(FORTUNE Magazine) – What a difference a deal makes. For ten years, Microsoft's power in the computer industry has grown inexorably and unchecked. The government tried, a couple of times, to corral the company in court. But now a real competitor may be emerging: IBM. Believe it or not, Big Blue is b-a-a-ck. With its purchase of Lotus Development, the industry-superpower-turned-bureaucratic-behemoth suddenly seems most likely to succeed in heading off Microsoft.

Even some at Microsoft evidently believe that the purchase of Lotus poses a threat. Says an internal Microsoft document obtained by Fortune: "Lotus if acquired by a major industry player or competitor could have a significant impact on Microsoft."

Every computer company today has a major Microsoft problem. The experience of an IBM senior manager illustrates why. On a recent flight, the IBMer found himself seated next to an engineering professor, who immediately wanted to talk about Microsoft. "They seem unstoppable," exclaimed the professor. "You guys must be worried." A bit irritated, the IBMer explained that IBM's revenues last year were $64 billion. "And how big do you suppose Microsoft's were?" he asked. "Oh, maybe $40 billion, $45 billion," the professor guessed. Correct answer: about $5 billion. The professor was astounded. "In the Sixties and Seventies people talked about IBM not as a member of the industry but as the environment," says Jim Moore, whose GeoPartners Research in Cambridge, Massachusetts, helps craft strategy for some of the biggest technology companies. "Now Microsoft has become the environment."

Nothing frustrates IBM CEO Lou Gerstner--and every other computer industry leader--more than the disproportionate mindshare Bill Gates and his crew at Microsoft command. How can it be, a reasonable person might ask, that such a relatively puny outfit matters so much in a $400-billion-a-year worldwide industry? Why, for that matter, is Microsoft's market capitalization $49 billion, practically equal to IBM's value of $52 billion? To understand the answer is to understand why IBM bought Lotus.

Your computer may come from Compaq, Packard Bell, Acer, or Zenith. Your printer might be made by Canon or Hewlett-Packard. Your monitor might be from Sony or any of a dozen other companies. But one product--made by one company--is common to almost every system: Microsoft Windows. Gates & Co. sells 80% of PC operating systems worldwide. The operating system is the glue that holds together the entire industry, ensuring that all the software runs on all the hardware. This explains why Gates is so rich, why Microsoft is so powerful, and why every other computer company is merely chopped liver. IBM, for its part, has failed so miserably and repeatedly to develop successful applications for the PC that almost nobody knows that it, and not Microsoft, is actually the world's largest software company. IBM's $11 billion in software revenue last year--mostly for large computers--was more than double Microsoft's total sales.

The only realistic way to challenge Microsoft is to usurp its role as provider of the software you see every time you sit down at your desk. Right now, just one product has that potential--Lotus Notes. With the purchase of Notes, IBM has finally found a way to get back on the desktop. Even more than market share, IBM is buying mindshare.

Notes is crucial to IBM for another reason. The industry is moving into a new phase--networked computing--in which an isolated PC is not much more useful than a paperweight. People want to link their PCs to corporate computers, online services, and other far-flung networks. To share work across such networks, they need special software, called groupware; by far the best groupware on the market is Notes, the flagship product of Lotus.

Think of Notes not in comparison with PC applications software like Microsoft Word, but rather alongside the big Kahuna--Windows itself. Notes is where a user spends his or her day, communicating with colleagues and launching other programs designed to work with Notes. In the Microsoft internal memo, dated less than two weeks before IBM's bid for Lotus and labeled Confidential, a senior financial analyst offers Steve Ballmer, of Microsoft's office of the president, a wide-ranging assessment of Lotus's market and financial position. In a section entitled "Implications for Microsoft," the analyst writes: "Notes still has the momentum to become the industry standard for groupware and the dominant platform for applications development, potentially reducing Windows's role as a platform for applications development." Microsoft downplays the report--"It's just water-cooler talk," says a senior executive.

