Is Ethernet Going Sour?

With Brocade’s cut in guidance on its earnings call, the company joined what seemed a parade of network equipment vendors who’ve called the future of network spending into question.  Most Wall Street analysts have suggested that Ethernet is coming under pressure and that corporate IT spending is likely to be weak.  Both are likely true, but I think the Street is (as usual) content to catalog symptoms rather than address problems.

In the enterprise space, our spring survey found that enterprises were still generally holding their capital plans but were slow-rolling project spending.  A part of the reason was concern over economic conditions, which was a visible issue even before the harsh political face-off that’s virtually killed market confidence this summer.  Another part was some concern over their cloud plans, concern that arose from getting more insight into cost and benefit as they got deeper into the topic.  Both these issues appear to have grown over the summer, and I expect our fall survey to show that.

Data center modernization is the only real driver of network change in today’s market.  Nobody has demonstrated any direct productivity gains out of network change, despite Cisco’s attempts to make telepresence the water-carrier for network expansion.  The problem is that virtualization as a driver for data center modernization appeared to have tapered off even this spring.  It’s not that people weren’t doing it anymore, but rather that the network change part was largely baked and they were back-filling into pre-existing plans.  Cloud computing was the big driver remaining, and the cloud has proved more complex than enterprises had expected.

In the service provider space, I’m seeing the result of five years of declining revenue per bit.  But the thing that’s really hitting now is a more subtle structural issue.  Content, which everyone knows means “video” is the driver of traffic in both wireline and wireless, to the point that you could almost neglect other growth sources in planning.  But content isn’t “Internet” traffic as most would know it.  More and more content is served out of metro cache points, and so it’s the metro capacity that’s consumed.  Metro means Ethernet, and the growth of Ethernet to support content delivery has been the driver in a shift for operators from IP-dominated planning and spending to capital planning that’s Ethernet-dominated.  That process at first tended to focus on more premium players and products because early metro/aggregation Ethernet was an expansion on the previous business-focused Ethernet services infrastructure.  In most metro areas today, according to our operators, the impetus for Ethernet growth is consumer video, and that’s the worst service in terms of ROI.  Thus, price pressure on Ethernet is inevitable.

Where the economic situation enters the picture is double-edged.  On the enterprise side, uncertainties about the revenue line will encourage most businesses to hold back, to delay spending as much as possible.  I think it’s pretty likely that Q3 will be soft for that reason no matter what happens at this point with the economic picture, simply because it’s not possible to wash all the uncertainties out of the market by the end of September.  The question will be whether project budgets take the larger hit, which they normally do.  That’s important because nearly all IT spending growth comes from the project side, because that’s where new benefits are typically introduced.  But in addition to representing the incremental spending for this year, the project budgets for 2011 reset the baseline for 2012.  If those dollars are not spent this year, then the later spending that depended on those projects is also curtailed.  That’s the risk, and it’s particularly acute given that enterprises will be starting their fall tech planning for next year in only a month.

The Ethernet shift for operators will be critical because it’s an illustration that spending is increasingly focused on places where the only differentiation is cost.  Ethernet features are virtually impossible to make meaningful, and “cost” beyond the direct cost of equipment has gotten tied up in conflicting vendor claims, none of which have been compelling to the buyer.  I asked operators what role vendor studies and figures on operations savings meant to their selection, and they told me that these were used where it was convenient to build a management justification for the choice they wanted to make, but almost never actually influenced that choice.  Thus, Ethernet is incredibly subject to pricing/commoditization, and that’s what we’re going to see.

Cisco’s earnings are due on Wednesday, and the Street seems to think they’ll roughly meet guidance.  If they fail to beat the estimates nicely or if they are also cautious in guidance, it will be an indication that the networking industry is in for a very tough patch in 2012.  Switching will be the place to watch too.  The word is that Cisco has been aggressively discounting its own Ethernet products and UCS servers as well, and that’s not helped the industry’s margins in the Ethernet space.  More discounting will confirm my thesis, I think.

 

 

Leave a Reply