A spinout's odyssey (Corporate Dealmaker)

A spinout's odyssey

[ Share ] [ E-mail ] [ Leave a Comment ]

This was supposed to be the breakout year for Platform Solutions Inc. After spinning out of Fujitsu Ltd.-owned Amdahl Corp. in 1999, PSI had an impressive syndicate of investors and alliance partners behind it, a product that was generating buzz and a management team with decades of industry experience. The company was finally ready to aggressively roll out its products.

Instead, PSI is tangled in a nasty legal dispute with IBM Corp. that threatens its very survival. You see, not only is IBM the company's primary competitor, it also grants mission-critical licenses that are central to PSI's technology. The company's mainframe platform allows customers to simultaneously run multiple operating systems, including IBM's System z (z/OS) and OS/390, Linux, Unix and, significantly, Windows, which IBM mainframes can't run. PSI's products, according to the company, are priced 15% to 30% lower than IBM systems.

In the lawsuit filed last November, IBM charges PSI with patent infringement and breach of contract over PSI's license for IBM's z/OS. It asks the court to terminate the license and bar PSI from using IBM's patented technology in its products. In the meantime, IBM has said it will not license its software for use on PSI machines, seriously limiting PSI's sales effort. With its life on the line, it's no surprise that PSI has hit back hard, filing a countersuit that charges IBM with antitrust violations, unfair competition and business torts.

"Technology companies are strange bedfellows," says PSI chief executive Michael Maulick, whose résumé includes 22 years at IBM. "All these large corporations have cross-licenses with each other because they want to stay out of the kind of litigation we're in. We wanted to do a cross-license with IBM from day one and say, 'Why fight? You know we're going to be here. Why not do what you do with HP, NEC, Hitachi, Fujitsu, Intel and every other major company?' Even though it was IBM's standard practice, they refused to do that with us."

PSI's journey from an orphaned project to a spinout to a contender legitimate enough to stir a giant such as IBM is worth exploring for more than just the David-and-Goliath nature of the legal battle. With its ties to legendary computer architect Gene Amdahl, its near-death experience inside Fujitsu and the many fateful decisions that kept it alive and growing through some very thin years, the company is a case study in the ways that alliances and networking can nurture a company. PSI's financial and operational partners include Intel Corp. and Hewlett-Packard Co.

At the same time, PSI offers a reminder of what a thorny issue intellectual property rights can become in an industry where a single product might be underpinned by technology from multiple companies. Indeed, the IP issues in play here probably create some complexities even within Big Blue itself. Though IBM wouldn't comment for this story, it appears its software division stands to gain, should PSI's mainframes do well, because any user will have to have an IBM software license. IBM's hardware division, conversely, could lose sales to PSI.

It's not as though PSI wasn't conscious of the potential pushback from IBM, particularly if its mainframe systems showed promise. From the initial spinout from Fujitsu to negotiating licensing agreements, establishing clear lines of separation between proprietary IBM intellectual property and PSI's technology was a priority. Beginning in 2001, PSI began requesting the software and patent licenses it said it needed from IBM to do business. Repeatedly, those requests were denied, and IBM began making requests of its own to examine PSI's products for possible infringement.

The mainframe business may lack the profile it enjoyed in the years before desktop computers, but it's still important and, for those who can play, attractive. It's a mature, multibillion dollar market with high barriers to entry. Machines can run into the millions of dollars, and product reliability is a must. IBM is estimated to own about 85% of the market, and nearly every executive, analyst or industry consultant has some ties back to Big Blue. Members of PSI's executive team have nearly 90 years of IBM experience among them; others grew up at Amdahl. That's a big part of why PSI even has a chance of succeeding in the mainframe business.

The roots of PSI's technology reach deep into IBM, where Gene Amdahl made a name for himself in the 1960s as chief developer of the IBM System/360 family architecture. Amdahl left IBM in 1970 to found Amdahl Corp., with significant financial backing from Fujitsu. The company's plug-compatible mainframes, so called because users could unplug an IBM machine and replace it with an Amdahl mainframe, were smaller, faster and less-expensive than IBM's big iron.

