bandwidth-dominated rather than a processor-dominated environment, when you talk to senior IT executives, bandwidth or not, they remain loyal to the Intel processor road map.
And the same loyalty is apparent in the mobile market. Steve Pintarich - a service delivery coordinator at Computer Sciences Corp (CSC), an outside IT management group for such companies as Otis Elevator and Raytheon - deploys hundreds of Dell Latitude series laptops and also believes in the Intel road map.
"[SpeedStep] is the next step, and we're always looking at what's available to give our users the most current technology," Pintarich says.
Intel's Dracott concedes that things are moving beyond the desktop, but that "[Intel will] move at a faster speed needed to empower employees. The model here is to support a broad range of clients, cell phones, PCs, and thin clients," Dracott adds.
But if Intel has an Achilles' heel in its core markets, some analysts say it may be in the high-end servers that both enterprise-level businesses and ISPs are demanding to run their e-commerce-focused data centres.
"Unix on an Intel box is not a first choice," says Linley Gwenapp, a principal analyst at The Linley Group.
Intel servers are not currently able to satisfy the performance requirements of e-commerce, Gwenapp says. And Intel appears to understand that "big hulking servers", as Intel CEO Craig Barrett calls them, are needed to fuel the new economy.
So over the past two years, Intel has moved 50 per cent of the research-and-development dollars spent on microprocessor research from IA-32 processors to IA-64. The big push starts in the second half of 2000, when the first Itanium processor ships.
Gwenapp believes that giving IT executives a choice between a proprietary box from Sun or IBM and an Intel box offered by multiple suppliers will change things.
"Above and beyond technology is the business model," says Gwenapp. "If an IT manager buys Sun e4000s, they just married Sun. If price/performance gets out of line or Sun falls behind though, they are in trouble."
Strength in diversity
Even if IA-64 fills a substantial portion of the growing need for high-end servers to feed the Internet economy, senior management at Intel, led by Barrett, recognises that in the Internet age, with alternative devices proliferating, increasing margin pressure from new competitors, and the sheer size of Intel, it is no longer reasonable to expect growth at the 30 per cent-plus rate the company experienced through most of the 90s.
"I'd be very pleased with 15 to 20 per cent growth. I think that is realistic for a company our size," Barrett says. To do that, the Intel road map is diversifying in order to supply silicon and products at all levels of Internet infrastructure.
In the second half of this year, the desktop division will launch not only IA-64 but an entirely new line of 32-bit processors starting with Willamette. Willamette performance is expected to have an initial clock speed of about 1GHz and a floating-point performance 20 per cent greater than Pentium IIIs running at the same speed.
StrongArm II, Intel's answer for the handheld market, is also set to ship in the second half of 2000 and is expected to offer about 400MHz performance at extremely low power.
Intel is also ramping up both research-and-development dollars and acquisitions to become a building-block supplier for the Internet infrastructure. The framework for Intel's expanding network infrastructure product line is the Intel IXA (Internet Exchange Architecture). This is a concerted effort to build a range of silicon solutions and products - from routers and low-end to high-end Gigabit switches to DSL and cable modems - that fit a single architecture.
"[IXA] is the same model they used in the PC space. They want to be the main supplier for all of the components inside the box," Gwenapp says.
James Smith, an Intel spokesman from the networking communications group, echoes the same sentiments.
"You can compare [IXA] to what we did with PCs, where we supply the processor, chip sets, and graphics chips, for example," Smith says. "IXA is the similar architecture for building communications equipment."
Intel is also going beyond simply supplying the building blocks to the new breed of e-business data centres by offering complete hardware solutions for routing, switching, Web caching, and load-balancing equipment.
According to Scott Richardson, general manager of the communications and Internet server division, the new Intel focus sits above the network plumbing, and the company will offer Internet infrastructure appliances that drop into the network.
"We are focused on delivering products and solutions to help small-to-medium enterprises get connected and do business over the Internet, and products will exist on both ends of the wire - on the business side and to service providers," Richardson says.
Testing uncharted waters
Internet age or not, no one denies that most of Intel's business is still focused on selling atoms, not bits. However, Intel is not taking itself completely out of the bits business. Last September, the company launched its IOS division for Web and database hosting.
The company will increase its investment in Online Services to about $US2 billion during the next three to four years, opening centres in Japan and the UK, says Mike Aymer, vice president of IOS.
IOS is the centrepiece of Intel's fledgling New Business Group, which is chasing what is projected to be an $8 billion data-hosting and $10 billion Web-hosting market by 2003, according to analyst IDC.
"Our goal is to be one of the leading providers," says Aymer, pointing out that a typical data centre with 10,000 servers requires about $150 million in networking and infrastructure.
However, to build this business, Intel needs to work alongside other application service providers (ASPs) at its centres. "ASPs will be hosted in our facility," says Aymer. "We are an ASP for the stable things, like Web hosting and databases. Our stable of apps will only be a partial solution. I don't view us as competition with companies like Oracle Business Online or MyASP."
Also included in this division are Intel Architecture Labs, which conduct most of the research, and something Aymer calls "Edge Services".
"The concept is to deliver services on the edge of the Internet, like streaming media," Aymer says.
Reluctant to give many details beyond calling it an "exploratory effort" that, for example, looks to deliver "video out to the edge", Aymer's explanation hints that Intel intends to stay nimble in the Internet age.
Intel invested in Quokka Sports, an Internet extreme sports channel that does Web-casting of live events, pushing multimedia content out to the final user.
"We've been working seriously with Intel Edge Services on improving the streaming media experience out to the end user," says Pascal Wattaugh, vice president of technology at Quokka Sports in California.
Building on the basics
Despite Intel's ventures in Web hosting, services to the edge of the Internet, and the communications equipment business for data centres, the company is still fundamentally about silicon. Take Intel Capital's seeding of local Web portals from China to India to Poland, for example.
"Why would we invest in a portal in Poland?" asks Stephen Nachtsheim, director of operations at Intel Capital, the company's investment arm. "Because local language will drive people to want to use the Internet. Since the predominant device to access the Internet is the PC, it will sell more PCs."
Patrick Gelsinger, vice president and general manager of Intel's desktop products group, explains it simply: "No matter how you slice it, the power behind the Web is silicon. The Internet runs on silicon - in the routers, servers - and that's what Intel is good at."
Timely access to information can give your company a competitive edge in today's business world, so speedy performance of key business applications is a critical factor. Although upgrading to the fastest Intel processor has been part of the solution, bandwidth is becoming an increasingly important factor in this equation.
Sluggish, poor-performing software is a pain to use, resulting in underutilised applications and frustrated users. To add burden, many business applications such as customer relationship management, enterprise resource planning and help desk systems are hosted on a server with a Web or client software interface - with most of the processing being done on the server side.
So even though the computers themselves are getting faster, application performance isn't. Because organisations are moving to a networked model more and more, the actual speed of computers has become less important, and the bandwidth available has become most important.
So what does this mean for the decisions you make to support your business? Investing in faster desktops for your workers will be less beneficial. Instead, make sure mobile workers are using the fastest modems available - and DSL, if possible. Remote offices will see a more significant benefit from moving to a faster link to the home office rather than moving to the newest PIIIs.
Of course, CPU performance is still very important on the server side. If your servers can't keep up with the network traffic, then your application performance will suffer for all users. Interconnection speeds between servers is equally important. At a minimum, Fast Ethernet is the only way to go for application and database servers.
When planning for annual upgrades, keep bandwidth at the top of your list. In the long run, faster connections between your sites will be of greater benefit to your users. And often the costs of these upgrades will be less expensive than trying to keep up with the newest expensive desktop PCs.