What exactly is the market for enterprise systems?

By all accounts, it appears that the US Justice Department's decision on whether to block Oracle's takeover of PeopleSoft will hinge on an interesting question: how do you define the market in which Oracle and PeopleSoft compete? PeopleSoft wants to define the market narrowly: large companies, such as the Fortune 1000, that are looking for a single application suite to serve the entire enterprise. PeopleSoft argues that this market is essentially served by three vendors: SAP, Oracle, and PeopleSoft. Allowing the deal to go through will reduce that number to two, a duopoly.

In the opposite corner is Oracle. According to an article Wednesday in The Deal, Oracle is expected to argue that there aren't any antitrust issues, because in fact, many other vendors serve the Fortune 1000 market:

Oracle is expected to argue the Justice Department's definition of the software market affected by the deal is too narrow and without legal merit. [Oracle] also will try to show that companies similar to those complaining about the deal have alternatives to Oracle, PeopleSoft and Germany's SAP AG for enterprise software solutions. This will include examples of companies that design their own systems and those that buy software from other vendors such as Microsoft Corp., IBM Corp. and Lawson Software Inc....Another line of attack for Oracle is likely to focus on whether the DOJ staff appropriately limited the market to software sold to large companies. It is expected to show it would be easy for other enterprise software companies to reposition their products to serve this market. Microsoft, for example, intends to market its enterprise solution to one-third of the Fortune 1,000.

I find this argument unpersuasive for two reasons. First, Fortune 1000 companies shopping for a complete enterprise system across multiple business units and even multiple countries do not consider writing custom solutions. It just doesn't happen. Second, to consider Microsoft's enterprise solutions (e.g. Great Plains, Navision, Solomon, and Axapta) on the same scale as SAP, Oracle, and PeopleSoft, is ludicrous. Microsoft's products are targeted for the small to mid-size company market: good solutions, but they do not scale up for companies that require tens of thousands of users. Similarly, Lawson often goes up against PeopleSoft in some deals, especially in some key Lawson verticals such as healthcare, but usually only for HR systems or perhaps financials, not the complete suite of enterprise-wide functionality. And IBM doesn't develop enterprise application software, so I don't know how IBM got lumped in as an example. (IBM competes with Oracle in the database and application server markets, but not in application software.)

SAP, Oracle, and to a lesser extent, PeopleSoft, have become entrenched in large organizations. Other vendors such as Lawson, QAD, Microsoft, or SSA, often co-exist at the divisional or plant level. But it is the rare exception for one of them to rise up and displace SAP, Oracle, or PeopleSoft, at the enterprise level. If anything, other vendors tend to get displaced when the corporate office moves in to standardize on fewer applications. This is why, among large companies, the big three continue to gain market share, not lose it. When large companies merge, and one is running SAP and the other is running SSA's BPCS, guess which system gets replaced. I know I'll get feedback, pointing out the exceptions, but those are just that, exceptions.

I agree that in the future other vendors may join the ranks of SAP, Oracle, and PeopleSoft, but it won't be easy, and it won't be quick. The enterprise applications marketplace has gone through three major transitions in the last twenty years, each tied to a major shift in underlying technology: from host-based architecture (e.g. COPICS, AMAPS, MANMAN, MAPICS, J.D. Edwards, BPCS, PRMS) to client/server architecture (e.g. SAP, Oracle, Baan, PeopleSoft, J.D. Edwards), to Internet architecture (e.g. SAP, PeopleSoft, Oracle). In the transition from host-based to client/server based systems, there was a nearly complete turnover of leadership, with J.D. Edwards as the only major vendor that successfully made the transition--but only after learning from the mistakes of other vendors, and even then, nearly failing in the attempt. But notice that the transition from client/server to Internet architecture has been different. The three current leaders -- SAP, PeopleSoft, and Oracle -- all started as client/server offerings. And all three have now completed the transition to an Internet architecture. This will ensure their dominance in the large enterprise market for at least the next five to ten years, or until some other architectural shift takes place.

Therefore, it will be difficult to understand if the DOJ decides in favor of Oracle.

Update, Mar. 1: Read my follow up post regarding the argument that DOJ actually makes in its complaint against Oracle.