Over the years, Amdahl and IBM battled for customers and underwent major restructurings. IBM began moving aggressively into software and services under turnaround CEO Lou Gerstner in the early 1990s, but didn't forsake the mainframe franchise that once defined the company. Amdahl became a wholly owned subsidiary of Fujitsu, which bought the rest of the company for $850 million in 1997 and was an important supplier of components, chips and subassemblies for Amdahl processors.

A huge financial gamble for IBM was about to pay off, though, and it would force Amdahl and Hitachi, IBM's biggest competitors, out of the mainframe business. In 1999, IBM introduced its z9 Series 64-bit architecture mainframe. The z900 made the 31-bit mainframes manufactured by Amdahl and Hitachi obsolete. IBM had invested an estimated $1 billion over three years on the upgrade, while Fujitsu and Hitachi simply weren't prepared to undertake the significant costs to re-engineer their own products. Had they decided to pursue an upgrade once the z900 was introduced, it would have taken years to accomplish.

Even before Fujitsu abandoned the mainframe business completely, it had been migrating engineering functions from Amdahl's Sunnyvale, Calif., labs to Japan, decimating the Amdahl work force. The restructuring prompted Ron Hilton -- by 1999 the last remaining member of the Amdahl development staff -- to approach management about spinning out the intellectual property associated with an Intel development project launched in 1995. That project laid the groundwork for the Intel-powered mainframe systems that PSI sells today.

"The complicating factor from our perspective was the transfer of technology in a way that didn't violate the restrictions on certain types of transfers that contractually existed between Amdahl and IBM," says Greg Handschuh, who at the time was general counsel at Amdahl, and today serves in the same role at PSI. "Once we were satisfied we could do that, we went ahead with the transaction. I never liked these deals, to tell you the truth. My view had always been even though it's technology that's no longer going to be exploited for one reason or another, I never particularly liked seeing it go outside the company because you lose control of it in one fashion or another."

The spinout ultimately occurred in two steps over three years. In 1999, Hilton was granted a nonexclusive, perpetual license to the IP as it existed at that time. In return, Fujitsu received a 35% stake in PSI and a seat on the board. Hilton also received a family of diagnostic and testing tools critical to the development of the Intel-powered mainframe.

"When you're dealing with proprietary architecture like we are in the mainframe business," Maulick says, "the ability to validate that what you're developing conforms to that architecture is as much a part of the development effort as creating the product itself. If we weren't able to inherit the base development tools and test equipment and programs from Amdahl, we would have had to develop those all on our own." Maulick says the work could have cost more than $400 million.

Between 1999 and 2002, Hilton and his team worked in Amdahl-owned labs in California developing the open-system mainframe platform. Still, PSI was far from on solid footing. The founders were bootstrapping the company, there was no experienced management team and no comprehensive business plan. The company was so low on cash at one point in 2002 that it only survived by licensing back to Fujitsu the technology as it had been enhanced. The transaction led to a cash infusion of $500,000, enough to keep PSI in business.

"We didn't expect a positive response," says Handschuh, "but Fujitsu has a near-compatible mainframe offering that they sold almost exclusively in the Japanese market. The technology that Ron was developing had potential applications for that effort. There was a field-of-use limitation on the license strictly to Fujitsu's noncompatible, near-compatible offering."

About this time, a former Amdahl controller contacted George Hoyem, managing director of Blueprint Ventures, with a proposition: Would Blueprint, a VC firm specializing in IT infrastructure startups, be interested in working with a fledgling company trying to break into the capital-intensive, highly specialized mainframe business?

"My first reaction was to laugh," Hoyem says, "but when I heard it was a spinout from Amdahl, it pricked my interest."

Conventional wisdom at the time was that the mainframe business was declining, as in fact it was in a variety of segments, says Hoyem. But in others, such as transaction processing and airline reservations, it was steady if not growing, and according to Blueprint's research, those customers were hungry for an alternative to IBM. It was enough for Blueprint to take on PSI as a client and do what it could to turn it into a company that could attract venture funding.

The first steps were to find a CEO and to "clean up the IP structure," says Hoyem. "For instance, the licensing agreement that existed between PSI and Fujitsu was not robust enough to make us comfortable. We wanted to enhance the source code rights for the tools. There were also some issues around technology that had been developed in conjunction with IBM, and we wanted to clean up the code and the technology to make sure there was no pollution of that from proprietary IBM information. We wanted to make sure we had a process to cleanse the IP from anything that had residual linkages back to agreements that Amdahl had with IBM."

While recruiting experienced senior-level executives to a startup with no customers can be difficult, getting Maulick on board was an easy sell. "When we sat down with Michael, we said we're starting this company, and it's based on all the bones of the software remains of Amdahl," Hoyem says. "And by the way we have all these tools, source code, technology, people and know-how, and we're going to take the hill. He was like, 'That's unbelievable.' "

Maulick had spent the previous seven years in CEO positions with various startups, but before that he held several executive positions during 22 years at IBM, including as a senior engineering manager heading up mainframe design and development. His knowledge of the industry and his relationships with IBM business managers proved invaluable.

"The one key thing that had to be negotiated, which I did before I [officially] came on board, was a software license for IBM's operating system to run on our hardware," says Maulick. "If we couldn't get that, we couldn't have a business."

Before Maulick's arrival, Hilton, who was then CEO, was interacting with IBM's legal department. "IBM saw PSI as a competitor and said, 'No.' I basically contacted my old friends and laid out the business proposition," says Maulick. "We talked about it as a business entity with IBM knowing full well we'd be coming into the marketplace. ... I said, 'You're getting money for the license. Why wouldn't you do this?' It was standard practice for them to do that with Amdahl and Hitachi. ... If I were to speculate, I'd think that probably their viewpoint was that we would never succeed because they knew how tough it was to develop a product in this space."

It took multiple face-to-face meetings over a three-month period, but IBM ultimately granted PSI a license for its OS/390. At the same time, PSI engaged HP as its platform partner, replacing Unisys Corp. "HP was building bigger machines with the capabilities we needed to tie together with the IBM operating system to produce a product with the reliability, availability and serviceability we needed," says Maulick.

"We didn't want HP as an investor," adds Hoyem, "because they were a potential exit candidate for the company. Intel [an investor] would never buy this company, but you don't want people to invest in a company at the VC level who might buy the company because you don't want them to get their nose under the tent."

There has, in fact, been speculation that HP and PSI may have been talking about a possible acquisition when IBM filed its lawsuit, though PSI says it can't comment on any talks because of confidentiality agreements.

The IBM license and the HP partnership were critical to PSI's ability to land venture funding. The company closed its first round, $10 million from Blueprint Ventures, Fujitsu, InterWest Partners, InvestCorp and Intel Capital, in December 2003. Bringing in Intel Capital as a strategic investor helped deepen the existing technology relationship between PSI and Intel Corp. The company closed a Series B round led by Goldman, Sachs & Co. nine months later for an undisclosed amount. Eventually, PSI bought out Fujitsu's 35% stake in the company.

Once the funding and key strategic relationships were firm, Hoyem and Maulick turned to building out the management team. They did so with a long list of IBM veterans including executive vice president Linda Zider, technical adviser Carl Caricari and strategic adviser-sales Bill O'Connell. The executive team was rounded out by former Amdahl employees Hilton, who is now chief technology officer, and Handschuh.

Of course, the licenses, alliances and impressive management team won't matter if IBM successfully shuts PSI down. And even if PSI survives, there's no guarantee that customers will buy its product.

"There are some unique things," says Gartner Inc. analyst John Phelps, who himself spent 30 years at IBM. "It can run Windows, that's an advantage. But they've kept it very close to the vest, so I don't know what kind of software pricing they'll have. Hardware will be expensive, but it won't be as expensive as [IBM's] proprietary chips. All the other costs should be similar."

The lawsuit has forced PSI to rethink its rollout timeline and to be more conservative in its new customer goals. "The lawsuit is a point of conversation, it's an objection to overcome," Zider says. "We have some customers who were really pretty close who chose for all kinds of reasons to wait and see as opposed to jumping on board. Our job just got harder in trying to find those who want to jump on board."

