There’s no doubt that Cisco’s entry into the compute side of the data center is an Atlantic City size gamble. But for the established vendors, “there’s trouble bussing in from out of state.” So who stands to win and who stands to lose?
I think the biggest winner here is the end user, Cisco’s entry in to the market, regardless of its technical merits or long term success, will shake up the other vendors and accelerate the development of more efficient, more manageable compute infrastructure. The x86 server market has been too stable for too long, and it takes a vendor with Cisco’s resources to really shake things up.
The potential losers are HP and IBM who currently own a substantial majority (~75%) of the blade server market. But picking off their customers may not be as easy as it sounds (and it doesn’t sound that easy!) The whole point of blades (well maybe not the whole point, but certainly a big part of it) is vendor lock-in. Investing in a blade solution means investing in an infrastructure such as chassis, blade switches etc, this increases the cost of entry, but more importantly in this context, it increases the cost-of-exit which may make the Cisco product a tough sell into HP and IBM’s installed blade base.
Companies like Sun and Dell with smaller blade businesses have less to lose, unless you believe that ultimately all servers will be blades (and I’m far from convinced about that). Also, Dell in particular does a great deal of its server business in the small/medium business market where the price of UCS is likely to be a significant barrier.
For the software companies that Cisco had on tap to say nice things (and boy did they lay it on with a trowel) this is a no-brainer. So long as Microsoft/BMC/VMware sell their software, they really don’t care who the hardware partner is. What they do care about is a hardware partner becoming too dominant, so Cisco muddying the waters suits them just fine.
For Cisco, the question is can they make enough net-new revenue from UCS to justify the costs of getting into and staying in the server market for the long haul. Cisco’s cost will include:
- Development costs: Non-recoverable engineering costs of developing the entire UCS platform are no doubt substantial. These costs are not a one time thing either, they will have to maintain investment to keep up with a rapid pace of development.
- Sales & marketing costs: Going after server business will entail substantial sales and marketing costs given that Cisco has to penetrate an existing market with established and deep pocketed incumbents
- Support/services/consulting infrastructure: Make no mistake, adding enterprise level service, support, and consulting for UCS will not come cheap.
In covering these costs, Cisco has to contend with the fact that it is only attacking a segment of the x86 market. According to IDC, blades accounted for less than 20% of x86 server revenues in Q4 2008 ($1.4B), and the best part of that market is currently dominated by HP and IBM (both in terms of quantity & revenues/unit shipped.) So Cisco either has to take customers away from those companies, or expand the market in a significant fashion in order to be successful. Generally, market expansion is a function of price, i.e. as the price drops, the market expands, but Cisco isn't trying to drop the price, if anything it's more likely to raise the price and try to make a "value" play (i.e. it may cost more, but you can do more with it, so you end up saving money.) This is always a tough sell even in the best of times, after all, who actually believes vendor "Return On Investment" calculators? But trying this approach in the toughest economy we've seen in my lifetime is going to be even more challenging.
Posted by: Nik Simpson