Cisco recently has very visibly admitted that recent financial performance fell short of their goals and announced a diverse set of actions to remediate the problems including shutting down the Flip camera business, reorganizing the engineering management (with the related departure of the Nuova senior team) and greatly simplifying the matrix management structure put in place over the last few years, an early retirement option and the related goal of reducing OpEx by $1B/year over the next year.
Three issues have been identified by Cisco:
- Weakness in the public sector
- Transitions in the switching business
- Issues with the “matrix” management model of Councils and Boards
Given the worldwide financial state it’s easy to believe that Cisco’s public sector business has been impacted, but it seems unlikely that this would motivate reorganization or the dropping of specific products by itself. I think it’s an important issue but probably not the cause for organizational restructuning nor the broad cost cutting Cisco has projected.
Innovation is vital for Cisco. It has been clear for years that the organic revenue growth in the traditional switching and routing markets wouldn’t drive revenue adequately to support a desired market valuation (the source of the 12%/year goal), especially given the predictable and quite dramatic commoditization of these traditional markets. As a result Cisco has aggressively explored and implemented diverse means of finding and developing new markets including (in additional to organic innovation and M&A activity) internal search for and funding of new ideas, external contests for the same, and an innovative refinement of what has been called matrix management in the past, which we’ll collectively call “councils” here. The essence of councils is the delegation of planning, strategy and budget activity from a traditional, top-down, command and control model to a broader set of committees incorporating a larger set of leadership. Cost organizations like engineering and marketing received funding through these councils. The councils were constituted both to broaden the decision making community and to force the collaborative cooperation between the various requesting organizations. Chambers first constituted councils ten years ago to address the cut-throat competition between different Cisco routing and switching product groups, and then over the last few years broadened the concept so that a much more diverse set of strategic opportunities could be driven.
Why change this now? There are a set of reasons. Although the structure was effective in accelerating innovative and entrepreneurial activity within the company and developing businesses that have contributed significantly to Cisco’s overall growth, Cisco learned that managing innovation is hard work, and that succeeding in new markets is often more difficult than was anticipated and requires more remedial management attention than expected (Flip, for example). But more seriously, the complexity of the council discussions (the necessary but probably not anticipated side effect of broadening the decision making community greatly) created a lot of “noise” in how resources were used and how basic engineering decisions were made. The result was increased cost, sufferable in good times but not sufferable in today’s challenging times and a less than optimal response to challenges in the core businesses, specifically switching because of delayed and defocussed management of these core products. So a decision was made to cut back on these structures greatly in order to simplify operations, reduce unproductive cost and simplify and accelerate decision making in the core businesses.
The most serious issues are in the switching market, the core of Cisco’s revenue and profitability. Chambers speaks often market transitions. John’s philosophy has always been to anticipate and respond to market transitions rather than focusing on competitors (“skate to where the puck is going”). There are important switching market transitions, to be sure, but things are considerably more complicated. The important market transitions underway include the virtualization of data centers and the transition from 1G to 10G Ethernet. But the situation is much more complex:
- When Cisco introduced the UCS integrated server and switch solution it created an unanticipated industry shit-storm in the data center business. Prior to the UCS announcement, HP and IBM – the leaders in data center servers – had often included Cisco switching in a procurement, if that is what the customer wanted (which was often the case given Cisco’s market leadership). Cisco believed that the introduction of UCS would cause HP and IBM to stop selling Cisco switches (but not stop the customers from specifying them) but what happened was considerably more dramatic. HP acquired 3COM and a quite valuable switch portfolio from the H3C JV between 3COM and Huawai. HP not only stopped selling Cisco switches but aggressively sold HP switches instead, often at a considerably lower price than the Cisco alternative. IBM pursued a similar path through an intensified partnership with Juniper. Considering that data center procurements are much more server driven than network driven, this was a problem.
- The problem is compounded by the virtualization of the data center, an initiative that is transforming data center architecture rapidly. This is the wrong time to have a new go-to-market problem because more of the procurements are “green field” and fewer incremental refresh of an existing network.
- The trend to virtualization is also driving evolution to lower cost infrastructure following a model developed and proven by Google, Amazon, Microsoft and the other huge data center operators. Cisco has always had some of the best products but not necessarily the cheapest products. Now the market requirements are shifting and price is a greater issue.
- Finally, merchant silicon parts that can be used to build high performance switches are getting pretty good as companies like Arista are demonstrating. Traditionally custom silicon parts delivered differential product value and Cisco’s large market share gave it an advantage in terms of silicon design competence and custom engineering cost compared to all the other smaller network equipment vendors. Using merchant vendor parts levels the playing field and makes software design an increasingly important part of switch development.
In the most recent financial call Chambers noted that Cisco has a proven history of producing market share leading, high-margin products and he expressed confidence that would happen after Cisco fully responds to these transitions, having solved the organizational “noise” and refocused the switching engineering effort. However, responding to follow up questions John admitted that there was no easy way to balance revenue and margin, especially in today’s switch market with rapidly increasing customer “value” (a polite way of saying rapidly decreasing prices).
When responding to another question John acknowledged the rapid “Moore’s Law” progress of the merchant vendor parts but felt sure that Cisco could do as well with their internal ASIC’s. This is a critical question. In the computer business merchant vendor parts (specifically Intel’s X86 architecture microprocessors) ultimately completely transformed the business. It’s all about part volumes. Suppose that today Cisco’s custom ASIC part volumes are greater than the combined volume for an equivalent merchant vendor part. In that case the NRE (the design cost for the part) amortized over the production is lower for Cisco preserving a basic advantage. But we know that some of the most modern Cisco switches (e.g., the Nexus 3000) use merchant vendor parts at the core, and there are industry rumors of a Nexus 7000 class switch (“Jawbreaker”) being also done of with merchant vendor parts (in both cases, Broadcom). If Cisco starts to use merchant volume parts in volume than the “volume curve” advantages could swing pretty quickly to the merchant vendor side. When that happens all bets are off, and quickly the differentiating focus becomes things like software, which it is fair to say historically has not been a Cisco technology strength.
So netting that all out, here’s what might be said:
- Cisco has a revenue issue in the public sector and in the switching market more broadly. In neither case is it a simple product or execution problem (although both exist in the case of switching). The switching case is the more challenging because it reflects a steep price/performance improvement which makes revenue growth in many segments difficult or impossible.
- These background challenges highlight some of the untended consequences of Cisco experimentation with a Nuevo-matrix management scheme called councils. The councils were put in place to accelerate innovation, which they did, but at what turned out to be a cost in resource redundancies and basic engineering inefficiency that is no longer tolerable given these core problems. As result, the council structure has been dramatically cut back and a more straightforward command-and-control structure put back in place.
- Cisco is confident that it can remediate the product shortcomings in switching with the simpler management, and that Cisco can maintain market share and gross margin in the switching business. But there is no obvious solution to the issue of decreasing price in the market and this is almost surely going to be reflected in a diminished Cisco growth rate and a renewed search for ways of broadening the product offerings and increasing growth, but going forward without creating the noise and inefficiency that the council system did.
- How Cisco increases stock value (beyond the ongoing stock buyback scheme) fares through this all is an unanswered and important question.
Cisco has always been an extremely competent and dynamic business, successfully responding to wave after wave of market opportunity and change. I wouldn’t want to bet against them this time, but the challenge is greater than in the past because of the combination of the overall market slowdown and accelerating price reductions in the core switching market.