TFH Flashback: Critical Mass

by Timothy Prickett Morgan

Editor's note: The following story is from the November 1993 issue of The Four Hundred, in which I talked about how IBM's future is in software and services and how the company should consider getting into the application software business, particularly on the AS/400, given this fact. I also talked about how IBM's love affair with the Smalltalk object-oriented programming language would probably not pan out. Java burst onto the scene two years later to obliterate Smalltalk, but the issues they both faced among midrange customers were essentially the same. I hope you enjoy this flashback and find it interesting.

The bigger the customer base, the better. You could probably sum up all of computer system marketing, in both theory and practice, in that single sentence. Or the selling of just about any other product for that matter. Size is everything, because reaching critical mass ultimately determines the long-term success of a product. The widespread acceptance of a piece of hardware, software, or any other system component has positive benefits not only for its creator, but for the customers that make use of these products, too. The reason is simple: Competition and secondary businesses follow where customers' pocketbooks lead.

IBM is justifiably proud of the fact that there will be some quarter of a million AS/400s installed at customer sites by year's end. Though 250,000 is certainly a large installed base, there are as many System/3Xs still plugged in around the world. We think that the number of AS/400s out there could be--and needs to be--much, much higher. Maybe closer to a million, which might be attainable . . . provided IBM changes the way it sells AS/400s.

IBM has enormous control over the AS/400 base. It makes all the processors and just about all the main memory. With the exception of a relatively few disks, tape drives, and printers, it designs and builds nearly all AS/400 peripherals. Big Blue owns the operating system, and more importantly, its only data base management program--the feature that gives the AS/400 its unique personality. IBM also controls most of the programming products that customers and third parties use to build applications software--RPG and Cobol mostly, plus utilities such as its Applications Development Tools. IBM's PC Support software, which allows information to be shared between the AS/400 and PCs, has competition in the market, but few customers go outside of the IBM fold. Perhaps the most important thing IBM has that no one else has is foreknowledge of how all the AS/400's hardware and software will be enhanced over time. The thing that baffles us is that IBM doesn't leverage this coveted position. In particular, we are astounded that IBM has not seen fit to do for customers the thing they spend so much money and time doing: developing applications software.

Figure 1: Right now, AS/400 software represents a very small part of the AS/400 division's revenues, but it is a very profitable business for IBM. These numbers are estimates based on market information supplied by IBM. Maintenance figures include fees for System/3Xs and AS/400s.

Sure, IBM has dabbled in the selling AS/400 applications. When the AS/400 family was announced five years ago, IBM ported its applications software for manufacturers, distributors, construction firms, and educators from the System/36 and System/38 to the AS/400. But IBM's heart wasn't really in the sales and support of applications software. The reason IBM didn't push hard with software was not a technical one--IBM has as good (or bad) a reputation for software development as Apple, Microsoft, and other prominent vendors--but rather a practical one. IBM intended to sell a lot of AS/400s by letting other companies--those with software expertise--sell them for it. As resellers like System Software Associates, J.D. Edwards, Software 2000, Lawson, and others have gotten stronger, IBM has gradually backed away from the applications business, and has even sold the applications software it did own to outside companies.

The AS/400 resellers are important to IBM, and one cannot underestimate their part in the success of the AS/400 so far. Upwards of 70 percent of the AS/400s sold in the world are sold directly or indirectly by third party resellers. These companies have developed over 20,000 AS/400 applications programs. They buy their hardware and systems software at a discount from IBM, put their applications software on it and resell the resultant systems to customers. Because IBM has essentially substituted an indirect sales force--resellers--for a direct one, it cannot in good faith sell applications software and systems in competition with those resellers. Unless it changed the way it sold AS/400s, that is. IBM could decide to sell AS/400s more directly, and bundle applications software with the machines. IBM would then have to compete with the resellers on the turf it gave them.

IBM could, if it wished, be to AS/400 users what Microsoft is PC buyers. Microsoft has shown that it is possible to make very big bucks from little pieces of software. The company holds what some would call a monopoly on desktop systems software and a strong position in applications by keeping software costs very low. It has the added benefit of knowing how the two most popular operating systems for PCs--DOS and Windows--are going to change over time, so it can never be blindsided by a new feature in the operating system (or one that has been removed) like competitors Lotus, Borland and WordPerfect have to deal with. Because DOS and Windows are cheap, and systems that use this software are based on Intel processors that are also inexpensive, Microsoft is able to sell customers loads of applications software. A large portion of Microsoft's growing profits come from applications software, not operating systems. IBM has even more control over the AS/400 platform than Microsoft and Intel together have in the PC world, but IBM doesn't use its position to full advantage. It gives its opportunity away.

IBM has good reasons for acting the way it has. For instance, IBM has an aversion to running its resellers out of business. And it seems that IBM isn't keen on being an AS/400 applications software developer because it views that business as too arduous.

Figure 2: Microsoft, the biggest personal computer software vendor, has grown at a slightly faster rate than System Software Associates, the largest AS/400 software developer.

As above chart shows, System Software Associates--the biggest AS/400 applications developer--has grown pretty much at the same rate as Microsoft. But its profits, as a percentage of revenue, are less than half that of Microsoft. The old IBM, with its big overhead and bureaucratic ways, couldn't afford to be in such a low-margin business. Though the new (and presumably more efficient) IBM could change its mind and go back into the applications business, we doubt it will. However, we still think that there is a big market for small AS/400s equipped with very inexpensive systems and applications software. This market may be, in fact, much more lucrative and important than the one at which that the new server machines, announced last month, are targeted. Every company that installed a big PC server to run its bookkeeping in the last five years is in theory a prospect for a small AS/400. So, too, is every business with a local area network.

IBM says that what AS/400 customers really want is good tools that let them build their own applications more easily. Well, IBM is right, but not completely right. Big companies that write their own applications want systems software and development tools. Software houses that write applications for other companies want IBM to stick to its present plan. For these customers, programming tools ought to be IBM's focus for the foreseeable future. And so far AS/400 customers that are not professional software houses or companies big enough to have substantial programming staffs have gone along. This scheme would work as well tomorrow as it did yesterday except for one thing: IBM is about to radically change the way AS/400 applications software is written.

IBM's vision of the future of software development on the AS/400 can be summed up in a single bit of computer jargon: object-oriented programming. IBM's first attempt at giving AS/400 customers OOP development tools is Envy/400, which is an offshoot of the Smalltalk/V programming language. (We have taken a careful look at Envy/400 in this month's issue to help customers figure out how--and if--this type of software will be useful to them.)

IBM is absolutely convinced that OOP software is going to revolutionize the development of applications, much as microprocessors have redefined the hardware end of the computer business. The company figures that the OOP tools IBM and its business partners create will boost programmer productivity by one or two orders of magnitude. (Similar claims were made about fourth generation languages and computer-aided software engineering tools. Not too many AS/400 shops use such tools, but instead stick to RPG.)

We think that it will be a long time before typical customers jump on the OOP bandwagon; maybe they won't get on it at all. System vendors and third party software developers can probably reap benefits from OOP because the emerging technology does seem to reduce the cost of developing software that will be used, with variations, by a large number of customers. But the only way OOP could become universal is if it is not only good but also economical. So far, the price of OOP development software is far too great to make it a hit among ordinary AS/400 customers. The AS/400 division has to ask itself if it wants to succeed as Microsoft has by offering applications software deals too good for customers to refuse, or if it will be content to reach a small audience with its new software and therefore charge more per unit to make money in a low volume business. If the future of IBM is in software and services, as two IBM chairmen have said in the past year, then the future is now and the only route open is the one with low tolls for customers.