Virtualization is the kind of red-hot tech market that startups seem to jump into weekly. But the newest entry comes with a twist--it's a spin-out from the huge investment bank Credit Suisse, where the IT team wrote its own virtual machine management software. Such spin-outs are still rare, and they face big obstacles moving from a corporate to entrepreneurial culture, but they may become a more typical source of inspiration for business software.

The IT team at Credit Suisse, a company with more than 48,000 employees, was struggling with the different management tools of each of its virtualization vendors, so it wrote a management console and tools to overlay them. The new company, DynamicOps, will develop and sell Virtual Resource Manager to global companies like itself. It's funded by Credit Suisse's in-house venture capital group, Next II, at an undisclosed amount.

"It was a vendor-agnostic, holistic approach," says Steve Yatko, managing director of Credit Suisse's Global Research and Development Group, who thinks the system will make it easier and cheaper to deploy virtual servers and desktops. But the spin-out costs more than venture capital; Yatko and CIO Karl Landert lose some key IT staffers to the startup, including the lead architect on the project, Leslie Muller, who's now CTO of DynamicOps.

It's not unprecedented but still rare for a user IT development project to spawn an independent startup. But Bart Schachter, managing director of Blueprint Ventures, a venture capital firm that specializes in such spin-outs, says DynamicOps has a few things in its favor. One, it's got the kind of blue-chip customer in Credit Suisse that startups struggle to land. Two, it has an engineering team that's worked together and understands real-world virtualization problems. Three, its software has been proven to meet the needs of a major player in one of the most IT-intensive industries. Given the growing importance of embedding industry and process knowledge in software, such inside information is increasingly important to development. "This is a very interesting path for an independent software company to come from," Schachter says.

So why don't we see more of this? Don't underestimate the gulf between IT department and IT startup, warns M.S. Krishnan, a University of Michigan professor and co-author of the book The New Age Of Innovation (McGraw-Hill, 2008).

When consumer stock trading started moving online, established brokerage houses attempted to match what the online startups were doing. When that didn't work, they thought the efforts could compete if they were just set free of the corporate shackles. Yet most of the online units spun out in the late 1990s didn't compete effectively with the firms that had originated as online companies and had a pure entrepreneurial mind-set, Krishnan says. The online firms had "a different dynamism" and the necessary flexibility, he says, that the spin-outs never could muster.

Likewise, DynamicOps employees will need to think of every would-be customer as a unique individual. "They need to personalize for each customer what they've already done," says Krishnan, rather than believe that what works for Credit Suisse will be good enough for everybody else.

The more common examples are of spin-outs from a tech company that develops something outside its core market, or from a university or government research effort. Blueprint Ventures helped the federal government's Jet Propulsion Lab spin out SpectraSensors, which makes a laser-based sensor that's currently on the Mars rover looking for water, but whose more marketable use is looking for leaks in oil and gas pipelines. Schachter hopes to take it public in 2009.

Few technology user companies have this kind of product development as much a part of their DNA as the University of Pittsburgh Medical Center. The health care provider and group health plan insurance company last week cut the latest in a series of deals to create startup tech companies, an effort led in part by CIO Dan Drawbaugh, InformationWeek's 2006 Chief of the Year.

The University of Pittsburgh Medical Center and GE Healthcare are each putting up $20 million to form Omnyx, a joint venture to make advanced digital imaging systems for pathologists, replacing glass slides with digital images. It's an effort that's been hampered by the slow loading of such images over a network. "Both organizations are tech savvy--one with best-in-class clinical experience, and the other with technology expertise globally," says Drawbaugh, one of the two medical center officials on Omnyx's four-member board. "It was a natural marriage."

Such partnerships are an explicit part of Drawbaugh's job. In 2005, when the University of Pittsburgh Medical Center inked an infrastructure contract with IBM, each company also committed $25 million to joint development of software and systems to tackle health care IT problems, using UPMC as something of a test bed. UPMC did the same twice in 2006: when picking Alcatel to upgrade its IP network, each committed $25 million to new product development; and when picking dbMotion for middleware, it took an equity stake and launched co-development projects. Stentor, a company that UPMC created in the late 1990s to sell medical digital imaging products based on technology it developed, was acquired by Philips Medical Systems for $280 million in 2005.

As software is delivered more often as a service, end-user companies may have more opportunities to enter new markets for IT-enabled services, and even spin them out as independent businesses. That's what American Airlines eventually did with its Sabre airline reservation system. It's a model that's flourishing in the financial services industry; insurance company Security Benefit, with $35 billion under management, offers business process outsourcing and digital imaging services mainly to other financial services companies, devoting more than half its processing capacity to the customers.