Description
Sirius Satellite Radio is one of two competitors in the satellite radio industry, the other (and far stronger) player being XM Satellite Radio (XMSR). The investing thesis that I will present here suggests a short sale of SIRI on the following merits:
• Sirius is a poorly managed company prone to both financial excess and the real risk that it will not meet its announced projections of 1 million subscribers by year-end 2004 and cash flow breakeven by 2005.
• Sirius is on track to burn through its existing cash by year-end 2004, and will need to refinance before then, which will result in additional dilution to current shareholders
• Assuming the financing window closes to Sirius between now and 2005, the company could face a cash crunch and potentially have to re-structure as it did in early 2003, when bondholders essentially bought 91% of the company for $450 million.
• XMSR is the clear, runaway winner in a two-horse satellite radio race with SIRI.
Further, while I am loathe to consider purchasing shares of a stock without a margin of safety in the stock price, I believe it advisable to hedge the Sirius short either fully or partially with XMSR shares in order hedge out industry-specific risk and to profit from the current quotational mis-pricing between XMSR and Sirius, which values a Sirius subscriber at approximately five times an XMSR subscriber (just to use one potential valuation metric.)
SIRI currently has a short interest of about 5% of the float, while XMSR has a current short interest equal to 33% of the float. This information makes a pair trade more interesting given that SIRI is massively overvalued relative to XM, but also in that a SIRI short doesn't risk a squeeze, while XMSR could easily benefit from one. More thoughts on the possible pair trade idea will presented at the close of this essay.
Sirius and XM are the only two competitors in digital satellite radio. Sirius owns three orbiting satellites from which the company broadcasts more than 100 channels, or “streams” of digital quality audio throughout the continental U.S. The company’s revenues are primarily from subscription revenue at $12.95 per month per subscriber. At December 31, 2003, the company reported that it had 261,061 subscribers, up from the December 31, 2002 report of 29,947 subscribers.
Sirius has agreements with Ford, DaimlerChrysler, BMW, Nissan, and Volkswagen to sell vehicles with Sirius radios installed. The agreements with DaimlerChrysler requires Sirius to share a portion of the subscription revenues received from subscribers via DaimlerChrysler, and Sirius reimburses DC for advertising expenses and hardware costs. Sirius also issued to DC a warrant to purchase 4 million shares of SIRI stock at $3 per share. This agreement extends to May 2007 unless terminated earlier. The agreement with Ford is substantially similar. Ford also has a warrant to purchase 4 million shares of SIRI at $3 per share, exercisable based upon certain performance conditions, and is fully exercisable after 1.5 million Ford vehicles carrying Sirius radios have been manufactured. The agreement extends to October 2007 unless terminated earlier.
The agreements with Volkswagen, Nissan and BMW also provide for revenue share and cost reimbursement. No warrants were issued to either vehicle manufacturer. Nissan and Volkswagen also have agreements with XM Satellite Radio.
Finally, Sirius has an agreement with Hertz to make Sirius radios available as a premium option to rental car customers. Sirius will own the radios and will install and service the radios. The service will be offered to rentals for a daily fee, a portion of which is shared with Hertz. At year end 2004, 24,011 of Sirius’s reported “subscribers” were actually Hertz rental cars, which provided an average revenue per month of $3.13 (versus average subscriber revenue of $12.02 for the rest of SIRI’s subscriber base.)
The CEO of Sirius is Joseph Clayton, best known for running Global Crossing into bankruptcy by massively spending in a land grab for market share for underwater cable. He appears to be following a similar operating strategy for Sirius. The company’s profligate spending quickly burned through the $1 billion in capital it had raised, forcing it to re-structure its debt in a recapitalization in March 2003, described in the company’s 2002 10-K filing, in which shares outstanding increased from about 77 million to over 911 million.
As of December 31, 2003, Sirius had increased its deficit to approximately $1.15 billion and reported net losses of $314.4 million, $468.5 million, and $235.8 million for 2003, 2002, and 2001, respectively. Operating expenses increased to $450.4 million for the fiscal year 2003 versus $313.9 million in 2002 and $168.5 million for the year 2001.
After its re-structuring in March 2003, the company lucked into a forgiving capital market environment where investors were once again eager to shovel cash into anything speculative, an environment that Sirius was quick to exploit:
• In May 2003, SIRI issued $201 million in 3.5% convertible notes due 2008, resulting in net proceeds of $194.2 million. The notes are convertible into shares of common stock at any time at a conversion rate of 724.6377 shares of common stock for each $1,000.00 principal amount, or $1.38 per share of common stock, subject to certain adjustments.
