IBM Corp. CEO Samuel J. Palmisano last year predicted that his company would spend $20 billion on acquisitions over the next five years, with a focus on software companies. Looking at the Armonk, N.Y.-based technology giant's record in doing software deals, the safe money would be on the IBM chief's being right.

The company has emerged as one of the top acquirers of smaller software companies, buying more than 65 in the past seven years. This has substantially changed IBM's mix of business offerings to the extent that software now generates 42% of IBM's profits, and is expected to compose half of them by 2015.

The company that invented the hard drive in 1956 and whose name was once synonymous with hardware is now the second-largest software company (trailing Microsoft Corp.) measured by revenue. It generated $22 billion last year.

An area of particular interest for IBM has been in business analytics, a realm of software that helps customers take reams of data gathered from past performance and use it to help make strategic planning decisions. During the past four years, IBM has sunk more than $14 billion into 24 analytics acquisitions, targeting companies including Coremetrics Inc. and Unica Corp. In one of the largest deals in this area, IBM late last year acquired Netezza Corp. for $1.7 billion. The deal provided IBM with a computer appliance that is said to run high-speed, high-performance business analytics queries far faster than comparable technology, and, perhaps more importantly, is relatively simple to install and use.

Watching over this flurry of deals is Steve Mills, senior vice president and group executive of IBM Software & Systems. The 59-year-old, who joined IBM right out of college in 1973, took the reins of the software group in 2000 and last year added responsibility for IBM's hardware group as well.

We recently spoke with Mills about IBM's interest in growing its analytics offerings, its overall M&A strategy and Oracle Corp. CEO Larry Ellison.

The Deal: Some of IBM's biggest acquisitions this year have been in business analytics. How has analytics evolved into a technology area of such interest, and why is it such a big revenue driver for IBM?

Steve Mills: We spend a lot of time on what our customers want, where they want to spend their money and where they will get incremental value from IT. Businesses have been using computers and software for capturing and organizing data and generating reports and doing various analytic routines to unearth interesting information and patterns.

They put a considerable amount of money into what is broadly defined as business analytics. It's about doing a better job of getting one version of the truth within their company.

One of the reasons we acquired Netezza, we recognized that customers have built hundreds of thousands of data marts and data warehouses, not necessarily having procured the data consistently. Netezza has a very nice model for a prepackaged data warehouse that improves consistency. We saw a lot of power in that.

Master data management is an important area. You can't get to one version of the truth if you can't reconcile data sources. Banks struggle with this issue. My full name is Steven Alan Mills, but what if a bank spells Steven with a "ph"? If you are the CEO of that bank, you can't fully capitalize on the relationship you have with me. Many financial institutions these days are working on these problems.

Other things are fueling data enthusiasm. Clearly, there's a desire to become more predictive. I want to know what to sell and when to sell it and where. Predictive analytics are taking on more importance. Financial institutions also need to see asset risk and market risk and try to be more predictive.

There's also an increasing focus on real time, meaning "I want to collect data in the moment, up to the minute." Algorithmic trading is a business analytics process, for example.

Every industry has examples of sophisticated business analytics, and they need to store more data and have tools to organize and structure it. Interesting things are emerging as a result of the new connectivity infrastructure. Public sentiment can be measured and captured almost in real time. If you're a retailer or consumer packaged-goods company, you think a lot about what people think about a product you are putting out or planning to put out.

There are huge quantities of information that often come to you unstructured. Tweets are unstructured, but I need to understand what is in a tweet and apply value to the adverbs and adjectives and nouns and create a rating structure on relative sentiment on any particular topic.

Why has IBM chosen to acquire analytics technology instead of developing it internally?

We are the company that invented the disk drive and the database. We run the world's largest private enterprise math department at IBM Research. But for as much as we develop, we can't seem to get it all fast enough.

At what point are you brought into acquisition discussions, or does it depend on the size of the transaction?

I'm involved in 100% of product acquisitions in terms of the process in which we analyze what we might buy.

It's a disciplined approach that begins with market requirements, structure, where there are missing elements. You either make, buy or partner. You can't make everything, but you can't buy everything. I'm involved in all these decisions.

The parameters of an acquisition and the level we are willing to spend follows a very clear set of guidelines for return on investment of capital vis-à-vis an alternate use of cash. It is a shareholder asset.

I worry more about integrations and how we will leverage an acquisition. The easy part of buying companies is buying; the hard part is creating a value equation better than what the company could have done on its own.

Is it accurate to say that IBM only acquires software companies with which it has some kind of history or relationship?

It's pretty close to being accurate. We acquire companies and then we invest in them. Our returns are determined by synergy or lift, not by creating more efficient ways to run that business by ripping parts out. We are not going to acquire a company that we don't have some degree of familiarity with.

In all cases, we feel we have a pretty deep insight into a company. We determine compatibility and a way we will fit the business in.

Oracle CEO Larry Ellison often refers to IBM as his company's top competitor. What's your take on Oracle now that it has absorbed Sun Microsystems? Is its ability to offer a full stack of software and hardware creating a difference in the marketplace?

It's wonderful rhetoric, but customers don't want to give any vendor pricing power. Larry refers to past eras in the computing industry. Customers today want a forward view that embraces open standards, modularity, freedom of choice. Relationships with IBM are fundamentally different than Oracle. Oracle wants to maintain as much control as possible.

Customers want better time-to-value. Nothing Oracle delivers is fast to deliver. It takes weeks to install an Exadata box [Oracle's database server product].

This "all-in-one" idea that shows up in Larry's ideas and ads in reality requires a lot of labor. We are a primary provider of that labor to that market; we have a huge services business that revolves around the technologies that Oracle creates.

It's wonderful chest beating. Oracle is a high-testosterone business and very declarative in everything they say. They always win, we always lose, they're the best.

What is the single biggest challenge IBM faces in doing acquisitions, and if there was something you could improve, what would it be?

We work on continuous improvement. For each acquisition, we can look at the process of bringing it into IBM, levering its creativity, taking advantage of customer relationships, and find examples of what we didn't do as well in some places as others.

We have wonderful examples of doing things remarkably well, but when you multiply that by as many countries we are in with the people we have, you will see some areas where we may have stumbled, and we learn from those experiences.

Are you collaborating as an organization on a global basis, are you getting feedback telling you where you aren't doing things well? Typically, this might happen in a small country: Why did we not do it correctly in Romania when we did it so well in Germany? Maybe there were only a few people there watching, but that's not acceptable.

You added oversight of IBM's hardware operations to your responsibilities last year. You weren't busy enough?

I had nothing else to do! I've always been very involved in hardware; IBM has always had strong focus on the development of hardware and software together. We try to build it in a way that matches the way customers use it.

Sam Palmisano asked me to take this one, recognizing that, looking ahead, we need to continue enhancing hardware design. Customers don't see [hardware and software] as separate technologies. I'm acting as the cheerleader coach to put more things together faster.