Microsoft to IBM: Tolerate PSI Mainframes or Quit Europe
Published: January 15, 2008
by Hesh Wiener
If you think Microsoft has an aversion to Article 82 of the European Treaty, the anti-monopoly law that forced it to open some Windows server APIs and also to live with others' media players on desktops, think again. Microsoft is backing Platform Solutions (PSI) in an Article 82 action that could loosen IBM's grip on the mainframe. If PSI is given the keys to every glass house in Europe, Microsoft will be one of its principal guests. In fact, PSI could get more from IBM than it wanted when it first announced mainframe-compatible systems.
Last November 27, PSI announced that it had raised more than $37 million in additional capital, and that one of the companies in the deal was Microsoft. Other investors in PSI include Intel, Blueprint Ventures, Goldman Sachs, InterWest Partners, and InvestCorp.
PSI has a firmware-based mainframe system that runs on Itanium servers from NEC and Hewlett-Packard. Beginning a few years ago, PSI put a few systems into user sites as test machines. IBM decided to shut down PSI if it could, and to do this via a Federal lawsuit filed in White Plains, New York, during November 2006. PSI answered and countersued. Since then, both parties have amended their original complaints. In the litigation, IBM says PSI has violated its patents and otherwise behaved illegitimately, asking the courts to affirm that IBM has every right to refuse to license its software to PSI and also to punish PSI for its transgressions. PSI, for its part, says IBM is all wrong, that it is obliged to license software for use on PSI systems, and that IBM should be punished for its abuse of a monopoly.
In other words, IBM and PSI are not getting along.
Meanwhile, PSI has kept up a sales effort that includes the marketing of low-end boxes through T3 Technologies. T3 used to be an IBM mainframe reseller and later added Flex-ES mainframe emulation systems to its product line. IBM eventually dropped T3 as a reseller and separately (and later) pretty much shut down sales of new Flex-ES systems. There are some Flex-based I/O control boxes still being sold and there's a related support business in which T3 participates, but it's not the business it once was and, from the looks of things, never will be again.
T3 didn't like any of IBM's behavior very much, and on November 26, 2007, the day before PSI announced its latest round of financing, it joined PSI's counterclaim. What that means is that it is asking the court that will try IBM versus PSI to hear its story as part of the IBM case. T3's filing in part echoes PSI's case and in part adds fresh, additional arguments to PSI's antitrust claims. T3 has added an economic angle to the PSI presentation, saying that it believes from its experience in the IBM mainframe market that IBM has been able to slow the rate at which mainframe price/performance improves since it squeezed out competitors Amdahl and Hitachi, despite rapid improvements in server value that have occurred in other parts of the computer market. T3's arguments, if accepted by the court, would show that there is a distinct mainframe market and that this is the relevant market for arguments about IBM's conduct. IBM spent many years persuading courts that the mainframe market is really part of a much bigger server market and that therefore IBM cannot be considered a monopolist.
(For those who want to read the U.S. case as presented by various parties, industry analyst Phil Payne has rolled the key court filings into one big hairball.)
The susceptibility to influence of pricing in one market segment by pricing in another, related segment (or a lack of such susceptibility) is crucial to the legal demarcation of a market than can be examined to see if any participants in that market have pricing power and if so, have they abused such power. American antitrust law outlaws the abuse of a monopoly position, not the holding of a monopoly per se. European law is similar, but there is a tendency among legal scholars to say that U.S. antitrust law in theory looks at the effect on consumers to determine if there is a monopoly and then if it has been abused, while European competition law looks at the set of suppliers in a market to see if there is a monopoly and whether it is abusing its position in ways that impact other suppliers. A lot of the difference comes down to semantics and to the methology of European legal proceedings compared to those in the U.S. Law in the United States has its roots in English law, and European law has roots in Napoleonic or civil law. Both seek justice, but English law is more of an adversarial process in which judges serve as referees, while Napoleonic law is often more accurately described as an investigation managed by judges or magistrates.
At this point, for IBM, PSI, T3, Microsoft, and particularly for mainframe users, it doesn't much matter how the court battle is fought. What counts is whether a court in the United States or Europe will tell IBM to loosen its grip on mainframes the way Microsoft was told last September it had lost a three-year-long court battle and that it had to loosen up (a decision it formally accepted in October). Microsoft not only agreed to provide lots of technical information, it also agreed to license patents covering technologies that were at issue at rates that were a fraction of the amount Microsoft originally asked. The patent royalty rate at the end of the battle was 0.4 percent of sales; Microsoft had originally demanded 5.95 percent. There are other aspects of the decision that affect the cost various parties must pay to use Microsoft's intellectual property, but the outcome is similar. Microsoft gets to keep its rights, but can only charge rivals a nominal amount.
