A lot of people were surprised when

IBM

(IBM) - Get Report

made three rapid-fire software acquisitions in August. They shouldn't have been. Big Blue's software group has been averaging 8 to 11 takeovers a year and has become IBM's most profitable business unit.

In 2005, the software group posted sales of $15.8 billion, or about 16% of the company's total revenue, but 37% of its profit. If it stood alone, it would be the world's second-largest software company behind

Microsoft

(MSFT) - Get Report

. It has a workforce of about 44,000, including 26,000 software developers and a sales force of 13,000.

Top-line growth is a major priority for Senior Vice President Steve Mills, who heads the group. Last year, the software business grew by 4%; going forward, Mills expects to grow revenue by 6% to 9% a year, with new acquisitions adding 2% or 3%. Significantly, 2005 marked the first year in which the newer, faster growing parts of the software business, including WebSphere, Lotus, Tivoli and Rational, contributed the bulk of the group's revenue -- 52%.

Mills' performance has won him lots of notice, but he's in no position to relax. The software group has been a standout at a time when Big Blue's growth has slowed; the company's stock is well below its two-year high of just under $97. He needs to keep that machine humming.

To watch Farnoosh Torabi's interview with Steve Mills about IBM's software, please click here

.

Mills talked with Senior Writer Bill Snyder on Aug. 24. Here's an excerpt from that conversation:

TheStreet.com:

IBM's software group has performed very well on your watch -- but IBM's stock has not. Would shareholders be better served by spinning it off?

Steve Mills

: This topic has come up on and off over the years. When Lou Gerstner came into IBM, his conclusion was that he would keep IBM together. The perspective was that from a value-creation perspective, both for shareholders and customers, IBM was uniquely positioned to meet market needs and satisfy requirements across the spectrum of information technology. People are looking for systems, not just products, and that's even more true in this decade.

Does that mean "no?"

That means no.

IBM has spent more on acquisitions this month August than in all of last year. What's the rush?

We did 11 acquisitions last year; we've been doing 8 to 11 a year for the last few years. Since the year 2000, we've done 41 of them. A number of them came together in the month of August because of how the negotiations worked. It was more of a circumstantial thing than anything we had planned.

Is it time to slow down?

No. We're on track to do the kind of numbers I mentioned

8 to 11. It's part of our business model.

At $1.3 billion, Internet Security Systems (ISSX) was a bit larger than many of your acquisitions. Do you expect to make more large buys?

TheStreet Recommends

We've made other billion-plus acquisitions in the past.

FileNet

was $1.6 billion, and earlier on there were the purchases of Rational and Lotus. I wouldn't rule out more multibillion acquisitions.

What do you look for in an acquisition?

We're looking at, how do we get leverage for our shareholders? First of all, we are buying for cash, and we think about delivering a return on that capital. Can we accelerate the revenue growth of the company we buy; does the technology fit well with the other technologies we are delivering so we get a lift in the other parts of my portfolio from the acquisitions? I'm seeking to make these things pay back as rapidly as possible. We've done deals that have been accretive in under a year. I'm always, always driven by both the top and bottom line.

Oracle (ORCL) - Get Report has been buying companies whose heavy streams of ongoing maintenance and service revenue flow to the bottom line. This is different than your strategy, isn't it?

Yes. We are trying to drive incremental revenue growth and accelerate the revenue of the company we buy. In many instances, it doubles. Contrast that to what Oracle did with the acquisitions of PeopleSoft, Siebel and so on. Many analysts will tell you that if you looked at the run-rates of those companies before they were bought out, and then see where they are today, you'll see that their revenue has actually dropped.

Oracle's approach is a derivative of the Charles Wang approach. He perfected that technique at Computer Associates

now

CA

(CA) - Get Report

, and he was able to demonstrate some very good margins to Wall Street.

How do you drive margins in acquired companies if you, unlike Oracle, don't make very heavy cuts in personnel?

We've been very focused on code-sharing and architectural consistency, though we certainly work on eliminating redundancies.

IBM exited the applications business a number of years ago, and you've been partnering with applications companies. But now it's getting harder to distinguish applications from middleware. Is your strategy of staying out of the applications space changing?

The dividing line between infrastructure and applications has never been a clear one, and if anything, it is becoming less clear as a result of things like service-oriented architecture. And by the way, we never said we would be operating in some sort of highly divided mode. We are out to grow our business.

However, we know that having an ecosystem that surrounds our business is important to our success. We are leaving room and not trying to take away the revenue of our business partners.

You mentioned service-oriented architecture, or SOA. How would you define it, and how does it relate to IBM's acquisition strategy?

Within an SOA, your applications act as services and can be mixed and matched to meet the needs of your business process. In practice, customers have acquired lots of applications that have proved redundant. SOA is about simplification. How do I eliminate the redundancies around applications I have bought or built?

We have put our money into the technologies that support SOA. So you see us do acquisitions like

Webify

, which helps customers identify reusable application components and helps them understand how they can logically be connected together.

Let's end by talking about Steve Mills for a moment. You're already running the equivalent of a Fortune 500 company. Isn't CEO, either at IBM, or elsewhere, the logical next career move?

That's not what I get out of bed thinking about. I'm a 33-year IBM vet, and since the late '80s, I've been involved in building the software business. When I came there were about 4,000 of us; today there are about 44,000.

Nobody has more fun than I do. I love to build things. My hobby is home reconstruction and building. IBM provides me with a great opportunity to build things every day. I wouldn't have that opportunity anywhere else.