But if IBM makes the right strategic decisions--if it holds enough hands at Lotus to make the acquisition work, if it sells Notes cheaply enough (until recently Notes was priced for the carriage trade, at $495 a copy)--it has a chance of regaining some of its glory as a new era in computing begins. "For the first time since the IBM PC was introduced in 1981, IBM is ahead of Microsoft on something," says Marc Schulman, a Stamford, Connecticut, consultant.

The question is, can it stay ahead? Gates, clearly, has no intention of sharing a spotlight. "Notes is a product, not a strategy," he told FORTUNE after the IBM deal. "It has a position we think Exchange [Microsoft's groupware] may dent as it comes to market." Gates has singlehandedly put a lot of dents in the business of every other software company, if not driven them out of business. Besides Lotus, his only real PC software rivals are Novell, whose Netware controls the basic functions of PC networks, and Oracle, which makes database software. Each is less than half Microsoft's size. Gates has left tire tracks on PC makers too. By establishing software, specifically Windows, as the keystone of the industry, he has turned PC hardware into a commodity.

Figuring out how to respond to Microsoft's power has become a near obsession for some computer industry colossi--among them AT&T (1994 revenues: $75 billion), Hewlett-Packard ($25 billion), and Sun Microsystems ($5 billion). All built huge chunks of their business by tightly linking their own software to their own hardware. AT&T's software and hardware are mainly in its telephone network, but its dilemma is similar to that of the systems makers. If Gates' software becomes all that matters, these companies could shrink. Some could even die.

Meanwhile, Compaq ($11 billion), the leader in business PCs, and Digital Equipment ($13.5 billion), a humbled veteran from the pre-PC era, have allied themselves with Microsoft. If Gates succeeds, so will they.

Gates self-consciously tries to sustain the company's image as dauntingly capable and his own as a youthful genius. He has become so identified with his company that in the public mind he is Microsoft. From a marketing standpoint, a bunch of faceless corporations are competing against a superhuman billionaire. George Bush to the contrary, the vision thing does matter. "Microsoft's greatest strength is perception," says Bill Raduchel, chief strategist for Sun Microsystems. "Our customers get lots of pressure from end users in their organizations to go with Microsoft."

Like Microsoft, Lotus Notes has buzz. It is the first popular product to take full advantage of what is called client-server computing. Extra-powerful PCs called "servers" (in contrast with desktop PCs, their "clients") tap into vast amounts of stored data and route it to individual PCs when needed. The client machine may not even contain the software for manipulating the data. That too may be on the server. Notes enables co-workers scattered around a building or across the globe to share thoughts, data, and programs almost as if they were sitting at the same desk. It functions as a kind of super-versatile online service for the company.

Because versions of Notes reside on both the client and the server machines, it provides a platform upon which other applications can be built. That's part of what makes it like Windows. An entire subindustry of third-party companies has created more than 700 commercial applications that work with Notes. It also has its own "user interface"--or onscreen appearance--as does Windows. Result: Users consider themselves to be "in" Notes rather than in Windows, even though Notes requires Windows or another operating system to work. Lotus CEO Jim Manzi has bragged for years about how Notes can supplant Windows in the user's mind.

The more minds Notes infiltrates, the better IBM's chance of diminishing the importance of Windows. IBM has already said that it will use all its formidable resources, including its sales force in 140 countries, to promote Notes. Unlike Lotus, IBM can market Notes the way Microsoft does Windows--by selling it cheap and counting on large volumes to boost profits. IBM could even consider giving Notes away. It might turn out to be the all-time best loss leader the computer industry ever had.

Lotus by itself couldn't sell Notes rapidly enough because it had its own crippling Microsoft problem: Its bread-and-butter business of desktop applications was being gobbled up by Gates. To make up for lost revenues as sales of its 1-2-3 spreadsheet and other applications tanked, the company charged way too much for Notes. Today Notes is on more than 1.5 million PCs, a sizable number until you realize that 65 million people use Windows. Total Notes revenues are around $300 million a year, vs. an estimated $1.5 billion for Windows.

While buying Lotus is smart, this deal alone won't save IBM from the devastating effects of Microsoft's software onslaught. Gates wants to impose the organizational dynamics of the PC business on the entire world of computing. Right now the PC industry is split into distinct subindustries whose products work together. Companies compete in four main areas: microprocessor chips, PC hardware, operating system software, and applications software. Microsoft and Intel have acquired unparalleled power by completely dominating their respective realms: operating systems and microprocessors. (For more on Intel, see the cover story.)

The PC economy, if it prevails, will tear apart old-style integrated businesses like IBM's. Says Steve Ballmer, Microsoft's top marketing executive (and its other billionaire): "This is really about new world vs. old world economics. There's a group of people who have never embraced the PC economic model." IBM, for example, still gets about 40% of its revenues, and even more of its profits, from computer systems made from its own software and hardware, estimates Bob Djurdjevic of Phoenix's Annex Research.

Ballmer has some fairly radical--and self-serving--advice for IBM: The company should get out of hardware and software entirely. After blithely suggesting that IBM abandon three-quarters of its business, he goes on to urge IBM to use its systems-integration expertise to help customers assemble components made by others, like you know who. "Maybe we should be partners in some way," says Ballmer. "All the time, I hear customers say, I want Microsoft's technology and IBM's handholding and installation." But what if IBM suddenly had the right technology, like Notes? Wouldn't customers want to buy PC software from the same hand they hold?

Microsoft has a simple strategy for dominating client-server computing: to establish Windows as firmly on the server as it has on the client. With the disarming bluntness typical of Microsoft's top mangers, Ballmer says: "Big theme No. 1 for long-term strategy is that Windows sits everywhere." The battle to control the server is undecided. Microsoft's gladiator is Windows NT (NT stands for new technology). The product has not figured in the past year's debate about the company's allegedly monopolistic behavior, but if NT succeeds, Microsoft could get a hammerlock on corporate computing. It could also potentially help achieve another of Gates' goals--to become the biggest wheel on the information superhighway.

The operating system on a typical server today is Unix, a program developed at AT&T in 1969, primarily for controlling the phone network. Hewlett-Packard, Sun, and IBM each sell unique versions of Unix that work best on their own computers. Unix servers sell for $40,000 or more each, making this a lucrative business for all these companies. The Unix server market currently totals about $14 billion a year, calculates International Data Corp.

Microsoft has given NT a set of features that outdistances any version of Unix. It combines three kinds of software in one. Like Unix, it resides on a server and manages applications so they can be shared by a group of PCs. Like NetWare, the product that made Novell the powerhouse of networking software, it enables client PCs in a network to share files and printers. And like its predecessor versions of Windows, NT is as easy to use as the operating system on a desktop PC. Microsoft intends NT to be the only operating system that companies need to create a client-server computing network. Furthermore, it aims to make NT both cheaper and simpler than Unix or any other alternative.

Unlike each version of Unix, NT will run on standard machines made by hundreds of companies. Computer makers will inevitably drive the price of NT hardware far below that of comparable Unix servers, just as competition drove the price of Windows PCs below that of Apple's Macintosh. Compaq already sells NT servers for as little as $3,500. Says Jim Allchin, who runs Microsoft's business systems division and is in charge of NT: "We're into commoditizing stuff that used to cost a lot of money."

To build a universe of application programs for NT rapidly, Microsoft requires any developer that writes software for Windows 95 to make it work on NT as well. Though only 1,200 applications run on NT now, company executives predict that within a year there will be 7,000 or more. That would easily outshine Sun, to take just one example: It has about 3,300 applications on its Unix machines.

Microsoft will try to use its position as operating system supplier to expand into server applications. This is the same strategy it used on the desktop. It bundled together its word processor, spreadsheet, and other applications into the Microsoft Office suite, and sold it to customers who bought Windows. Now it is building a client-server bundle, called Microsoft BackOffice, that contains Exchange (Microsoft's E-mail and groupware product), programs for managing and supervising a network, and a database. This last is critical because many of today's most popular client-server applications, from companies like PeopleSoft and SAP, require a database to function. Microsoft will price its combination of NT and a database, called SQL Server, lower than Unix and a database from Oracle or another company. Allchin, who also supervises BackOffice, says he is creating an environment for developing client-server programs that is far better than any alternative system, including Lotus Notes.

