Juniper reported its numbers, which showed better profits and a slight improvement in revenue, and then issued a pretty nice 3Q outlook to boot. The initial reaction of the Street was mixed; some hoped for better performance given Juniper’s multiple, and others were happy. But earnings may not have been the big news. Kevin Johnson, Juniper’s CEO, announced he would be retiring.
There have been rumors of Johnson’s departure emerging this year. He arrived from Microsoft to replace Scott Kriens, the original CEO and now Chairman, and many thought he might push Juniper out of the box-fixation mindset that has been its legacy. He didn’t, and in my personal opinion he didn’t really grasp the difference between “software”, “embedded software”, and “network software” in an SDN and NFV age. Juniper may have embraced software in an organizational sense, but not in the sense that it needed to.
What should have been done? Clearly, Juniper like other vendors was facing pressure from operators to support the new operator monetization goals. Logically, that meant providing service-layer software that would allow operators to build new services that were competitive with those of OTTs, but also to recast current services in a more modern, cost-effective, profitable, and flexible way. Juniper had an initiative, “Junos Space” that could easily have done that, and when I reviewed their concept at the launch almost three years ago I believed they would take the steps they could have taken. They did not. Space became a very simple “operations” tool, a slave to cost management and TCO and not even a factor in monetization.
When SDN and NFV came along, Juniper embraced the former and at least in a positioning sense ignored the latter. Service chaining is an NFV use case, but Juniper presented it as an SDN application. Yes, you can chain services with SDN, but unless you frame service chaining in the operations and deployment context of NFV you don’t have the savings that made it interesting in the first place. I called Juniper out on their tendency to sing SDN songs about NFV concepts, but they’ve really not changed that theme at all.
I don’t know what Kevin Jonhson thought Juniper software would look like. Like a Windows ecosystem? Some inside Juniper have told me that’s exactly what he thought. Like operations glue to link Juniper to OSS/BSS? Some say that too. The problem is that the ideal Juniper software story isn’t either of those, or perhaps it’s both but at another level. Network software is about the virtual world that networking lives in, and in particular about the elastic and critical boundary between SDN and the network, and NFV and the cloud. NFV, which to be fair came about long after Johnson joined Juniper, defines a framework that is aimed at costs but can be applied to revenue. The critical error Juniper made under Johnson’s command was to ignore NFV because it seemed to be about servers, and embrace SDN because it seemed to be about software and networks. Semantics are a bad way to engage the customer.
Cisco is the ranking player in networking, in SDN, and even in NFV even though its positioning is as vacuous as that of Juniper. Why? Because they’re the incumbent, and all they have to do is kiss the right futuristic babies and they can hold on. Juniper has to come from behind. Its earnings are not a reflection of it’s strategic success—it’s losing ground steadily in strategic influence. The earnings reflect the inertia of the industry, an industry that buys stuff on long depreciation cycles. It will take years for operators to wean themselves out of Juniper gear even if they try, and in that time Juniper needed to be darn sure they didn’t try. That’s what Kevin Johnson was likely hired to do, and he didn’t do it successfully. Juniper used to be a box company that couldn’t position strategically. Now they’re part box company, part traditional software company, and still groping with the real problem of defining “network software” and their role in it.
The Cisco acquisition of Sourcefire is even more logical in the light of Johnson’s departure. If Cisco can kill Juniper in enterprise security and cloud security while they fumble for CEO candidates it won’t matter much who they end up with. And security is only one of three or four cloud-related killer areas where Juniper needs a cogent strategy to develop a lead. If they miss any of them, they’re at risk to losing market share, and if their P/E drops to the industry average they’re a seven-dollar stock. Think M&A, but think of it under decidedly buyer’s-market terms.
Watch the CEO choice, and watch what they do in the first hundred days of their tenure. This is do-or-die time for Juniper.