Jefferies & Co. networking analyst George Notter today reports on having met with Ishmael “Ish” Limkakeng, who runs the Insieme data center switching effort at Cisco Systems (CSCO), a visit that reassured him about Cisco’s position in data centers and the cloud.
Notter, who has a Hold rating on Cisco shares, and a $29.50 price target, writes the company is “executing” very well, he writes, and may win back some business lost to Arista Networks (ANET) and to so-called white box switching equipment:
Cloud computing operators are seeing limitations with white boxes, he notes:
And custom-build cloud computing switches are having a hard time being re-used:
Arista is actually much more exposed to cloud than is Cisco, he notes — a reason he rates Arista stock a Sell:
Even better, Cisco is having success selling customers on “ACI,” or application-centric infrastructure, its vision of programmable networking:
ACI is helped, he writes, by the fact the alternative, “software-defined networking” is “complex, the standards are still getting solidified, and interoperability between vendor solutions remains tenuous.”
Cisco stock today closed down 4 cents at $29.31.
Notter, who has a Hold rating on Cisco shares, and a $29.50 price target, writes the company is “executing” very well, he writes, and may win back some business lost to Arista Networks (ANET) and to so-called white box switching equipment:
Based on our conversations, we remain
very confident in Cisco’s positioning within the Data Center Switching
market. Overall, it feels like the organization is executing exceedingly
well in this business. We expect that – on the back of the Nexus 9000
product cycle – the organization will be able to hold, or even improve,
its market share in the Data Center Switching business over time. Below,
we’ve outlined number of the key takeaways from our conversations [...]
The BroadCom-based version of the Nexus 9000 is performing well also.
The strategy of building a higher performance, higher port density Data
Center switch is paying off – customers can deploy the Nexus 9000 while
getting a no cost “insurance policy” of sorts.
While White Box certainly remains a
threat to Cisco (and Arista) among these ICP customers, a number of ICPs
are feeling the inherent difficulties associated with adopting the
technology. Specifically, White Box solutions tend to force the ICP (or
any customer) into a “technology integrator” role.
Separately, we note that
internally-developed Switch solutions (such as Facebook’s “Project
Wedge” Top-of-Rack switch) remain centered on a very narrow set
of applications. As such, it’s difficult for an Internet Content
Provider to take a solution like Project Wedge (which Facebook recently
contributed back into the open source community via the Open Compute
Project) and use it for their own purposes.
We remain concerned about Arista’s
business with these customers longer term. The company generates roughly
50% of its sales via Cloud Providers – the most sophisticated buyers of
Ethernet Switches in the world. Moreover, they’re getting 27-28%
operating margin across the business – these types of margins are
inconsistent with this customer base. Over time, Arista could feel
margin and revenue pressure from these customers via alternative
solutions like White Box or internal-development. In Cisco’s case, the
organization generates roughly 5% of its $14 billion/year Ethernet
Switching business with the ICPs (or roughly 1-2% of company revenue).
It’s a very small exposure.
Cisco ended the April quarter with
2,650 Nexus 9000 customers and 580 ACI/APIC customers. Internally,
they’re very happy with these results. The ACI/APIC version of the 9000,
in particular, is tracking ahead of their internal expectations. This
is good news. As we’ve said previously, there’s a lot at stake here for
Cisco. If they can’t convince customers on the merits of ACI (and the
Cisco ASIC-based architecture in general), the Data Center Switching
market devolves into a price-per-port game with White Box solutions
setting a low reference price versus Cisco. Limkakeng noted that over
half of the 580 ACI/APIC customers are proof-of-concepts (this has been a
big area of inquiry from investors). The comments make sense – Cisco
grew this 580 figure from 300 Q/Q and customers typically take 2-3
months trialing the system before production deployments. As such, they
have a lot of POCs out there. Also, it jives with Arista’s recent
comments that they aren’t really seeing ACI competitively right now.
Limkakeng commented that these POCs – for the most part – will turn into
full-blown ACI implementations.
Cisco stock today closed down 4 cents at $29.31.
No comments:
Post a Comment