fantasy vc - cumulus
Continuing a series on startups I'd put a bet on if I could.
Pundits and analysts like to cite SDN, NFV, Open Compute Networking, and the ever greater capability (thus usage) of merchant silicon as harbingers of commoditization. I think "commoditization" is the wrong word, used loosely without any regard for what it actually means. Rather, what we're seeing is [maybe] the onset of a kind of x86-ification or open-systems-ification of networking. Given networking (telegraph, 1844) is historically behind compute (Babbage, 1830) systems, this is not without precedent. ;-)
Either you disrupt by doing something new. Or you disrupt by changing the supply chain, removing middlemen, disintermediation (or consolidation of intermediaries unto yourself?). When was the last time any significant disruption happened in networking? Arista looked like it was going to stir some things up, but has more or less ended up as a niche version of the typical vendor. But it did take a step in the direction Cumulus continues down by abandoning spinning it's own silicon and focusing on Linux-based OS and hardware packages.
[This is where people with history in networking start berating me for glossing over all the other attempts and examples and acquisitions and research and….]
Put these things together:
- Network management is a mess, has been a mess, and continues to be a mess
- Config automation is ascendant
- Application-centrism is ascendant
- Network gear is made up of specialized computation machines; why should it be isolated from the same historical progression as has happened with other kinds of computation machines?
- There's precedent for OS/hardware independence
- Has anyone ever cut out OEM's before and made distributers/VARs into the only packagers for product? What happens when Channel gets to command the margin?
- There is a market [small but potent] for mix and match network hardware and operating systems
- Every network incumbents' margin structure is someone's opportunity and those margins are fat
- Mainframes, Power Systems, Superdomes, etc., are niches while most servers sold are a variety of x86 hardware + standard OS packaging exercise
That's not to say that they're guaranteed success. Or won't get crushed by an incumbent or other party. Or even scooped up before they become too successful. Just that I would've placed that bet.
Disclosure: Cumulus is not in my former coverage area and I have no financial interest in them.