|Shares Out. (in M):||46||P/E||0||0|
|Market Cap (in $M):||1,342||P/FCF||0||0|
|Net Debt (in $M):||360||EBIT||0||0|
|Borrow Cost:||General Collateral|
I recommend shorting Cogent Communications (ticker CCOI), a provider of Internet traffic transit services. Cogent’s Net-Centric segment (44% of revenue) is facing an increasing disintermediation headwind as some of its key customers choose to connect directly with each other rather than use Cogent’s network. Moreover, Cogent barely generates FCF (LTM FCF yield is negative), and its stock price has been supported by hefty dividends and share repurchases that have entirely been funded by increases in net debt.
CCOI has decent liquidity (avg daily trading volume >$15 million) and is available to borrow at general collateral rates.
Through its network of leased and owned fiber assets, Cogent provides Internet transit to two types of customers:
For a fuller (and better) description of Cogent’s segments, read jdr907’s CCOI write-up from January 2015.
Cogent’s Net-Centric business is getting tougher:
As explained by a Level 3 executive in 2011: “In the early days of the Internet, to reach the Internet and each other, content providers and ISPs were in large part dependent on purchasing IP transit from Tier 1 backbone providers which acted as middlemen, exchanging their customers’ traffic between themselves through peering. Today’s ISPs and content providers have much more choice.” To reach the end user, a content provider can purchase IP transit through a third-party provider like Cogent, it can connect directly with the ISP, or it can pay a Content Delivery Network (CDN) to store content close to the ISP and end users, bypassing backbone transit providers completely:
I believe that content providers are increasingly availing themselves of the alternatives to purchasing IP transit through providers like Cogent, as evidenced by decelerating growth in Cogent’s Net-Centric revenues:
The bull case rebuttal is that this deceleration is a temporary result of certain ISPs refusal to upgrade their port connections with Cogent.
By way of background, usually Tier 1 Internet backbone networks agree to “settlement-free peering,” whereby they send and receive traffic from each other for free. The logic here is that the data sent from one network to another is usually roughly within balance. In Cogent’s case however, many of its high bandwidth Net-Centric customers (think Netflix) send large volumes of traffic through Cogent’s network to the ISPs such that the volume of traffic between Cogent and the ISPs is not remotely within balance. The ISPs don’t like this, as they are forced to spend money to upgrade their connection capacity with Cogent while not being able to charge Cogent’s clients for the incoming traffic. As a result, several ISPs stopped upgrading their connection capacity with Cogent’s network. The resulting bottlenecks worsened Cogent’s value proposition and drove some of its customers to choose alternatives to using Cogent’s network (ie, direct peering with the ISP and/or use of CDNs).
In February of this year, the FCC placed the interconnection issue under Title II jurisdiction, which means there is increased regulation of such peering arrangements. Likely in response to this, Verizon and AT&T, which were among the ISPs that had stopped upgrading their connection capacity with Cogent’s network, subsequently signed interconnection agreements with Cogent that should alleviate the bottlenecks.
While the newly signed interconnection agreements are a positive for Cogent, I’m skeptical that they will arrest the deceleration Cogent is seeing in its Net-Centric business. Cogent has disclosed that roughly 5% of its revenue (implying ~10% of its Net-Centric revenue) over the last year has been stunted due to port congestion. I think the amount of deceleration in constant currency Net-Centric revenue growth (from nearly 10% at the beginning of 2013 to 3% so far this year) is too large to be explained by headwinds impacting just 10% of the Net-Centric revenue base. Rather, I believe the increasing trend towards direct connections between content providers (Netflix, Facebook, Google, etc) and ISPs is a secular headwind for Cogent that is not going away.
Cogent barely generates free cash flow:
If you briefly skim Cogent’s financial statements, you might think that the Company generates much more FCF than it actually does. If you just take “Net cash provided by operating activities” and subtract out “Purchases of property and equipment,” it looks like Cogent generated $36 million of FCF in 2012, $33 million in 2013, and $13 million in 2014. However, this significantly overstates Cogent’s FCF, because much of Cogent’s capex does not show up as “purchase of property and equipment.” Instead, much of Cogent’s capex is funded via the incurrence of capital lease obligations. Explains Cogent in its 10-K: “The Company has entered into lease agreements with numerous providers of dark fiber primarily under 15-20 year IRUs typically with additional renewal terms. Once the Company has accepted the related fiber route, leases that meet the criteria for treatment as capital leases are recorded as a capital lease obligation and an IRU asset.”
Cogent discloses its capital lease obligations occurred in the supplemental disclosure section at the bottom of the cash flow statement. This item represents a real impact to cash flow. While Cogent might not have to fork over the cash immediately upon incurrence of a capital lease, it does have to fork over the cash at future dates over the life of the lease. The interest portion of capital lease payments are included in cash flow from operations, but the principal payments show up in the “Cash flows from financing activities” section of the cash flow statement.
To calculate Cogent’s FCF, I contend that you have to either include capital lease obligations incurred or include principal payments on existing capital lease obligations. If you do so, FCF generation looks much worse. In the graph below, the red line represents LTM FCF/share if we exclude capex funded via capital leases. In recent years, FCF/share under this calculation method has ranged from ~$0.30 to $0.80; not great considering CCOI’s $29 stock price, but it’s something. The blue line represents LTM FCF/share if we account for capex funded with capital leases by subtracting out principal payments on capital leases. The green line represents LTM FCF/share if we account for capex funded with capital leases by subtracting out capital lease obligations incurred (as well as “PP&E obtained for note payable”). Under these two methodologies, CCOI FCF is actually running negative currently:
This may seem like an obvious observation, but I’m not sure all market participants are aware of this. Several sell-side analysts calculate FCF in a manner that ignores capex funded via capital leases, and jdr907’s January 2015 VIC write-up on CCOI makes the same mistake.
Cogent ‘s stock price is supported by unsustainable dividends and share buybacks:
Over the past year, CCOI has spent $63 million on dividends and other $53 million on share buybacks. This has entirely been funded by increased borrowing, however: