Lenovo Deal Done, Power Systems Takes Center Stage

October 6, 2014 Timothy Prickett Morgan

Well, the deal is done, and IBM is completely and finally out of the X86 server business now that it has sold off the System x division to Lenovo Group. Or, to be more precise, has sold off the System x business in most of the major countries where IBM and Lenovo both had operations and could do the transfer and get the approval of regulators. Now, we find out exactly how committed Big Blue is to the Power Systems platform.

Ever since the homogenization of the server lines that started at the end of the 1990s when Sam Palmisano was getting ready to take the helm of IBM, the company has had to be very careful about putting any one of its server platforms ahead of any of the others. Back then, IBM had AS/400 and RS/6000 product lines based on essentially the same Power-based iron, plus System/390 mainframes and System x X86 servers. The machines did not have much in common and had different design cycles, supply chains, and routes to market. IBM decided, either rightly or wrongly, to not try to push its own platforms against those based on processors from Intel. In the mid-2000s, like other server makers, IBM added chips from Advanced Micro Devices just to be able to apply some pressure to Intel and to offer customers whatever unique advantages that came with its Opteron processors over the Xeons at the time. And without knowing any numbers, Intel is generous to those server makers who toe the Xeon line and who buy in volume. IBM was never as close to Intel as Hewlett-Packard and Dell, and we have to presume that the co-marketing dollars and volume discounts for raw chips and chipsets played into the favor of HP and Dell and against IBM to a certain extent, especially after IBM sold its PC business to Lenovo a decade ago.

It is unfortunate that the Power Systems division has the Xeon muzzle removed now rather than a decade ago. Back then, the Cell Power chips were being deployed in the world’s fastest and first teraflops-scale supercomputer, Cell variants were being deployed in the major game consoles, Apple was still on PowerPC chips, and PowerPC embedded processors were the natural choice for a lot of industrial and consumer devices. Had IBM launched the OpenPower Foundation then and opened it up, ARM might have never gotten into all of our cell phones and tablets and might not be a threat in the datacenter as it currently is. The Power chip business might be generating billions of dollars in its own right, and IBM might be able to do very sophisticated things with the architecture it as Intel is doing with its X86 architecture, ranging from low-power processors for microservers and switches to hefty number-crunching, massively parallel chips aimed at supercomputing work.

But that didn’t happen, and IBM is trying to reinvigorate a much-diminished Power ecosystem. To Big Blue’s credit, it has at least seen the complete picture and has chosen a course of action–I hesitate to say the right course of action because only time will tell–to try to move Power forward.

IBM is hosting its Enterprise2014 event in Las Vegas this week, and the new Power8 machines aimed at big data workloads with FPGA and GPU coprocessor acceleration as well as the enterprise-class E870 and E880 machines, which replace the Power 795 and Power 770+ and Power 780+ machines in the prior lineup, are the stars of the event. And now it is up to the IBM top brass to demonstrate the performance and value advantages of the Power Systems platform. They need to be aggressive and engaging in this, and they need to get their channel partners fired up about taking on X86 platforms with Power. This means arming them with reams and reams of data about how Power can–and does–beat X86 machinery.

It also means focusing on IBM i and AIX as well as Linux, even though IBM wants to characterize this fight as one taking place between Linuxes on X86 and Power iron. Linux drives a little more than a quarter of the server revenue in the world and soon it will be a third; the majority of the rest of is driven by Windows on X86 with a smattering of mainframe and Unix platforms. IBM is going to need to preserve its AIX and IBM i bases and grow them if it wants to be able to continue to invest in the Power platform, particularly if it is not able to sell off its foundry operations. This cannot be only a battle for the hearts and minds of the Linux community, even if, as I suspect, that Linux will eventually account for half of worldwide server revenues as clouds and hyperscale web applications continue to grow faster than enterprises.

As part of the deal last week, Lenovo paid IBM $2.1 billion to buy the System x division, which was $1.8 billion in cash and $280 million in Lenovo stock. This is about $200 million shy of the expected payment for the deal, but these things happen. Adalio Sanchez, who used to be the general manager of IBM’s System x division, is now the Lenovo senior vice president of Enterprise Systems, reporting to Lenovo president Gerry Smith. Sanchez, who was also previously general manager of the pSeries division (one of the many names of the RS/6000 line over the years) and the mainframe division at IBM, among many other jobs. Lenovo has not just acquired the System x division but also a portfolio of software including IBM’s General Parallel File System (GPFS) and Platform Computing cluster management and Java messaging software. All of these components are, not surprisingly, a part of IBM’s own analytics stack on Power8 machines and by licensing them to Lenovo, IBM is helping to foster its own direct competition on an X86 platform.

You might ask why IBM might do this. The reason is simple. Without this software stack and without licensing of Storwize storage arrays and LTO tape arrays, Lenovo would not do the deal. And IBM wants cash and it wants out of the System x division that it may have given away a little too much. Only time will tell. What seems obvious is that IBM is going to have to sell Power against Lenovo, and often using the same essential software stack. Maintaining the AIX and IBM i customer bases and growing them will seem easy by comparison.

There are billions of dollars at stake, and perhaps the very future of the Power platform. And all readers of The Four Hundred have a stake in that. Hopefully IBM plans to do something useful with that $1.8 billion and whatever money it is no longer losing in the System x business to market the Power platform and expand its market.