Most people in the IT space understand the trend towards bare metal everything (servers, switches, etc.). The movement to commodity hardware drives down costs in infrastructure where price is disproportionately determined by the cost of hardware. While most of the industry dialogue is around price, there are strategic implications of this movement that extend well beyond the mere procurement of networking hardware.
Pricing in networking
For anyone looking at the broader implications of bare metal switching, the first thing to understand is networking’s pricing model. While the vast majority of investment (as much as 90% of R&D at large companies) is on the software, the hardware accounts for the lion’s share of pricing. Because the pricing mix is unjustifiably skewed towards the hardware, energy behind bare metal switching is even higher than it would otherwise be. When hardware is the major (or only) means of monetization, it will tend to be priced higher.
This creates an abnormally large bogey. If you take something that is priced high in a market with a large addressable opportunity (TAM), you create a very attractive space for would-be competitors. The appetite for lower-cost platforms is further fueled by the presence of a dominant incumbent making 65 points of gross margin.
With an opportunity this large and attractive, it was only a matter of time before companies targeted the space. For awhile, Arista was the primary price competitor in networking. Now, with bare metal switching taking root, they are joined by companies enabled by the likes of Cumulus, Big Switch, and Pica8.
The likely end game here is that everyone converges on a narrow set of common components, which would mean that hardware costs will eventually normalize to a fairly narrow band. There will always be someone who is willing to drop a couple of points to buy additional market share, but even these drops will asymptotically approach some margin threshold.
The real difference here is that, given the large addressable market, the margin threshold to play a significant role is dropping. Where Cisco might operate at 65 points of gross margin, other players might be willing to drop that to the mid 50s. How low people go is largely a game of chicken with Cisco. So long as competitors keep pricing close, it is in Cisco’s best interest to do nothing with regard to pricing. Changing pricing across the board results in lower margins, and spread over high volumes, this constitutes a material change in business model.
However, if a competitor drops too low (and starts making real inroads based on price), then Cisco is forced to respond. By pricing their products even close to the price leader, the delta in price becomes too small to justify a move away from a tested incumbent. This effectively squelches price as a differentiator. Oddly enough, this means that those who would compete on price actually have to price higher than they otherwise could to avoid unnecessarily pulling Cisco into the pricing gambit.
The real disruption from bare metal switching is not in price so much as it is in margin. The difference between a company willing to price low and a dominant incumbent with healthy share is not really related to whether the hardware is bare metal or not. In fact, with everyone building very similar hardware based on essentially the same components, pricing favors those that ship in volume. Cisco will get better per-unit pricing on Broadcom chips, for example, than anyone shipping (and thus buying) in lower volumes.
If not price, what is the real impact of bare metal switching?
That there are companies willing to give up margin to take a meaningful piece of the Ethernet switching pie is the real difference maker. Depending on the game of pricing chicken, it could materially change the margin structure for the entire industry.
Interestingly, this creates challenges for companies who both want to compete on price and be acquired. While it might be possible to buy market share, this creates a less lucrative business for would-be acquirers, because it could mean buying into a large but low-margin business. So companies have to be a bit careful when setting price. In an industry where pricing follows a cost-plus model, this is very strategically important.
The likely response to pricing and margin pressure
The first thing a dominant incumbent will do is squeeze the supply chain. There are points of margin to be had by taking advantage of shipping volumes. This means that Cisco should be able to drop price to stay in the ballpark (they have to be close, not the best) with those who would compete primarily on price. But what happens when the system has been optimized?
Cisco can essentially do three things:
- Do nothing and cede market share – This works (and is what they did with Arista) so long as share movement is small. If the threat to business is high, this is no longer a viable option.
- Exit the space and retreat to higher-margin markets – With so much attention paid to classic disruption strategies (thanks, Clayton Christenson), companies are loathe to abandon their cash cows lest they get painted into a deteriorating business strategy.
- Drop price and compete – The only real option in the face of significant pricing threats is to stay and fight.
Dropping price but maintaining margin
To compete in this type of climate, Cisco’s volumes will allow them to drop price and still maintain healthier margins than anyone else in the space. But if they have to go lower than their volumes can support, what will they do?
Strategically, the next thing to do is to look at where you are shedding margin. The biggest contributor after actual development costs? The channel. For each deal that Cisco wins through the channel, they lose points to fulfillment, support, and services. If you are Cisco and you are looking for where to claw margin back, the channel is absolutely the next target.
In fact, the success of endeavors like VCE and Exadata prove that handling integration yourself and selling direct is a lucrative model. This is likely why Chambers talks about becoming the biggest IT company (not just networking). It reflects the re-verticalization of the IT stack, and the re-emergence of a direct sales model.
The impact on resellers
Cisco is unlikely to go after their largest resellers. This leaves the smaller guys exposed. Their business will either be gobbled up by a smaller number of large resellers (operating themselves at lower margins), or taken by Cisco directly. Either way, Cisco wins.
Resellers should be looking at the current climate with a very wary eye. As a result of this shift, there will absolutely be resellers that fail and go out of business entirely. Watchful resellers should be planning their moves deliberately.
How do resellers respond?
The first thing to do is open up the product catalog. But this has to be done carefully.
Simply adding competing products to the catalog threatens to upset the apple cart prematurely. Instead of adding general networking products to compete directly with Cisco’s mainline products, consider offering new niche products. Very narrow use cases that deliver value in ways that Cisco’s mainline products cannot represent a lower-risk means of opening up the catalog without biting the hand that feeds you.
These niche solutions then form the foundation for a mitigation strategy when Cisco comes knocking looking for more margin.
The bottom line
These dynamics should make the channel a vital part of the upcoming war for datacenter networking. While the discussion now is on switch pricing, the real battle is going to be over margin. And it won’t be waged just inside vendor walls; the reseller space will be pulled in. Alliances are forming, troops are moving, and those who see it all will be in the best position to command their armies.[Today’s fun fact: While chopping onions, hold a piece of bread between your lips to keep your eyes from watering. Minimally, it gives you something to chew on.]