|Shares Out. (in M):||5,456||P/E||11.0x||0.0x|
|Market Cap (in $M):||91,743||P/FCF||8.9x||0.0x|
|Net Debt (in $M):||-32,043||EBIT||9,150||0|
Cisco is a great company. It is dominant in a large and growing market, albeit a mature one. It sports very high ROIC (>100% little book definition, and 40%+ including intangibles) and cash flow conversion (>100% net income). It has a rock-solid balance sheet. And in my opinion, it is significantly undervalued (6.6x Ebit, 5.2x Ebitda, 5.7x Ebitda-Capex –all TEV/ltm metric).
+ Cheap on an absolute and relative basis
+ Substantial scale and scope advantages
+ Leader in a market with net favorable secular growth trends
+ No imminent threat to its market dominance
+ Top notch management team, whose interest is well aligned with that of shareholders
+ Rock-solid balance sheet ($32b net cash)
+ Strong cash flow generation, majority of which goes to share repurchases and dividends
+ Continued share repurchases to meaningfully contract share count in light of a rapidly diminishing stock of options outstanding
CSCO is the 800lb gorilla in the switching & routing market. As such, it enjoys significant economies of scale and scope. In my view, the most significant is the amount of resources it can afford to channel every year into R&D. This (and its acquisition capabilities) is what enables CSCO to stay at the leading edge of technology. Importantly, its R&D budget is a multiple of all of its competitors combined:
R&D expenditures (most recent fiscal year)
HPQ $0.7-0.8b (estimate based on rev, op margin, and 3com historical)
Its installed base ($200b, 50m devices, 6m annual customer interactions) and partner/distributor ecosystem is unrivalled. More IT people are familiar with CSCO, standards setting, “you won’t be fired for buying CSCO”, etc.
Services are not only a fast-growing and high-margin recurring revenue stream but also an avenue to further its competitive position.
Strong competitive position
CSCO has 70%+ market share. HPQ and JNPR are relatively distant #2 and #3 respectively. Further down the food chain, you see BRCD and other niche players. In Gartner’s magic quadrant matrix (Enterprise LAN), CSCO is clearly the leader. The leader in this matrix is the most visionary with the strongest execution capabilities (and not necessarily the largest player). HPQ is the only other company in the “leaders” quadrant. Tellingly, the whole matrix is pretty empty. It seems to me that a mostly empty matrix combined with HPQ as your major competitor is a relatively benign situation.
Despite recent hiccups, CSCO has not really ceded market share. If you compare the growth rate over the last few years of the four major players, it is hard to argue that CSCO has lost market share.
The service provider (e.g. telecom) segment’s competition includes Huawei, and ZTE. These two Chinese companies are relatively new entrants. CSCO recognizes them as aggressive price competitors in some markets. In the long term they could become a not insignificant threat.
Despite meaningful competitive pressure, a relatively recent product transition, corporate restructuring, and not insignificant exposure to Europe, fy12 will be a record year for CSCO.
Strong corporate governance
+ The CEO, John Chambers, is the single largest individual shareholder.
+ For all practical purposes, the company stopped granting options four years ago. Instead, most of management’s compensation is in the form of restricted stock (>90% fy11). For example, the CEO’s salary is $375k. The remainder of his compensation is in the form of restricted stock ($12.5m fy11). This, obviously, is a much better way to align management’s interest with that of shareholders.
+ It initiated a dividend in fy11 and increased it in fy12.
+ It has repurchased large amounts of stock.
A mountain of options is mostly in the rearview mirror
One common criticism of CSCO is that its share count has not decreased in proportion to its massive share repurchase program. It is a fact that dilution has mitigated to a significant extent the benefits of share repurchases. However, this is mostly in the past. Over the last few years, the company has had to issue large quantities of stock to satisfy the mountain of options that resulted from the late 1990s/early 2000s collective delusion that argued that options/stock-based compensation is not an economic cost. The number of options outstanding peaked in 2006 (~23% of diluted share count). With the elimination of option grants (and a lower share price), it has naturally seen a steeper decline since July 2008. As of April 2012 options outstanding are <10% of diluted share count. Looking forward, about half of the options outstanding as of July 2011 expire within the next year or so (and the last batch of options expires in 2016). Notably, all of top management’s options outstanding have exercise prices at least 10 percent higher than the current stock price; and the vast majority above $20. Clearly, a higher stock price is in the interest of senior management. And for all practical purposes, management “won’t see a dime” before a long position at the current price sees a nice ROI.
