Technology News Technology News



Another Perspective



IBM v PSI:  THE OPERATION WAS A SUCCESS BUT THE PATIENT DIED

On July 2, IBM and Platform Solutions, Inc. (PSI) ended the legal battle IBM had initiated on November 29, 2006 without a scintilla of nobility.  IBM capitulated, buying its way out of the legal morass by acquiring PSI, possibly for as much as $240 million.  A portion of this sum went to PSI's lawyers; the rest went to the company's investors and, possbly, to its employees.  PSI capitulated, too.  In return for the money, PSI employees, investors, and representatives took a vow of secrecy.  Both sides called the result a victory.  What else would they say?

In the short run, the outcome is unlikely to have any visible impact on the mainframe business, which includes not just hardware and software but also a huge market for hosted computing and support services.  In the long run, though, mainframe users, mainframe software developers, independent mainframe hosting providers and others affected by the vaporization of PSI will have to adjust to new conditions.  Any hope these parties might have had for more competition in the mainframe hardware market is now likely to be extinguished.  While mainframes may in some ways compete at the margin with alternatives such as Unix, Linux, or Windows platforms, customers with legacy software, legacy personnel, and legacy information processing strategies will just have to live with Hobson's choice.

PSI had become the only possible commercial alternative to IBM in the mainframe market, despite the potential availability of emulators in the form of Hercules and Flex.

Hercules is an open software mainframe emulator that works under Linux, Windows, or Mac OS X.  IBM will not license its systems software and middleware for use with Hercules, but that has not stopped illicit experimentation.  People who ought to know what they are talking about say Hercules works and that it can run pretty much all the code IBM 64-bit and 31-bit hardware can run.  Nobody who knows about Hercules systems running current or recent IBM software in use by developers or by end users will go on the record about their experience because their conduct might be illegal and, worse, IBM might seek to enforce the rights it claims.  Still, there does not appear to be a single instance where this has occurred, so for the moment it looks like a Hercules underground will continue to exist and possibly, in some locales, thrive.

Flex-ES from Fundamental Software is a mainframe emulator IBM used to support in 64- or 31-bit mode for approved developers and in 31-bit mode for end users.  IBM and Fundamental seem to be in conflict these days, and the result has been an end to Flex licensing.  Installed end user systems are still allowed to run.  Developer systems, distributed under a time-limited license, are now history.  Some developers may have found a way to keep their old Flex systems running.  They are pretty quiet about this and IBM seems to have decided to look the other way.

IBM has not said what it will do with the PSI systems it apparently now has the rights to market.  These mainframes run on Itanium servers from NEC or HP.  IBM does not offer servers that use Itanium processors.  There is no technical reason PSI's technology cannot be ported to Power chips, but that would take time and also a change in IBM's internal politics, because IBM's charges for Power Unix servers are much lower than those for its z10 mainframes, which are, in IBM's words "siblings," of Power servers.

IBM's decision to buy PSI rather than pursue its attempt to crush it in the courts came not long after PSI mounted a vigorous effort to trim down IBM's legal claims.  PSI argued that IBM's trade secret claims were worthless for a number of reasons including IBM's longstanding full knowledge that PSI's had access to what IBM said were its secrets.  PSI said IBM had lost its right to complain that PSI misappropriated its secrets by tolerating the situation years after it had a possible right to object.

In essence, PSI said IBM was trying to rewrite history.  In the past IBM tolerated competition it later felt it could eliminate.  The change at IBM is a response to the change in the way the US Department of Justice felt about the consent decree that restrained IBM's behavior beginning in 1956.  PSI argued that IBM was obliged to stick with its past commitments, on which the PSI had relied.  IBM thought this was nonsense and that it could do what it pleased as long as the Justice Department remained silent.  Now, with the IBM v PSI case closed, the issue will not be heard by an American court.  However, there may be vestigial aspects to this dispute that will live on in the court system of the European Union, where competition cases can survive even when the parties that gave birth to a complaint have settled their differences.  And there is another lingering aspect to this affair, one that involves a company that marketed a midrange version of the PSI platform.

PSI had cut a deal with T3 Technologies, of Tampa, to sell a mainframe small enough to fit in a small rack.  T3 was once an IBM reseller and more recently served as a distributor of Flex boxes.  T3 not only has a business in the USA, it also has viable operations in Europe, the same legal venue where the embers of PSI's competition complaint may still be smoldering.  T3 had filed its own complaint against IBM in the US courts and, more recently, in the EU, too.  For now, T3 says it plans to pursue its legal battle and it might even ask courts on both sides of the Atlantic to compel PSI to keep providing mainframe-compatible technology for use with Itanium servers.  T3 might even be able to seek a way to be granted additional rights that would prevent IBM from suppressing PSI's work, which has roots going back to the days when Amdahl gave IBM quite a run for the mainframe customer's money.

