Feisty Ballmer Talks Cloud, Office 365, And Big Data

Microsoft’s cloud computing business is on a “steep ascent,” in the words of CEO Steve Ballmer, but what does that mean in terms of revenue, profit, per-seat economics, and rate of growth?

Quantifying its cloud business -- providing hard numbers on specific services -- isn’t something that Microsoft is ready to do. Ballmer says “millions of seats” have moved to Business Productivity Online Services, which includes software-as-a-service versions of Exchange and SharePoint, and the company’s other SaaS offerings, but that’s as far as he would go. “I don’t know whether we’d give you something more specific or not,” Ballmer said in a June 9 interview with InformationWeek editors at the company’s Redmond, Wash., headquarters.

Microsoft’s progress in the cloud, including the pending launch of Office 365, which will replace BPOS, was top of mind for the InformationWeek team (Rob Preston, Doug Henschen, Paul McDougall, and me) on this visit. Just 14 months ago, Ballmer told us that Microsoft’s cloud business was on the cusp of “hockey stick” growth. We’re monitoring the pace of adoption closely because of the massive change it represents not just for Microsoft, but for ITS business customers.

Ballmer was in a recalcitrant mood -- parrying some of our questions, calling others “unanswerable.” But he was also forthcoming, providing new insights into the company’s Windows Azure business (it’s growing slower than expected), its hardware strategy (it wants more control over the hardware), the secret to making SaaS pay off (minimize churn), and more.

Following is a lightly edited transcript of our conversation.

We started by asking Ballmer about his forecast last year that “hockey stick” growth was ahead for Microsoft’s cloud business, which includes BPOS, Windows Azure, and the soon-to-launch Office 365.

Ballmer: We've got a lot of room still to go, but we've definitely seen a hockey stick in terms of adoption of Business Productivity Online Services, soon to be Office 365. Literally rising to millions of seats effectively in a very short time. In that case, we have seen absolutely good liftoff, hockey stick-like liftoff.

In terms of Azure, we've seen good acceptance. But if you were to say, What percentage of enterprise computing is likely to move? Mission critical apps as a percentage will move to the cloud more slowly than information worker infrastructure. We’ll see a different pace and cadence.

Now, we're a big company; the numbers are all big. Every growth in Office 365 is something else that somebody didn't buy. You can look at our financial results by segment or the whole company, and you may not say, yes, I got it, I see it, it's right there. On the other hand, there is seat explosion, and if you look at the list of reference customers, people have moved, people are moving. We are in a very steep ascent, both in terms of commitments and deployments.

InformationWeek then asked whether there were any surprises in Microsoft’s cloud strategy, any areas where adoption was actually faster than expected.

Ballmer: I would have said, we were expecting faster growth [rather] than we're seeing faster growth. If anything, if you ask a year later about private clouds and public clouds, particularly as it relates to line of business applications, I'm smarter than I was and we're all smarter than we were a year ago.

The private cloud thing was important; it's still important. Is it a private cloud, or is it really just a virtualized data center? There is a conceptual difference between a virtualized data center and a private cloud, yet a lot of what’s getting sold as private clouds are really virtualized data centers. And a lot of what’s being done in the public cloud is the hosting of virtual machines, which, again, is a little bit different. We would not run our big services with the same kind of technology that are running enterprise data centers.

So, the move to next-generation operations and management infrastructure is going to be a little tougher, because there's a lot of legacy, a lot of understanding [required] of current approaches. Things will move a little more slowly, and people aren't willing to commit line of business apps en masse to the public cloud anyway right now. They're willing to commit websites. Startups are willing to move in that direction. But there's not an en masse move to the cloud versus virtualization.

Virtualization is in full flight, of course. The Hyper-V private cloud stuff that we've announced with our System Center Management tools I'm real excited about. But we're smarter 12 months into it than we were when I saw you guys last.

Ballmer and other Microsoft execs continue to talk about cloud growth, but where’s the evidence of it? Microsoft doesn’t break out cloud services in its financial statements, and quantifiable data is scant.

Ballmer: We've got literally many millions of seats that moved to the cloud. Millions of seats -- stop and think about that, that's a big statement, millions of seats from enterprise customers. I'm not talking about small and medium business. I don't know whether we'd give something more specific or not. I'm not close to what we're saying publicly on that, but I'll let Frank Shaw [Microsoft’s VP of corporate communications] get the correct public pronouncement that's not going to cause us a problem with the financial guys who haven't heard it before. But I am really excited about the speed with which that's happening. You'll see it mostly as reference customers as opposed to the numbers.

