Until recently, the tech market focus on processors surrounded two current processor technologies: ARM, which exists mostly for mobile devices but is trying to move up market to PCs and servers, and X86, which is mostly on PCs and servers and trying to move down market to mobile devices. ARM has the advantage, largely because moving up market increases margins and thus is far easier to fund, while moving down market reduces margins, making it hard to get the funding needed to compete. However, Intel, the dominant vendor in the x86 space, due to those larger margins, can put more resources on defense than ARM can put on attack. This suggests that Intel would be strongest on defense in the space it occupies and ARM the strongest on offense.
In the midst of this battle are a number of alternative processor technologies. The most powerful is arguably IBM’s Power because it is backed by one of the most powerful and the most experienced technology venders in the segment. Its success was emphasized when IBM passed its 100-year anniversary and celebrated the 50th for its most profitable platform: the mainframe, now called the System Z Server. But even though Power is currently eclipsed by Intel on servers and ARM on volume, it suddenly may be able to move around both technologies due to IBM taking its focus off Intel and putting it on unmet customer needs.
To succeed, IBM has to make big gambles in order to move where Intel can’t or won’t and to reach manufacturing volumes that put it back into the race. Well, it opened its microprocessor technology (OpenPOWER), which is incredibly risky, but likely critical if it wants to achieve its goal. Let me explain. I think this is bigger than folks realize, so I’ll cover that this week.
The Problem with Intel
If you talk to server or PC OEMs, the biggest complaint they have is that Intel is inflexible and expensive. This comes with the turf of being dominant because on servers and PCs, Intel faces little real competition (though AMD has been stepping up its game of late). And due to its massive volumes, Intel has no real desire or motivation to create custom parts that address individual OEM needs and, as all companies do, it remains tightly focused on maintaining margins. It is also burdened with a near legendary love/hate relationship with Microsoft which, as a software company, isn’t interested in protecting Intel margins and wants hardware to get as cheap as possible so that Microsoft sells more licenses and can charge more itself, while keeping server and PC prices down to maximize volumes.
This makes it really hard for Intel to innovate on anything but process technology and areas of common demand like security and performance. As a result, Intel is more likely to tell customers what they are going to get than to listen to what they want. Now, be aware that even though OEMs want advancement, they often complain when Intel makes major changes because those changes create support problems for the OEMs. VMware was apparently created initially just to programmatically deal with and cover the changes Intel was making at the time.
Emerging large-scale segment-leading OEMs and cloud service providers like Apple, Google, Microsoft, Samsung and Amazon have increasingly been developing processor engineering organizations in-house because they either want more control over the technology or need to differentiate from their competitors more. They don’t have the time or resources to build a processor from scratch and need at least some of the economies of scale that a common platform provides. ARM gives them this latitude but doesn’t really step up to the performance threshold they need, particularly for servers. But this does also speak to why Intel is having difficulty penetrating these new segments.
Power has the performance that these folks need but it lacks consistency with x86 and thus doesn’t have the economies of scale, in terms of market numbers, to reduce the risk enough to make it an Intel alternative, and IBM competes with some of the key vendors, making them leery of taking the products from the firm. To break out, IBM needed to break the model.
IBM created the OpenPOWER Alliance, mirroring to some degree what ARM had done and opening up its processor architecture to companies that had the capability to design their own chips. Almost immediately, two of the vendors that wanted and needed more flexibility than Intel provided, Google and Samsung, signed up. Google is the most powerful of this initial set and will either be a king maker for IBM or use the IBM effort as a stick to beat Intel into submission. Either way, IBM’s OpenPOWER platform nearly instantly went from near irrelevance to market driver because it is in a position to force Intel, which desperately wants to hold on to Google, to react.
Now a risk clearly exists as these vendors are likely to create services and products that rival IBM’s, but if they don’t take this risk, they don’t get to relevance. Interestingly, this mirrors somewhat the moves Intel made initially when it licensed its technology to IBM and AMD in the early years to make a market. Had it not, it likely wouldn’t be where it is today.
Wrapping Up: Thinking Differently
Often when you see companies compete, those that follow the leader tend to increasingly emulate that leader. This behavior assures that the leader continues to lead because its competitors train themselves to follow. The way firms like Intel, Amazon, Microsoft and Apple took the market from prior leaders wasn’t to chase them but to change the path and then get to a new goal, that they defined, first. This isn’t easy to do because the path of the leader is so compelling and a tight focus on it typically prevents seeing any other path. In addition, the leader’s path is safer because any problem is segment-wide while a firm going its own way and tripping up will look stupid for not following the trail the leader has blazed.
It is interesting to note that virtually every market leader got to be a market leader by blazing its own trail. Few vendors, even those that were market leaders, are willing to take that risk once a market has matured. Apple thought differently and took the market Microsoft had taken from it back, putting Intel at risk in the process. This weekend, it might be interesting to think about whether your firm is on the optimal path and the opportunities that thinking differently might allow.
With OpenPOWER, IBM is thinking differently and for the right company this has clearly created interest that otherwise wouldn’t have been possible. It will still need to translate into volume, but the first milestone, making Power relevant again, has been achieved.