Systems Of Engagement
March 17, 2014 Timothy Prickett Morgan
The language may change from time to time in the computer industry, but the problems do not. Vice presidents of data processing are now chief information officers. Instead of trying to create massive systems to deal with telemetry from an array of radar stations, as IBM did with the SAGE project in the 1950s and what at the time was the most ambitious computer ever built (with over 60,000 vacuum tubes and an estimated cost of between $8 billion and $12 billion when it was completed in 1963), we have the Googleplex of well over 1 million servers that spans the globe and chews on all of our web surfing, search, and demographic data to feed us the right ad.
Well, some of the time, at least.
My point is that data is always big, and all of this talk about the exponential growth in data and how we chew on it is, well, just a bunch of nonsense. Computing has always just barely kept up with data and that is absolutely nothing new.
What is new is that the incumbent players and technologies for the data processing era that started in the late 1950s and ran into the early 1990s are all struggling to find relevance in an era where the largest hyperscale datacenter operators code their own operating systems, middleware, and applications largely using open source tools and more times than not they also design their own systems and data centers and contract the manufacturing for systems, storage, and now networking to original design manufacturers (ODMs) in Taiwan and China. The incumbents, like IBM, scaled their businesses across hundreds of thousands of customers and generated decent gross margins from their investments. Data processing was a product that they manufactured and sold. With Google, Facebook, Amazon, and others, data processing is the manufacturing process and the product they make is a service that is encapsulated in software that no one ever buys and that in turn provides the raw material–data about you–that they convert into money.
IBM never learned to be a low-cost manufacturer, and has exited every hardware business that moved down the food chain from innovation to commodity–low-end printers, networking equipment (twice now, once to Cisco Systems in the 1990s and again this year with the Lenovo sale announced in January), memory chips, PCs, disk drives, retail store systems, training, and now the X86 server business. So why should we believe for a second that IBM can become a low-cost manufacturer of computing services? Everything in IBM’s history suggests that it will eventually walk away from any rapidly commoditizing market, and there is little doubt that cloud will commoditize over the long haul.
IBM’s choice to focus on what it calls systems of engagement–the machines that chew on big data to help companies locate and serve customers rather than process the data of a sale and make sure the product is manufactured through supply chain and ERP software–and the cloud delivery model is not one of convenience or choice. In the opening letter of IBM’s 2013 annual report, CEO Ginni Rometty explained how the world was bursting at its seams with big data and that IBM was positioning itself to become a vendor of choice for systems of engagement, whether they exist on premises or run in Big Blue’s SoftLayer cloud.
The truth is, IBM has little choice but to focus on cloud infrastructure and applications and big data. IBM does not sell an X86 operating system, as do Microsoft and Red Hat do, although it does have WebSphere middleware and DB2 databases that some enterprise customers want. Moreover, the current strategy of exiting the commodity hardware business that represents the dominate platform in use by corporations the world over is, ironically as well as sadly, IBM’s only option as the world’s largest provider of IT services and one of the world’s largest and certainly most profitable system software makers. The company has spent billions on various companies that peddle marketing and sales applications in the past few years, and it is most definitely back in the application software business. But this time, the applications are going to be running on the SoftLayer cloud and IBM is going to try to keep all of the money for itself rather than share it with downstream partners. There are more big shifts coming, IBM has to find something for its hundreds of thousands of Global Services employees to do as companies shift from paying IBM to help them integrate systems to simply buying a cloudy app from Big Blue or helping them host their own applications on SoftLayer.
All that Rometty cares about is that infrastructure, platform, and software services running on SoftLayer will be counted as cloud revenue, and she has promised Wall Street that Big Blue will break $7 billion in cloud sales by 2015. As the company exited 2013, it had $4.4 billion in cloud sales, Rometty said, up 69 percent from 2012 and bolstered by those online applications as well as SoftLayer, to be sure. But don’t get the wrong impression. IBM has to be using a very loose definition of cloud, and SoftLayer was a tiny slice of revenue in 2013, At the end of 2011, when I talked to SoftLayer, then a private company, it had $85 million in sales for the fourth quarter, Call it a $320 million business. The company’s installed base did not grow all that much in 2012 and 2013, so let’s be optimistic that SoftLayer was at an annual run rate of between $400 million and $500 million in sales when IBM acquired in in the summer of 2013. IBM added 2,400 customers to the SoftLayer fold since the acquisition, which is 11 percent growth, and is shelling out $1.2 billion to build out from 15 datacenters across its entire services organization to 40 datacenters by the end of 2014. IBM will have hundreds of thousands of systems running in those datacenters, and this represents a huge expansion in its processing and storage capabilities. It is about 15 times the spending rate that SoftLayer could afford to do on its own, and IBM will have to keep this pace up if it hopes to ever catch up to Google, Amazon, and Microsoft. It is already outspending Rackspace Hosting.
The smartest thing that IBM could do to build out its cloud is buy Rackspace, and then snap up Infor and bring all of its application software back home after a few decades out on their own and run it all on the combined SoftLayer-Rackspace cloud. In fact, given the amount of investment that Rackspace needs to make to stay in the game against Google, AWS, and Microsoft, Rackspace should be happy to do a stock swap and Rackspace’s shareholders should be happy to take the deal because the company’s other option is to get stomped on by three much larger competitors. Such a three-way combination–SoftLayer plus Rackspace plus Infor–would give IBM an additional $3 billion or so of revenue, at the minimum, that it could claim to be “cloud.” Or a lot of it over the long haul, anyway. The Global Services people could be deployed to help move Infor’s customers to the cloud, many of them running on IBM i-powered Power Systems machines.
But you have to remember what IBM is throwing into the cloud revenue basket is application and system outsourcing, in the traditional sense, as well as infrastructure services on real clouds. This is done to make IBM feel bigger and stronger than it is. Amazon Web Services has hundreds of thousands of customers using its infrastructure and platform services and is probably bringing in over $3 billion in revenues from this as 2013 came to a close. Amazon does not break AWS out as a separate unit in its financial reporting, and neither do Microsoft or Google reveal their revenues for raw cloud services.
No matter how much IBM wants to think about $1.2 billion being a lot of investment in cloud infrastructure, Microsoft, Google, and Amazon are spending a lot more and have been doing so for years now. It has a lot more catching up to do, and its Watson Group is not going to fill in any revenue and profit gaps any time soon–and perhaps not ever. Google, for one, is spending plenty on machine learning, and so is Microsoft. Watson has yet to be considered a platform like Apache Hadoop is, and if IBM is not careful, an open source alternative will emerge. Actually, based on history, I would say that IBM keeping Watson closed source (even though it is based in part on open source technology including Hadoop) will absolutely guarantee that an open source alternative will emerge. This is the battle between SNA and TCP/IP all over again, and SNA lost even though it was a much better networking protocol. If IBM had open sourced SNA, we would have had an Internet many years earlier and perhaps a better one, but Big Blue may not have made as much money.
This is always about making money, and the fact is that Microsoft, Google, and Amazon have other businesses that can subsidize and justify their cloud-building efforts, and as yet, IBM does not. That is what is missing from Rometty’s letter to shareholders. All she wants to do is settle everyone down, make them believe in the 2015 Roadmap to profits, and reorganize the company as painlessly and as quietly and as slowly as possible. I am not convinced that this transition will be as smooth as Big Blue would like it to be. I do think it is ironic that IBM could end up spending $1 billion a year on X86 servers and vanity-free storage and switches instead of selling $4.5 billion a year on X86 servers.
Rometty was clear–sort of–that the company was going to stay in the hardware business and called out Power Systems specifically.
“We are accelerating the move of our Systems product portfolio–in particular, Power and storage–to growth opportunities and to Linux, following the lead of our successful mainframe business,” Rometty wrote. “The modern demands of Big Data, cloud and mobile require enterprise-strength computing, and no other company can match IBM’s ongoing capabilities and commitment to developing those essential technologies.”
Many have questioned IBM’s commitment to machines for many years, and the $2.3 billion deal with Lenovo to shed the X86 server business makes many question IBM’s commitment to Power chips and systems.
“This divestiture is consistent with our continuing strategy of exiting lower-margin businesses, such as PCs, hard-disk drives, and retail store solutions,” Rometty explained. “But let me be clear–we are not exiting hardware. IBM will remain a leader in high-performance and high-end systems, storage, and cognitive computing, and we will continue to invest in R&D for advanced semiconductor technology.”
None of that means IBM will actually stay in chip manufacturing, you will note. This is the next division of IBM that will go, if Big Blue can find a buyer, and after that, or maybe before, IBM could easily exit server and storage manufacturing. Stranger things have happened. The important thing, should this come to pass, is that IBM will continue to support its software stacks, which are quite profitable, and to provide tech support for its systems, no matter who is building them.