A layman’s view of secular trends
Perhaps the two most important trends in the switching & routing market are exponential growth of data traffic and cloud computing. Obviously, data traffic growth is a positive. It seems to me that cloud computing is a net positive for market leaders. On one hand cloud computing adds nodes (data centers) to the network. This is a positive because each new node needs equipment, a meaningful portion of which is high-end equipment. On the other hand, data centers not only purchase and use equipment more efficiently but also reduce the equipment intensity of non-datacenter nodes. In summary:
+ increased volume,
+ premium on leading edge technology
- pricing pressure,
- less favorable mix in aggregate,
Why is it trading at these levels?
There has been a meaningful adjustment to growth expectations and some gross margin compression. Coming out of the depths of the 2008/2009 financial crisis, management was targeting mid-double digit+ revenue growth. Recently the growth rate was lowered to 5-7%, for at least for the next couple of years. The European crisis and weakness in the public sector are the main external factors behind this downward adjustment. Aggressive competition primarily and a product transition are the main factors behind lower gross margins. Management expects the competitive environment to remain aggressive. In turn, it has adjusted its business model to 60-62% gross margin, compared to the 64%+ of the previous decade.
Naturally the stock has been punished. I don’t think CSCO was priced to grow 15%+ prior to the downward adjustment in guidance. Notwithstanding, the result of punishment on top of an undemanding valuation has created the current opportunity.
Importantly, the company has mostly completed a restructuring initiative centered on refocusing on its core markets and lowering it expenses. As a result, operating margins have moved upwards and are pretty much of par with those preceding the financial crisis. The same is true for “ebitda-capex” margins.
A word on relative valuation
In my view, CSCO is clearly undervalued in absolute terms. Its FCF yield is 17% ($10b cash from ops-capex/$60b TEV). It is also inexpensive on relative terms. It trades at a multiple (of ebitda-capex) that is a fraction of JNPR or BRCD. Compared to other large tech companies, it is trading at a similar level to HPQ on an ebitda-capex basis, and at a huge discount on FCF yield. This is clearly myopic. The quality of the company, its competitive position, the competitive pressures, etc are completely different. I would argue that a better comp would be INTC; although it is not necessarily cheap and faces much more significant competitive threats.
It seems to me that taking it all into consideration, fair value in this market is somewhere around $25. This would still be below 10x ebitda-capex and FCF/TEV ~10%.
In terms of macro downside, even if the euro went to US$1.0, I don’t see how the cash generation power of this company would warrant a price below $20.
Key man risk. I think John Chambers is a top notch CEO. While the record is not stellar, in balance I think he is in the top 2% or so of CEOs worldwide in terms of bang for the buck. It is hard for me to name another non-founder/owner CEO of a sizeable company whose cash compensation is <$500k and the rest is restricted stock. For instance, I have a hard time finding a better CEO among the DJIA components, especially adjusting for amount and structure of compensation.
Service provider consolidation is a negative; especially in accounts susceptible to competition from Huawei et al. If one considers significant further service provider consolidation a meaningful risk, then shorting AMT could well serve as a hedge to this risk.
Most of the cash pile is outside the US and subject to tax if repatriated –an issue common among large tech companies.
Consistent execution in the form of earnings and guidance (Aug 15 - Q4 and FY12 earnings).
|Subject||Random competitive thoughts|
|Entry||06/26/2012 06:18 PM|
While I am not very close to Cisco in particular, I have spent a lot of time inside data centers as a quasi insider. They always consider Cisco equipment as incredibly expensive for what you get. It's also very hard to program compared to a server.
Cisco servers, while not material to sales, are incredibly expensive compared to say a Dell server.
From a broader perspective, the article below from a top guy at Amazon's data center group lays out an argument that Cisco is basically selling mainframes, as in expensive and proprietary equipment, while other hardware has been commoditized. The article is dated, but it does point out a long-term risk to the story.
There is also the threat from not just cheaper routers, but actually fewer routers. Like server virtualization has greatly slowed server unit growth, network traffic virtualization (WAN optimization) is hitting mainstream, and offers good consolidation of IP (internet) traffic.
Finally, the Infiniband market (Infiniband is a competitor to Ethernet, which is the switching tech used mostly by Cisco) is dominated by Mellanox, which shows that new angles can be taken by relative newcomers. I am pretty familiar with Infiniband, which I don't think will be super mainstream since it caters to very high performance users at a high cost, but they are targeting Cisco's market at the edges, and if you look at its stock price (MLNX), you can guess they are doing pretty well taking some share. since MLNX since IPO in 2007 is up 230% while Cisco is down 38%.