PSI began its Article 82 action on October 19, according to wire stories carried a couple months later by Reuters. Compared to U.S. court actions, the European process lacks transparency, but key aspects will become more visible in due course. The main point is that PSI started a formal complaint in October, and about a month later, after interested parties had a chance to think and perhaps nose around the European competition bureaucracy, T3 joined PSI's countersuit and Microsoft joined PSI's capitalization. If PSI prevails, IBM could be compelled to choose between licensing the intellectual property in question at low rates, paying huge fines, or the unthinkable: pulling out of the mainframe business across the European Union.
The timeline doesn't provide answers, but it does, finally, allow users to see the big questions:
- What is it about the mainframe business that keeps PSI, its business partner T3, and its backers alive and feisty?
- What threat does IBM see in PSI that persuaded it go to war in the first place?
- What connects recent events with the imminent introduction of the z6-based mainframe?
The first question is easy. Every mainframe user knows that any IBM-compatible system that works and costs less than an IBM one will get 10 percent to 30 percent of the market by footprints. (In the old days, when compatible makers pushed very big iron, they got this kind of share by MIPS, but PSI systems come up from the bottom not down from the top, so the target will be measured in footprints rather than MIPS.) For PSI, it means there's a business out there. For T3, there may even be dominance of the low end, where IBM has not had a good product in some time. For PSI's backers in general, there's a legacy market in the U.S. and Eurozone and a booming developer's market in India and China. For Intel, there's the hope that the Itanium is not the Itanic. And for Microsoft, there's a way to get Windows into glass houses where mainframe and Unix boxes still dominate.
The second question is tricky, because when IBM began its litigation it didn't look like PSI could get a very large piece of the market, Flex boxes with all their quirks held sway among small users and developers, and IBM seemed to have more than enough to do getting its base to migrate to 64-bit technology. There might even have been advantages to having an alternative hardware supplier as a compatible lightning rod in shops where IBM had displeased a customer. Without any other mainframe options, an unhappy IBM customer has to move to another platform, one from which it will never return. This doesn't seem to add up to a big enough problem to provoke IBM into the morass of litigation. It might come down to the possibility that IBM's market models showed PSI getting a lot more business than PCMs did in the past, even if its machines could only run MVS-z/OS family operating systems. Or it might simply be that IBM believed PSI would quickly go broke once the court case was underway.
The third question is really about IBM's worries surrounding the transition from its current unique mainframe processor chips to new CPU circuits that are a branch of the Power family used in System p and System i boxes. Only IBM can address this issue, and IBM isn't talking except about how wonderful its z6 technology will be. An outsider can only speculate that there are dangers in this migration, and that the dangers are not due to flaws in z6 but rather the difference between its nature and that of current mainframe hardware. Uncertainty about the best way to make z6 a success despite its novelty could explain why IBM seems to have such a bad case of nerves.
Still, the most curious aspect of the whole matter may be IBM's persistence. It seems that it could long since have begun working toward some kind of settlement with PSI that leaves everyone alive and, if customers are pleased, healthy, too. Instead, the ante has been raised to absurd levels. PSI could be crushed. PSI's marketing partner could be forced to reinvent itself. PSI's investors could see their equity wiped out. On the other hand, Big Blue could end up being forced to not only license for low fees whatever IBM intellectual property PSI uses, but also to license other related technology that would enable PSI and others who wanted to get into the mainframe game to use IBM technology, IBM software, and IBM know-how. Having started out with an urge to curb PSI, IBM could end up making legal z/OS on the Hercules emulator a real possibility.
It has to make outsiders wonder whether IBM would have gotten itself into its current position if it had brought in wiser counsel in the first place. Some IBM watchers, including me, were a bit surprised when the initial lawsuit didn't bear the imprimatur of Cravath, Swain & Moore. If things start to go badly for IBM in this matter, it's hard to imagine which lawyers or advisors could pull its chestnuts out of the fire better than Cravath.
PSI Adopts NEC Itanium Servers for Mainframe Clones
Flex-ES And IBM: Can They Bend? Are They Broken?
Platform Solutions v IBM: Estoppel, Old Show Key
IBM Sues PSI: You Say Emulate, We Say Litigate
A Joint Assault on the Mainframe Hardware Market
Platform Solutions: Waiting for the Other Shoe to Drop
Post this story to del.icio.us
Post this story to Digg
Post this story to Slashdot