Like PSI, T3 might work out a settlement with IBM, but, as PSI learned during the past few years, no agreement governing intellectual property or competition in the big iron business comes easy, not for IBM's adversaries and not for IBM, either.

To end its dispute with PSI, IBM had to pay a tidy sum.  The amount, thus far, is undisclosed, and that is permitted under the one American law that might bear directly on the acquisition, the Hart-Scott-Rodino Antitrust Improvements Act of 1976 (HSR).  That law requires parties to certain mergers that might eliminate or reduce competition against the public interest to seek permission in advance from the Federal Trade Commission and the US Department of Justice Antitrust Division.  However, there are financial thresholds below which an acquisition is considered too small to require reporting.  If the deal was worth less than roughly $60 million, it would be exempt under the chump change provision of HSR.

If the IBM-PSI deal closed at more than $60 million but below $240 million, it would only have to be reported under certain conditions.  One rule involves the size of the acquiring party, and IBM is more big enough to attract the attention of the law.  But the acquisition also has to involve the acquired party, and its size must be big enough to draw in the law, too.  The general rule covering the smaller party is that it must have sales or assets of at least $12 million.  (All the size figures are approximate and they are adjusted yearly based on US GDP.)

PSI's sales were essentially zilch.  The PSI systems in test were all loaners or systems for which a purchase may have been contemplated but not completed, or systems for which there was only nominal initial payment.  In any event, no matter what its arrangements with users, it is doubtful that PSI had sales of more than the trigger level of $12 million.  However there is a question of PSI's assets, which might by some reckoning be above the trigger threshold of $12 million.  The most recent round of financing, which included money from Microsoft, brought in about $37.5 million (some of which took out prior investors).

Could IBM have bought PSI for no more than $60 million and thus have its deal fall below the lower threshold for reporting?  A lot would depend on the deal PSI had with its lawyers, Susman Godfrey, who had taken on the case on a partially or wholly contingent basis.  They apparently earned their keep, in effect beating IBM's lawyers, Quinn Emanuel Urquhart Oliver & Hedges, and ought to have emerged with a fair piece of change.  Susman Godfrey's massive investigation during the discovery phase of this trial had to have run the clock pretty hard, putting the litigators' financial position at risk.  The acquisition also required anyone inside PSI or closely tied to it to agree to keep mum about pretty much anything to do with the company, its technology, its history, and its legal case.  In addition to its current payroll, PSI might have had to go back to former employees and get them to agree to silence, too.  Sure, there might have been confidentiality agreements in place, but it's a safe bet that IBM, having blundered into an ill-fated lawsuit and having chosen counsel that was unable to vanquish little PSI, would insist on the strongest vows of silence it could possibly obtain.  These agreements might not have come cheap.  PSI's people would certainly have sought legal advice before signing away their precious rights to tell a compelling story.

All this suggests that IBM might have had to fork out more than $60 million.  If IBM's purchase exceeded that amount, or if it could not be structured in a way that the cost of buying PSI's equity came to under $60 million, the next stop on the HSR Threshold Express is $240 million.  Would employment contracts with PSI employees count in the acquisition total under HSR? Most likely not, and that might mean IBM could pay, say, $15 or $20 million to Susman Godfrey, leaving $40 to $45 million for investors while slipping under the $60 million wire.  Still, we doubt IBM got off that cheaply.  Before IBM took PSI to court, there was speculation that Hewlett-Packard would buy PSI and use its mainframe technology to get into some glass houses it could not otherwise enter.  The Register thinks HP was going to pay $200 million for PSI.  That seems like a high price, but perhaps one should think about a mainframe's coattails.  IBM would still get the software revenue if HP sold a PSI box, but HP would get a good shot at a support and services deal and so would another potential customer for the machines, EDS.  Until IBM sued, the PSI situation had a lot of glitter.  And, if PSI didn't sell out and managed to beat IBM in court, the glitter might turn out to be gold.

So it is entirely possible that IBM could have bought PSI for as much as $240 million, or, with the help of hard working accountants and lawyers devising a legal structure that shielded aspects of the deal from HSR, for a sum between $60 million and $240 million.

— Hesh Wiener July 2008


Copyright © 1990-2017 Technology News of America Co Inc.  All rights reserved.