IBM has chosen to challenge the hegemony of the x86 chip in the server market – and, as far as I can tell, this is a serious challenge. In short:

IBM will be offering Power Series servers to compete aggressively with x86 servers, at price levels that significantly undercut the price of a typical x86 server.

The What, How and Where?

First of all, IBM will only compete in the Linux area of the market. This constitutes less than one third of the x86 server market, if measured in terms of servers shipped. However, that statistic excludes the cloud, where Linux deployments dominate and where IBM will surely be trying to attract attention for Power Series.

IBM’s surprise attack on this market has clearly been thought through. The Power Series will be aggressively priced in terms of TCA (total cost of acquisition) with typical pricing more than thirty percent lower for virtualized environments than for equivalent x86 servers. The TCO (total cost of ownership) is likely to prove more advantageous than that, but in the mid-market IBM will promote TCA, because the mid-market thinks in TCA, not TCO. 

IBM’s edge is primarily the capacity/performance advantages it delivers with its efficient implementation of virtualization (PowerVM). This is likely to attract attention for several reasons, aside from lower cost.

• It is more flexible than an equivalent VMware virtualization capability, offering dynamic deployment of virtual machines, arguably superior security and better scalability.
• In terms of benchmarked scalability the PowerVM (compared to VMware’s vSphere 5.0) tends to deliver 100% improvement in throughput – with the advantage gradually increasing to 131% as the number of cpu cores grows.

IBM is partnering with a number of key of Open Source vendors and products to target specific application areas (Red Hat, SUSE, Hadoop, Apache, MySQL, Samba, etc.) and to provide a comprehensive and inexpensive software stack that includes infrastructure and web capabilities.

IBM’s Targeting

The main stumbling block for IBM is that it already has a healthy share of the x86 market via its System x platform. So IBM needs to ensure that it does not cannibalize that revenue stream, or indeed cannibalize its traditional Unix business – otherwise it will simply compete with itself to its own detriment.

For that reason, IBM is specifically targeting new customers. That presents its own difficulty, simply because the cost of customer acquisition is always higher for new customers, and deals normally take  longer to mature.

The specific markets and application areas IBM is targeting are:

• The x86 Linux ISV market. Here the TCA and virtual machine capabilities are likely to be attractive. Additionally the rewards for success in this market will be high in terms of revenues (from multiple server sales) and sustainability.
• The so called “big data market”. Sales here often involve large numbers of units and this application area is richly served by Open Source products and large scale Linux databases.
• “Smarter Planet.” IBM will be targeting applications associated with and related to its smarter planet initiative (large scale infrastructure projects). In this area it has specific credibility because of its early consultancy engagements.
• Watson. IBM has its well publicized Watson AI application, which gives it a unique offering in some verticals – particularly in health care, pharmaceuticals and areas of the financial sector  (for risk management, for example).

Although it has not publicly stated this, my expectation is that IBM will make an aggressive push to large scale cloud vendors and large web related businesses. Here Linux dominates and actual the cost savings on IT go straight to the bottom line. Why would the cloud vendors ignore that?

Why has IBM Chosen To Do This?

The x86 market is very broad and has many niches. The x86 architecture is already being challenged “from below” by the ARM family of chips and Intel has been unable, thus far, to penetrate the mobile market. This fact is important because it pinpoints the difficulty of quickly establishing a technology ecosystem around a chip architecture, or a hardware architecture for that matter. This is a difficulty that IBM faces and which it appears to understand well.

Revenue growth in the x86 market is consistently higher than it is in the Unix server market. As x86 chips increase in performance they eat into the lower end of the Unix server market. For that reason, IBM’s challenge to the x86 market with its Power Series platform can be seen as both defensive and aggressive.

The Unix server market as a whole is exhibiting modest growth by revenue and has not, in revenue terms, returned to the pre-recession level of 2007/8. Within that market the Power Series is dominant, taking business in roughly equal measure from the two other major participants; HP and Oracle.

Technically there is no indication that the underlying hardware technology of HP and Oracle (the Itanium and SPARC cpus) or the associated software stacks, will be able to compete with IBM’s Power Series in the near future. More likely is that IBM will establish a similar level of dominance in the Unix server market as it eventually achieved and now enjoys in the mainframe market. This may guarantee Power Series growth for years to come. But even so, it will likely be modest growth.

Additionally IBM has established a market for its Power PC chip on games machines (Nintendo, Microsoft, Sony). However,  while IBM has achieved a significant volume market here, sales of gaming consoles have stalled in recent years, and the market – although delivering a pleasant and reliable revenue stream to IBM – does not appear to offer strong growth prospects.

Clearly IBM has an incentive to establish new markets for its Power PC chip and also for its Power Series hardware. And clearly, that is what it is attempting.

Conclusions

Our conclusions are as follows

1. This is a bold move and will likely disrupt the x86 server market.
2. We expect the x86 server vendors to ignore this challenge until it impacts their business in percentage terms rather than by the loss of a few sales.
3. IBM will, in our view, experience a level of success. It is difficult to predict how successful it will be, but we should accept from the outset that it takes time to penetrate a new market, both in respect of marketing (getting out the message) and sales infrastructure. Nevertheless, we believe that IBM’s chosen pricing is far too aggressive for this initiative to be ignored by customers.
4. Unless IBM pulls off some unexpectedly large deals (for example, winning business with a major cloud provider or social network site) we expect to wait a couple of years before concluding whether this initiative is successful. We’ll know then from the sales curves.
5. This is a very refreshing development from an industry perspective.