Whether the number was one million, 5 million, or 10 million it's hard to assess the difference. Across the planet, there's about 1.2 billion computers installed. Roughly a third would be in some kind of managed enterprise environment, say 400 million-ish. Millions is a big number. It's a big number. Next, we tried to determine whether Microsoft stands to make more or less money from customers in the cloud than from those that license its desktop software. We posed the question this way: In terms of revenue, is a seat in the cloud worth the same to you as a seat in the enterprise, is it worth more, or is it worth less?

Ballmer: Well, they are both seats in the enterprise. I'm not following you. The same person who comes to work and wants to use information and collaborate.

So we tried again. Do you make as much money from that seat if it moves to the cloud? It should be less, right, because of the shared model of the cloud? Ballmer refused to answer the question and said that anybody who does is “just BSing you.” He jumped to a whiteboard and sketched out two hypothetical situations, one where the customer made a one-time software transaction, the other where the customer subscribed to Microsoft’s software as a service.

Ballmer: Our value prop on this is cheaper than them doing it themselves, because we take out labor, we take out capital, plus we subsume the software question. You can see our prices on the website for Office 365, and most people have a pretty good guess at what our Office products are like. If this is being renewed all the time, if we're running this as low COGS [cost of goods sold] versus high COGS, it gives you a different view of what profitability looks like. If this customer turns in a month, that's not a very good thing. If this customer turns in a very long time, we're obviously going to be better off, as long as we're running things correctly from a COGS perspective.

So, there's no easy answer. You have different timeframes, different cost structures. There's nothing that's similar about the economic model. The value props are very different. In one case, we're selling software. In one case, we're selling the value of software, labor, and capital. In one case we get our money up front. In the other, the potential for churn before we get there is higher, but the potential for keeping that customer regularly paying us, and working with us, is higher.

That's why it's an unanswerable question. I'm sure when we look back in five years we'll do the post-mortem and we'll say, hmm, we made more money.

We penciled it out. We’ve got these complicated models, big spreadsheets that all say we should make more money if we do it right. But we have to do it right, and there's no good apples and apples comparison.

We wanted to know how Microsoft’s business strategy might change given the expiration in May of the Department of Justice’s Consent Decree, which had imposed restrictions and regulations on the company for the past 10 years. Has the expiration of the Consent Decree changed the way Microsoft now operates?

Ballmer: Operate, no. Do we get to make a set of decisions differently than we might have made before the consent decree, based upon obligations that are no longer there?

There are things that we get to consider, but when you say change the way we work, our people are still going to work every day, they're still transparent and open in our prices with our OEM customers, and the like.

There are certain restrictions that are gone. You can read the decree and say the obligations we had or the things that we were effectively banned from doing are now gone. It doesn't mean we're going to do all of them, but it does mean that there are new approaches that are available to us.

We did have an obligation in the consent decree to have an antitrust compliance committee. We chose to keep it around, not because there's an antitrust issue, but there's a set of issues now that the board meets on and thinks about regularly. We may have renamed the committee, but it does consider a certain class of issues on behalf of the board, and we decided that's worth preserving.

Among Microsoft’s big successes in the past year have been Xbox and Kinect. Microsoft exercises tight control over the design and manufacture of its gaming platforms, in contrast to its approach with Windows PCs, which are available from a wide variety of manufacturers. Does Ballmer take anything away from that experience that might apply in other product areas?

Ballmer: Yes. We're smarter about that. I don't think we want the seam between hardware and software be an innovation barrier for us. It doesn't mean we have to make any particular class of device, but thinking through, end-to-end, how the hardware and software come together, as opposed to artificially boxing the way we think about systems design -- the Xbox and Kinect are good reminders of that.

Perhaps an even better reminder is the work that we've done on so-called chassis specifications for Windows Phone, where there's a base-level hardware we require before you can run Windows phone. For better or worse, we like it. Sometimes OEMs would prefer to do things a little differently, but we decided that that led to a better kind of integrated experience and allowed us to think about our innovation in a different way.

So there is a set of learnings that are valuable. You could say we're in the business of selling devices, whether we build them and price them ourselves or not, we sell phones, PCs, and I'll call them TV companions. It's as much our business as it is HP’s, Dell's, HTC's, or Samsung's. It's only a good day when a lot of Windows PCs sell. So how we continue to invest end-to-end in thinking through those devices is important. And the need to think through systems design, hardware, and software on the back end is as important as it is on the front end. Given that philosophy, is there any chance Microsoft could look at hardware makers or chip companies as possible acquisitions?

Ballmer: I didn't say that. I didn't say it, and it doesn't suggest that. It says we've got to think through the hardware and software design end-to-end. It doesn't suggest anything, in fact, quite the contrary, because we have three different models. The chassis spec model for Windows Phone that works, that is starting to work. We've got an Xbox model where we do hardware and software ourselves and sell it. And we've got a licensing model [that’s] pretty open in terms of hardware design for Windows. Each of them has worked pretty well in some ways.

We observed that IBM and Oracle are both in the server hardware business, based on the argument that there are advantages in owning all the pieces. Does Ballmer disagree?

Ballmer: I don't. You don't have to [own all the pieces], but you have to think through all the pieces. Whether you have to bend the sheet metal and do all that is another question. And Oracle doesn't have the pieces.

It's sort of ironic. Neither of those two companies have the key pieces. Neither are really in the networking gear business, yet if you look at the back end of computing, compute, storage, and network are all pretty important. I think that's an opportunity for everybody.

The conversation turned back to Office 365. Microsoft announced Office 365 last October, and on June 28 it will formally launch the SaaS version of its flagship applications suite at an event in New York. (It’s unclear if Office 365 will actually become generally available at that time.) We wanted to know, what’s the business case for Office 365?

Ballmer: The pitch has to talk about speed of innovation, agility, cost, availability, reliability, and security. We've done a lot of work on certifying our stuff. It helps from a security and availability. We have a track record. It's sometimes imperfect, but we've got a better track record than anybody else on the latter front.

The biggest problem in making the cost argument is a lot of customers don't have refined cost accounting that lets them understand how much it costs them to run a service. If you’re the biggest enterprise, you might know that. If you're not, you probably don't have very good cost accounting on what it's costing you to run your information worker infrastructure. So we have to work with the customer individually so that they get a clean comparison. This is what it costs me today; this is what this will cost me; this is what I can save on people and capital.

In terms of agility and the benefits of thinking from the cloud, I'd say that tends to be pretty well accepted by the bigger customers, and increasingly by smaller enterprises.

Finally, we asked Ballmer about the big bets that the CIOs who are Microsoft’s customers are making. On this subject last year, he pointed to cloud computing and collaboration. Now, a year later, are there other big bets he’s seeing?

Ballmer: The cloud remains valid, and collaboration. And collaboration I want to underscore, because the range of things we can do just keeps going up. The number of devices from which you can collaborate, the people you can reach with the Skype acquisition, the ability to connect people inside and outside an enterprise, since a lot of what you want to do is talk to your customers and trading partners, that gets more powerful. I may double down on virtual meetings, and Kinect is potentially an interesting peripheral in the enterprise. There's a lot of things there to further double down.

The other thing I'll underscore is data. And I'll use the word "data" instead of "BI" because data somehow says I want to use all the world's information, and BI somehow says I want to use the information that we figured out how to capture inside our corporate system.

The opportunity to use large amounts of data, generated internally and externally, continues to grow. I'm excited about the fact that we have a search engine. Without a search engine, I don't know how you play in the data that lives outside your company all that well. The search engine gives us a platform to get after that. Excel gives us a platform for analysis. SharePoint gives us a platform for presentation. And SQL Server for storing and capturing data that we want to formally manage.

There's a lot of work we're doing with Windows Server and our high-performance computing work to try to push that in the direction that enables more processing of large amounts of real-time data.

With a large gaming concern, with the chief marketing lines and what they want to do, it turns out, getting people to come back to Vegas, or Macao, the problem isn't to get people to come back, the problem is to get them to spend everything they can spend while they're there. It’s almost like you want to form a profile. This person just checked in there; we saw their player card. They want to go to a restaurant. We want to know what they're doing. We want to process all that data in real time, whether it comes in our systems or not, and make sure we have an opportunity to sell them something else, get them to spend more money in our casino, in our hotel, in our theater. This explosion of the use of data in not always traditional ‘BI-ish’ ways is a big thing for us.

Microsoft a heavy hitter in big data? The company didn’t used to be considered a key player in that market. We asked Ballmer if that perception has passed.

Ballmer: You've got to ask our customers. I think so, but we're in bigger data than anybody else out there. Nobody plays in big data really except us and Google. Now, you could say, what about that Oracle company? What about IBM? Well, they do data in the enterprise, but they don't do big data. Big data is indexing all the world's websites. Big data is tracking all the world's business. That's a real big data experience. The largest data sets in the world that are being mined are those kinds of data sets.

If you asked the average enterprise customer, they'd say, our data warehouse is still on Oracle, or IBM, but I think that whole dialogue is changing quite a bit. People are evaluating our stuff. People are looking at what to do. People are doing a lot of things that are a little different than they thought about as being data before.

Security concerns give many companies pause as they consider migrating portions of their IT operations to cloud-based services. But you can stay safe in the cloud, as this Tech Center report explains. Download it now. (Free registration required.)