In the 1970s, a small company called Digital Research thought life was simple. Microsoft did languages, DR did operating systems, and a bunch of companies did applications.
These applications included Wordstar word processing, Software Arts' Visicalc spreadsheet and Ashton-Tate's dBase database.
In the 1980s, Novell came along to provide Netware networking, and in the 1990s, Netscape set a standard for browsers.
However, it will not have escaped anyone's notice that now operating systems, languages, office applications, networking and browsing capabilities are all provided by one company - Microsoft. No wonder it has grown so large.
Unfortunately, just about everyone who wants Microsoft's PC client software already has it, and the main source of new cash is getting all those who have it to pay for it. Hence the introduction of licensing schemes and product activation. The software may see no real growth unless sales of PCs take off again.
Over the past decade, Microsoft has found growth on the server side, with Windows and Back Office. It hasn't done badly. Before NT, Microsoft's server market share was more or less zero.
But now it is being squeezed between IBM, which has a monopoly of large-scale corporate computing, and GNU/Linux and the Apache Web server, which are coming up from below.
So, where can Microsoft find the growth it needs? Part of the answer lies in home games consoles (Xbox) and mobile phones (Smartphone 2002). But the company is also placing its hopes on another area - business applications.
This is where Microsoft Business Solutions (MBS)comes in. The plan is to bring functions such as customer relationship management to small and medium-sized enterprises.
MBS is not starting from scratch: Microsoft has bought Great Plains and Navision to get it going. And the huge Microsoft marketing machine could help turn this foothold into a global business.
The target is not the giant corporations that buy from IBM, Siebel and other heavyweights - or not yet. It is companies whose CRM systems are a Rolodex or, perhaps, a copy of ACT. This is a fragmented market that is ripe for consolidation or, to use another word, takeover.
In the last three months of 2002, MBS had revenues of only $139m (£84.8m), and it made a loss of $93m. These results place it a long way behind the office applications division, which generated $2.4bn in sales and $1.9bn in profits in the same period.
However, MBS' sales almost doubled from $73m in the same quarter in 2001. And it's the growth that matters. When you have $43.4bn in cash, short-term losses are much less important than the opportunity to build a $10bn business.
Jack Schofield is computer editor of the Guardian