Sweeping Transformation

Besides the move away from IT outsourcing, the key elements of Mott's plan include:

>> Data center consolidation: GM plans to go from 23 sizable data centers worldwide to just two, both in Michigan. In the process, GM will replace servers, storage, and networking with today's more efficient gear, in hopes of reducing costs and using more automation so that fewer of GM's IT pros are involved in "run the business" kinds of IT operations and more are involved in new development and innovation.

>> Application consolidation: Mott estimates that GM can cut 40% or more of the company's 4,000-plus applications that have sprung up in various regions and divisions by moving to standardized, global applications for everything from financials to factory-level processes.

>> Hiring: Mott plans to dramatically increase the IT organization's hiring of new college grads and will look to top computer science schools beyond the universities with which it now has recruiting ties. Most of GM's development will be done in the U.S., even though the company's growth is very much tied to China, Brazil, and other countries. Mott does plan to hire internationally to bulk up an "IT planning community" that can gather the requirements needed for GM to build global apps.

>> Three new software development centers: One will be in the Detroit area, with the other two in U.S. locations still to be determined. They'll be chosen based on the kind of development talent the company can attract there, including their proximity to key universities. (Silicon Valley, anyone?) Dispersing IT development will be a cultural change for the still Detroit-centric GM.

>> Data warehouse consolidation: GM has about 200 data marts today, and Mott wants to move all of them to one architecture. The goal is to make data easier to access and use, and to integrate it better across groups and divisions to reveal more insights. CEO Akerson told Fortune in a May interview, "We're looking at our IT systems real hard. Data warehousing of customer information was not a strength in the company. Maybe it wasn't even existent, but it will be in the future."

Mott's ambitious transformation plan won't be universally popular. End users will complain: You're taking away the application that I love for the greater good of cost savings? Business unit managers will complain: You're making me do a cost-benefit analysis for my little upgrade? Isn't this the kind of bureaucracy a nimble GM needs to avoid? Even Mott's IT team will grumble: You're messing with the fiefdom of IT contractors I manage?

Mott's critics at HP complained that he cut too fast and deep, leaving business units vulnerable, or that he spent precious resources on internal bean counting.

There's one additional, controversial piece of Mott's plan: It all happens at once, over the next three years. Mott described it in the past as "choosing is losing." For example, if you move to an insourced IT staff and away from outsourcers, but you don't modernize the data center to increase IT automation, then you won't get the benefit from those new staffers working on new projects instead of maintenance and support work.

It fits Mott's general philosophy that many IT projects fail because they just take too long. It's why he favors putting more staff on a project for a shorter period--doing fewer projects at a time, but faster. "Every month that goes by, every quarter that goes by, the ROI goes down," he says.

The clock has started ticking. Mott launched this IT transformation plan internally in late June on a town hall webcast with GM's 1,500 IT employees worldwide. The world will be watching: If Mott and his team botch any of this IT transformation, it could put the company's historic turnaround in jeopardy.