While comparing notes with Brad Casemore and a others a few weeks ago after his VCs Weigh SDN’s Risks and Rewards post that added commentary to my What does Arista, Insieme & Vyatta have in common? Implications for SDN post about the suitability of network virtualization for venture capital investment, a number of blog-worthy points surfaced that were not discussed in either post.

1. Distribution channel is controlled by incumbents.  Enterprises we speak with expect network virtualization to be bundled with the virtualization solution they consume.  These whole solutions are created and sold in large part by the likes of Cisco (w/ VMWare + EMC via VCE), IBM, HP, or Dell.  Or by up and comers like Arista who partners with VMWare and others to build whole solutions. Arista has succeeded at building distribution because they have a 'hot' box that customers demand - which is harder than it looks.  This is also increasingly true in the service provider market where large carriers outsource the development, deployment, and in some cases operations of their networks to companies like Ericsson, Huawei, and ALU.
2. The 'S' in software-defined networking is for software -- which means unless your company is one of the few on an amazing revenue ramp, most acquirers will look at your SDN company from a build vs buy lens.  Meaning the cost to buy the company can't be dramatically more than the cost to build the product.  One rule of thumb to consider is acquirers typically pay 2X - 4X the development cost of a product -- unless there's a huge potential bump in revenue or strategic risk for major revenue loss.
3. Most of the logical acquirers are either a) financially hurting (e.g. they can't afford to pay up, like a HP, JNPR, Brocade etc); or b) are bottom feeders (e.g. historically pay low multiples for companies - VMWare, Oracle, IBM, etc).  Brad adds a good point by adding that this a sign of the times. With the macroeconomic picture a muddle, theres's a limit as how much anyone will pay for software-based infrastructure.

During our exchange, Brad brought up another valid point which is, it's not clear that hardware-driven networking vendors can adapt to a software-driven business model. If they acquire a successful SDN startup, regardless of what they pay for it, and then try to graft that software onto a box-based business model, the acquisition will ultimately fail. These aren't bolt-on acquisitions. The acquirer might have to change how it does business to succeed in this market.

Another strike against large SDN / network virtualization acquisitions is, given my knowledge about the personal and institutional investors of the hottest SDN startups, an industry dynamic has been unintentionally created that limits those startups ability to structure an effective bidding war for their startups.

I think the only place where Brad and I may disagree is his statement that  "the VC investments we’ve seen heretofore in SDN already suggest that it is perceived differently from the linear networking markets that have preceded it."  That may true for Andreessen Horowitz and their investment in Nicira though a) Nicira is AH's first networking investment; and b) from my customer inputs, Nicira today appears more like a company that uses professional services to finance software development, than the VMWare of networking.  There is nothing wrong with this approach, it's just not revolutionary.   If we look at other venture capitalists who've invested in SDN -- they are primarily the same investors who've financed the previous round of networking companies with the same linear investment structures and thesis.

What does this mean?  Software-defined networking and network virtualization are more likely than not, evolutionary changes that favor incumbents.  While I can't predict the future, it feels like there is: a) a relatively small pool of buyers for SDN companies; b) that if they acquire, expect mostly $20M - $100M exit valuations;  and c) maybe one or two SDN companies get big enough to go public; but even then Cisco who has a history of containing these companies to relatively limited markets (like ADC or WAN Acceleration) they become too big to be acquired, yet, too small to inorganically expand into sizable adjacent markets to challenge Cisco (or others) for networking dominance.

Those conversations further reinforce my belief that the networking market has limited options for over-sized returns expected by the limited partners in venture capital funds; making venture capital an inefficient vehicle for financing networking innovation, including network virtualization; and as a result, people should choose to join a venture-backed SDN startup because of passion; while understanding that the odds of a financial outcome greater than their current gig are small and that the risks taken for an opportunity for a life-changing financial outcome are quite high.

Where do we as an industry go -- and how can we finance networking innovation while rewarding those who take the risks?   I'm in the middle of researching that right now, using other industries as examples and models.  I'm optimistic and will share details in a future post.  If you have an opinion or thesis, would love to have a conversation.