The M&A in network equipment continues, but it also continues to involve primarily specialty firms being picked up by the giants. The latest is Cisco’s buy of optical specialist Lightwire, a company that’s specializing in creating optical interfaces whose low power dissipation (heat) means they can be packed more densely. That’s a pretty solid indication that Cisco believes that they’ll be deploying higher-density optical interfaces in products, and that in turn could offer some insights into their view of the market.
This sort of high interface density could arise in two areas of the network—the aggregation edge, driven largely by demands for mobile deployment (particularly small cells) and in the data center. For Cisco, a higher-density product in either space could give them some additional differentiation at a point where success will likely be key—the point where new service plans first impact infrastructure.
The question now is what’s driving Cisco here, other than this obvious differentiation point. Do they intend to try to drive the market themselves with something that could improve their cloud momentum or mobile/behavioral creds, or do they simply plan to let nature take its course and use the interface-density point as a way to intercept opportunity? If they do the former, they gain not only the ability to time the market on their own terms but also a chance to get the higher-level story that is most likely to generate press attention. The price is the greater risk they then take; if you want to jump-start a market you have to give it the Kiss of Life for a while no matter how repulsive that may seem.
This comes as UBS reports that Cisco is at least threatened at the switch level. Campus and LAN switching accounts for most of the spending in the enterprise Ethernet space but our surveys have consistently showed that it’s being increasingly driven at the requirements level by data center modernization, and that is increasingly driven by IT business model changes and cloud computing. In 2011, UBS says that Cisco lost some share to rivals like Juniper, and so some strong play in the data center might be a very good idea for Cisco. That’s particularly true given that Juniper was downgraded just yesterday by one Street research firm for its delays in getting its own new products out there. QFabric, Juniper’s own data center switching keystone, is one of the products now expected by the Street to lag. Does that give Cisco time to counter, and force rivals into counterpunching against a Cisco initiative? Could be.
Then there’s wireless and backhaul. We’re seeing most of the big RAN players focusing on making their wireless gear more suitable to smaller-cell deployments. Even LTE cells are easily congested; users who expect to get double-digit download speeds could use up a complete cell with even accidental concentrations at a pedestrian crossing, much less a collective rest in a hospitality location. The more cells, the more backhaul, the more potential for optical feeds…the more cost. This also ties back to Cisco’s Lightwire strategy; high-density cells mean high-density backhaul. Huawei has joined the rest in coming out with a small-cell strategy, so you know the market is getting serious.
Might we then expect to see more equipment startups, more “network technology” instead of social-overlay stuff, in the future? There have been a few indications that could be true, one of which is the sudden interest in social IPOs. A new one is the momentum of “do not track”, both in terms of Federal policy and in terms of “voluntary” industry steps. The fact is that absent continued additional targeting and manipulative data on which to differentiate, the online ad market will see price commoditization in a couple of years that will largely offset its growth in appearances of online ads. I think this is why we already see a few brave VCs looking at things like OpenFlow switching. Frankly, I’d love to see some innovation at the lower layers here; we might even be able to do a better job of making the service layer work with something like OpenFlow.