I don't have an opinion on the company valuation nor on any immenent threats to Cisco's dominance, but just some thoughts.
|Subject||Agree / thoughts|
|Entry||06/27/2012 09:58 AM|
Good write-up. This is one of my larger positions. The way I see it, the market is pricing this for a 5% perpetual decline in earnings, using a 10% discount rate. I find it hard to imagine that despite all the knocks on the company, that it will indeed shrivel 5% per annum forever. (And if it did, your IRR would still be 10%.)
On the other hand, to edward's point and to quote Greenwald, in the long run these routers are toasters, which is to say that they become commoditized. Such is the nature of tech. But with good management the company can continue innovating and growing its customer relationships.
|Subject||RE: RE: Agree / thoughts|
|Entry||06/28/2012 06:59 AM|
It's also important to keep in mind that there are widely varying profiles of buyers for equipment, with varying levels of price/performance sensitivity. Many buyers are highly risk averse, and get enormous comfort from the vastly superior post-sale support that an organization like Cisco provides. If you listen to Chambers' comments on recent calls regarding the (favorably) shifting landscape with respect to customers that may have initially favored low cost alternatives like Huawei, but are now realizing that when things go wrong they don't get anything near the support that Cisco provides, it gives some comfort in this regard.
Yes, there will always be customers like data centers (who need to push the edge on price/performance), or Google (who have both the internal know-how to handle their own aftermarket issues, as well as push the edge on design), who may shift towards other providers.
However, there is massive demand from customers without these internal capabilities for whom an IT outage of even moderate length is far more costly than the difference in gear pricing. Whether or not Cisco sells a more reliable product (which they would argue they do) is almost irrelevant; what they sell is the concept of a more reliable product backed by aftermarket support that is clearly superior.
This dynamic is a big reason why Cisco has such a massive presence in the public/government vertical. It's full of IT directors who are far more worried about job security and/or possibly being called to task for that one catastrophic data outage than they are about taking credit for a 10% increase in price/performance stats as part of a budget that isn't terribly meaningful in the grand scheme of their department's overall P&L.
I think about the job of IT director the same way I think about the job of fund of funds manager. It's not about shooting out the lights. It's about keeping your head down and making sure you don't fuck things up.
|Subject||RE: RE: Random competitive thoughts|
|Entry||06/28/2012 11:25 AM|
I don’t personally think Infiniband (IB) is ready for mainstream data centers due to cost and ease of use compared to Ethernet, but it can still grow a lot in specialty applications, hence I brought it up as it could chip away at a profitable margin. Still, Larry Ellison of Oracle said, “InfiniBand is by far the fastest and most efficient switch fabric for running enterprise data centers.”
Regarding your observations that Ethernet is already as fast as IB and will soon be much faster, what chart are you referring to? If you are referring to page 8 of the June 2012 Mellanox Corporate Update which shows 10GigEthernet to be 75% efficient vs. 90% for Infiniband, that is a HUGE difference. Means that (basically) if you buy 10,000 servers, under 10GigE then 2,500 will be idle at any one time while only 1,000 will be idle with IB, and by idle I mean that latency in the network holds up the CPU which is idle since it already finished its tasks and is waiting on another task. Or put another way, you’d need to purchase 15% more servers/storage to have the same performance. That said, that is specific to High Performance Computing, which is much different than mainstream and I'd expect an unvirtualized typical web server to show zero difference.
Also, IB is not for connecting with the Internet, hence only applicable for internal data center switching – but not sure what % that represents. Mainstream data centers don’t have too much internal traffic, but High Performance (especially) has it, virtualized and cloud also have it to a real but lesser extent. In these cases, IB is many times faster than Ethernet. That is distinct from saying “Ethernet and IB both have 40 Gigabyte per second pipes” because that just refers to how much data arrives at any one time, it doesn’t specify how long it took to get there. An analogy would be saying that 1,000 cars drive through a town every hour in each of two lanes, but in the IB lane the cars took 10 minutes to get from the last town, while in the Ethernet they took one hour. That latency matters a lot if a compute unit has sent of a request for data, and sits idle until it comes back. However, mainstream data centers are typically way overprovisioned, so things are idle all the time so adding IB would make no difference.
Also– you asked about pricing. I don’t unfortunately have that, that is just based on offhand comments from data center folks. I did price out Cisco’s servers maybe 18 months back, and I seriously think they were double what I could get at Dell. No one who knows Cisco well is surprised that their servers cost that much, and nor have they sold any that I am aware of.
Overall – my gut would be that Cisco is like IBM server hardware was in the 1980s- still a/the dominant player, but will slowly lose share and margins over time due to both current and future threats. No idea if the stock price already has priced in enough, per your -5%/yr comment of what is priced in. At one time people said “you can’t be fired for buying IBM”, which was and is still true, but since that statement was made I don’t think IBM’s hardware department has done great.