Because IBM is refusing to grant user licenses for use on PSI machines, PSI is targeting its sales effort toward customers with IBM mainframes because their existing licenses would transfer from the old machines to the new. Still, the company has "a couple of handfuls of customers," says Zider, and the hope is to double that number each quarter through 2007 -- a respectable goal in the high-margin mainframe business. It helps that PSI has a formidable web of alliance partners. Both Intel and HP have a vested interest in seeing PSI succeed.

And so, of course, does Maulick. "When people ask why, with a 22-year history at IBM, I would compete against IBM," he says, "I tell them the same thing I told IBM's CEO: 'You are underserving a customer set that used to be my customer set. Part of doing business over time is being loyal to your customers, and I feel you're letting down certain customers, including your largest customers, by not having an alternative. They are begging me to bring something back, and I want to serve them.' "

The parties are scheduled to meet for the first time on May 4 in U.S. District Court in the Southern District of New York. It could take months or even years, however, for the suits and countersuits filed by IBM and PSI to wend their to a resolution. It's not clear just how long PSI can hang on with a hamstrung sales effort, but the PSI executives seem determined to fight. - Suzanne Stevens

PLATFORM SOLUTIONS INC.

The world's oldest startup

PSI's roots trace back to the company founded by Gene Amdahl and later acquired by Fujitsu. When Fujitsu left the mainframe market, PSI was spun out.

Date	Action

1964	IBM introduces the System/360. Created by famed computer architect Gene Amdahl, the S/360 was one of the first mainframe computers.

1970	Amdahl leaves IBM after his computer development ideas were rejected. He launches Amdahl Corp. a few weeks later.

1971	Fujitsu Ltd takes a 42% equity stake in Amdahl.

1975	Amdahl Corp. ships its first plug-compatible mainframe. The Amdahl 470 V6 was a faster, less-expensive, smaller alternative to IBM's mainframe that allowed customers to run IBM software without purchasing Big Blue's big iron.


1980	Gene Amdahl resigns from Amdahl's board.

1984	Fujitsu increases its stake in Amdahl to 49%.

1993	Collapse of mainframe market sends Amdahl stock tumbling 33%.

1995	Amdahl launches a technology development alliance with Intel.

1996	Amdahl's share of the mainframe market drops to approximately 7%, down from 12% a year earlier.

1997	Fujitsu buys Amdahl for $850 million. IBM commits to a major R&D effort to upgrade its 31-bit architecture mainframe to a 64-bit architecture.

1999	Platform Solutions Inc. is born after engineer Ron Hilton receives a license from Fujitsu to the IP developed through the Intel alliance. Fujitsu receives a 35% stake in PSI.

2000	IBM introduces its z9 Series 64-bit mainframe. Fujitsu and Hitachi Corp choose to exit the mainframe business rather than upgrade their own technology.

2002	PSI receives $500,000 by licensing back to Fujitsu the enhanced open-system platform technology. BluePrint Ventures signs on as an investor and strategic adviser to PSI.

2003	IBM veteran Michael Maulick signs on as CEO and helps secure a critical license for IBM's operating system. DEC.: PSI closes a Series A, $10 million venture round. Hewlett-Packard becomes a platform partner.

2004	PSI buys out Fujitsu's 35% stake in the company. AUG.: Hilton presents PSI's technology at the SHARE conference, an annual gathering of IBM users.

2005	JAN.: PSI ships its first open-system platform to Beta Test Partner L.L. Bean. OCT.: Intel delays launch of the Dual Core Itanium 2 chip, impacting OEM delivery by nine months. This creates a one-year product-launch delay for PSI.

2006	NOV. 26: IBM sues PSI over patent infringement and breach of contract. NOV.-DEC.: PSI receives its first four purchase orders for Open Mainframe Servers.

2007	JAN. 19: PSI countersues IBM for antitrust violations, unfair competition and business torts.

Source: Corporate Dealmaker