Oracle, led by CEO Larry Ellison, is now selling a vision of integrated systems combining software and hardware, a strategy that many have said simply mimics what IBM did long ago. What do you think of what Oracle is doing?
Steve Mills: Oracle's got a very simple strategy, and that's make Larry rich. It's the reason the company was founded, it's its core purpose. As the biggest shareholder, it delivers value to Larry. It makes sense to Wall Street.
They're not a customer-friendly company. They want the customer's money and they're very creative with coming up with different ways to get the customer's money. They cannibalize their own ecosystem, they compete with their partners. It's a very pragmatic strategy. Whatever it takes to get money, that's what they're going to work on. If they think they can get money putting things together, they'll put them together. If they can only get so much money out of that, they'll take them apart and do it another way.
It's not driven by a strategy, per se, on some vision of where information technology is going. It's driven by the money-making motivation. All public companies have a motivation to make money. It's not a matter of, they're the only ones that think about making money. But some companies are more aggressive in their singular pursuit of that goal than others.
Oracle does a lot of this at the expense of their clients, which makes them my favorite company to compete against. They use schoolyard bully tactics, they make a lot of claims they can't back up, they tend to back away from benchmarks and any real-world evaluation of their technology.
We find they're a great opportunity creator. They put a lot of silly statements out that are easy to challenge. Customers don't like negotiating with them. They're difficult to get along with. They anger clients.
What is your opinion of Oracle's decision to stop developing software that supports Intel's Itanium chips, which Hewlett-Packard uses in server systems?
Mills: [The decision] essentially put a sharp stick in HP's eye. You look at that from a customer perspective and you say, Oracle may hate HP, but by the way, they just put a sharp stick in my eye, because I bought a bunch of HP Itanium boxes, I'm running HP-UX and Oracle stuff on top of that and now they're saying they're not going to support that. In effect, it's an attack on the customer. Oracle didn't ask their customers whether ending support on Itanium was something that they would like. There were no customers consulted on that decision. They did it because they wanted to pick a fight with HP.
Where does IBM stand on Itanium support in the long term?
Mills: We support Itanium. We support HP-UX. We also, by the way, try to displace Itanium systems. We're selling our products against HP. We do compete against HP but we also support HP. We've taken that view for a long time, that we're going to have a complex ecosystem. We're committed to [Itanium] as a function of what the customers do with it. If a customer has an IBM product running on a system, we stand behind that and provide long-term support. We have some clients that will keep some things for years and years and years longer than we're actually producing anything new but we still provide them with support. We have customers running things that we delivered in the 1960s.
What about brand-new technologies, such as the Watson supercomputer, which can understand natural language and provide answers to questions? How is IBM planning to commercialize that technology?
Mills: You saw it on Jeopardy as a [single] thing. Under the covers there are a lot of pieces to Watson. We're using Hadoop, we're using IBM DB2. It has a 'big data' underpinning to it. It's able to search and retrieves information, and is able to iterate on pieces of data. So more than just simply being able to find an answer, it's able to relate verified answers to other pieces of information that come in. It builds an ontology. It builds a structure.
As a topic gets introduced into the system and you keep adding more and more data, you keep building out the knowledge base. What that does from a question-and-answer perspective is it makes the system ever more accurate. The system is in fact designed to tell you what its statistical confidence is. That's important when you start to look at commercial uses. The computer is a great tool and a wonderful assist to people, but it doesn't necessarily always replace people.
If your job had been processing checks in a bank, it's not that complicated to do, let the computer process the checks. But what if you think about creating a physician's assistant -- which is one of the things we're doing with Watson -- that can help a doctor deal with systems and treatment. So I observe your particular condition and put information in. Watson then comes back and provides multiple choices as to what is the likely ailment you're suffering from. The system's also built such that it could give you clues on what questions you could ask the patient to refine the symptomatic information.
Unfortunately sometimes in the medical profession, there are inaccuracies. Nobody can know everything. That's why people seek second opinions and go to specialists. So can you lift everybody's knowledge by using a computer? The answer is yes, and that's what Watson was built to do, to provide that augmentation to human knowledge.
We're moving it from the research effort into formal productization. We'll begin to deliver various implementations of Watson later this year and into next year, and you'll see product announcements. We'll do some things with packaging, using a T-shirt size model -- small, medium and large.
While spending on SaaS (software as a service) remains a fairly small percentage of total enterprise software revenue, analysts expect adoption to grow rapidly over the next several years. Where does IBM sit inside that scenario?
Mills: [IBM's SaaS portfolio] represents about $200 million in revenue. We expect that business to get bigger and bigger. But this whole world of cloud and cloud things is hard to calibrate to any degree of precision. Many of the analysts want to view SaaS as a fairly narrow definition. Some have taken a broader view and said there are thousands of business process outsourcing companies that are effectively offering SaaS. Their origins predate the popularity of that term. ADP is a favorite example. They're a cloud company. They don't call them themselves that, but that's what they do. It's your payroll, your payables and receivables. They do that for you using computers and software, and it's the same computers and software for everybody that comes in. It's a SaaS model no different than Salesforce.com.
So when people talk about how it's a nascent market, it only represents X percent today, actually it's a huge market when you measure it the way I just described it. The good news is that we're in that space as a business process execution company. We have something called Global Processing Services that is a range of offerings packaged up that way. They're repeatable, you put your work on our systems and we'll execute that for you.