The Unix camp's response is that Microsoft is limiting choices for the customer and moving back toward the old vertically integrated model of so-called proprietary systems. Sun CEO Scott McNealy says Microsoft wants to put corporate technology buyers in the same position Nintendo puts parents. "The purchasing agents have zero power against Nintendo, because their kid is screaming at them. They have no choice but to buy Nintendo game cartridges. That's where Microsoft wants to go, because in that environment a company doesn't have to compete." McNealy has a point. While Lotus Notes works on a variety of operating systems--including NT, IBM's OS/2, Apple's Mac OS, and Unix--Microsoft BackOffice does not. Customers who buy applications built around BackOffice are locking themselves in to NT.

But corporate computing managers are so fed up with the complexity of today's systems that they may be willing to tolerate less choice. Gary Fernandes, who heads marketing for EDS, the largest computer services company, predicts that NT and BackOffice will be a big success. "What customers want is results," he says. "They're far less concerned about competition in the market."

Gates has made no secret of his desire to to keep Microsoft growing by developing ongoing revenues from transactions, as opposed to one-time product sales. NT dovetails with this strategy. If it becomes entrenched in corporate computing networks, it could help Microsoft negotiate a cut of any business done across the networks. Such transactions could include very common activities--like routing phone calls--that today generate lots of money for Microsoft's rivals. Gates downplays this possibility. "Capturing transaction revenue requires a lot more than selling an operating system," he says. But Allchin, the guy who's actually in charge of NT, mentions projects that could lead to royalties for Microsoft when, say, voice or video messages pass through an NT system.

Where does this leave the rest of the computer industry? Scrambling, except for Compaq and Digital Equipment. Compaq can go with the Microsoft flow because it has proved it can thrive building hardware systems many see as a commodity. Digital fell far behind in the Unix market, so it too has intertwined itself with Microsoft. Although it competes with Compaq to sell NT servers using Intel chips, it is heavily promoting the fact that NT also runs on machines built with its own Alpha microprocessor.

Other computer companies are working hard to oppose Microsoft. "All major computer industry participants other than Digital and Compaq view Microsoft as a threat," wrote Marc Schulman in a recent essay, published by Boston's Patricia Seybold Group, that examined the company's dominance. What follows is a look at the stop-Microsoft strategies of three of the biggest. SUN MICROSYSTEMS hopes to defeat its rival in cyberspace. "The Internet is bigger than Microsoft," insists chief technology officer Eric Schmidt. Since the Internet is based entirely on publicly available software structures and standards, he argues, Microsoft can never dominate it. Says Schmidt: "The way you fight Microsoft is by offering services on the public networks." Consequently, Sun spends considerable time developing tools for using the Internet, a field executives say Microsoft has deliberately neglected since it too realizes the Internet is beyond its control. Sun has had tremendous success selling its Unix servers as hosts for Internet content, for example. A recent Internet Society survey found that 56% of Internet servers are Sun machines. In late May, Sun introduced HotJava, software to bring animation and multimedia onto the Internet's World Wide Web.

Sun also has vastly more experience than Microsoft in networked computing within corporations. For years its slogan has been "The network is the computer," a phrase that only recently began to make sense.

If NT succeeds, however, Sun is uniquely vulnerable, which may explain why the company's attitude toward Microsoft is uniquely belligerent and indignant. It is the only major maker of computer systems that is totally reliant on its own system software. Both IBM and Hewlett-Packard have large PC hardware businesses that can profit if NT does. Asked to discuss Microsoft, Sun's McNealy displays an almost manic defensiveness. He completely dismisses Microsoft's thrust into corporate computing, fuming that the company is nothing but a maker of desktop software, and overrated at that: "They're providing activities for individuals. We're making companies more productive. There isn't a company in the world that wouldn't be happier if they unplugged Microsoft Office. It's an activity generator. I've never used a word processor. I don't have time for it." As far as Sun is concerned, anything that slows down Microsoft--like the IBM-Lotus deal--is good.

HEWLETT-PACKARD, which has always been the consummate corporate pragmatist, is institutionally ambivalent about Microsoft. It intends to be a survivor whether or not NT succeeds. It is the leading maker of Unix computer systems, a highly profitable business that represents an estimated 15% of sales--about $4 billion. At the same time, H-P's PC business is growing rapidly (sales: $2.2 billion, according to a Salomon Brothers estimate). In May the company announced a partnership with Microsoft "to work together on the development, marketing, and support" of products based on NT.

CEO Lewis Platt refuses to say an unkind word about Microsoft, and insists he has no qualms about supporting both Unix and NT: "To deny that NT is making inroads into the client-server environment is absolutely ridiculous. We're serious about offering both Unix and NT. I'm not requiring either group to compete with its hands tied behind its back. Internal competition is a good thing. I've got the world's best insurance policy. If NT blows away Unix, we're fine."

But so great is the sensitivity inside H-P about the incipient head-butting between its Unix and NT businesses that the company wouldn't let anyone other than Platt speak to Fortune for this story. The company's greatest fear seems to be of signaling customers that it favors one system over the other. As Platt knows and as his competitors at Sun are eager to point out, H-P is walking a fine line. "They're just going to be an integrator of Intel and Microsoft products," chortles Ed Zander, who heads Sun's systems business. Says Platt, referring to the competition between NT and Unix within H-P: "That is a challenge. I will admit that."

AT&T displays its obsession with Microsoft not in what it says, but in what it does. It figured out even before IBM that Notes was a critical product for networked computing, and joined with Lotus to develop a service called AT&T Network Notes, now in market testing. Typically, a company installs Notes on servers scattered throughout its organization and then supports the systems and data itself. Network Notes would allow it to forgo using its own servers and keep all its Notes information on a network maintained by AT&T.

A company spokeswoman claims that the IBM/Lotus deal is good for Network Notes because "more people using Lotus Notes on the premises could create a larger market for Network Notes." But IBM is almost certain to offer a similar service on its own vast data network, which would put AT&T in the awkward position of paying royalties to one of its chief competitors.

AT&T is also market testing a related business it developed with Novell called NetWare Connect Services. Customers can use AT&T's network to bring together their widely dispersed computers. Some AT&T watchers see in these deals a strategy to shift the locus for corporate software off the company premises and onto AT&T's network. In effect, it is telling companies to give up their servers and instead pay it by the minute for computing. That would mean far fewer servers on which Microsoft could sell its software. John Petrillo, who runs AT&T's business communications services unit and supervises all these activities, denies any such grand scheme.

Though Petrillo refuses to comment directly on Microsoft, some detect a hint of desperation in AT&T's dealmaking. Says George Colony, president of Forrester Research in Cambridge: "AT&T's greatest fear is that it just has a big clear pipe, while Microsoft is building all the value outside that pipe."

Microsoft is not invincible. This spring it struggled to support 400,000 users for the test version of Windows 95. It will likely have more problems if thousands of corporations put NT at the center of their networks. Customer service is a crucial part of the computer business where mindshare alone won't do Microsoft much good. Because the company can expand only so fast, its big competitors still have time to mobilize. Meanwhile, Intel, Microsoft's longtime ally, is prodding it to bring better software to market faster.

What Microsoft can do is spend freely in its effort to prevail. Its desktop businesses remain hugely profitable. It has about $4.5 billion in cash. It has invested $500 million so far developing NT and BackOffice, and 1,000 people are working to improve and refine them. Says Ballmer: "We want to add more value every year for less money. That's goodness."

Yes and no. Good for consumers, but also a means for Microsoft to steamroll every company that gets in its way. Since the threat now extends to the industry's most powerful giants, the potency of Microsoft's approach will meet a severe test. In buying Lotus, IBM has thrown up the biggest roadblock Microsoft has seen in a while. A historic struggle is joined.