• In June 2003, SIRI sold 86,250,000 shares of our common stock in an underwritten public offering resulting in net proceeds of $145.5 million.
• On November 19, the company raised an additional $150 million through the sale of 73.7 million shares at $2.10 per share, and reduced debt through an exchange of company stock for $65 million in outstanding convertible notes.
At year-end 2003, Sirius reported cash and marketable securities of $550 million and long-term debt of $145 million. Two events since year-end will impact Sirius’s capital structure:
• Immediately after year-end, Sirius announced an exclusive seven-year agreement to broadcast NFL football games and to create the NFL Radio Network, which will broadcast NFL content 24/7. Sirius intends to offer this content to its subscribers at no additional cost. While this content may indeed provide Sirius with some differentiation from XM and may help attract more subscribers, Sirius is paying a lot for it: $188 million in cash during the term of the agreement, of which $10 million was paid on closing and $85 million to be deposited in escrow. The remaining $$93 million would be due sometime after August 15, 2009. In addition, SIRI issued $32 million in stock to the NFL, and warrants to purchase an additional 50 million shares at $2.50 per share. These warrants are exercisable upon the satisfaction of certain performance criteria by the NFL.
• On Friday, February 13th, Sirius sold $250 million in five year, 2.5% convertible notes with a conversion price of $4.41 per share.
SUBSCRIPTION GROWTH AND REVENUE TREND
Sirius estimates that it needs 2 million subscribers to achieve cash flow profitability, a figure which is derived by taking $12.95 per subscriber per month and annualizing it to $155 times two million subscribers, which equals $310 million in annual revenues. Obviously, this would also require the company to achieve an annual operating cost strucuture of less than $310 million. The company’s 2002 10-K (and most recent press release dated February 12, 2004) states that it expects to reach this level by year-end 2005.
Since the 2 million subscriber mark at $12.95 per subscriber is the magic number, let’s look at subscriber growth in 2003. Here is the subscriber data provided by Sirius in its Q4 press release:
Quarter Starting Subs Ending Subs
Q1 2003 29,947 68,059
Q2 2003 68,059 105,186
Q3 2003 105,186 149,612
Q4 2003 149,612 261,061
One thing that is clear about the Sirius subscription acquisition numbers is that the Hertz rental agreement is responsible for 22,067, or almost 10%, of the new subscribers acquired in 2003, and comprise 24,011 (9.2%) of the total subscriber base. These subscriptions offer only a fraction of the economics of real subscribers – the company reported average monthly revenue from Hertz subscribers were $2.65 per month in Q4 versus $11.99 (before effects of rebates and hardware) for the rest of the subscription base. The bottom line here is that I’m suspicious that the Sirius will be able to accomplish the $155 yearly subscription revenue per subscriber. In addition, I am deeply skeptical that SIRI will be able to reach anything close to the 1 million in subscribers they project for YE 2004 and the 2 million they project by YE 2005. To reach the 1 million subscriber number would mean that they will have to add 738,936 subscribers in 2004, or 184,734 per quarter. In 2003, the company added a total of 207,103 non-Hertz subscribers. In addition, I am convinced that satellite radios / subscriptions will be a hugely seasonal business, with a very high percentage of new subscribers coming on during the Christmas season. XMSR’s subscription patterns seem to confirm this – XM added 145,000 subscribers in Q4 2002, but then only added 135,000 in Q1 ’03, 209,395 in Q2, 209,178 in Q3, and then exploded for 430,580 subscribers in Q4 ’03.
It strikes me as incredibly aggressive to expect to average 184,734 subscribers per quarter in 2004 when they failed to reach even 75% of that number in the seasonably strongest quarter just past. If XM’s pattern from last year holds, I would expect SIRI to have a seasonally slow Q1 2004, likely with fewer subscribers than in Q4, then have two slightly better quarters followed by a blow-out Q4. My most optimistic estimate for SIRI would be 100,000 subscribers in Q1, 150,000 in Q2 and Q3, and then 200,000 in Q4, for a total of 600,000. This number does not take into account any churn in the customer base. It appears to me that there is the potential for significant churn, given that many SIRI subscribers are on some sort of promotional or trial plan, and that many may choose to discontinue the service when the monthly bill starts to get on their nerves or the novelty wears off. In any case, I am assuming that given an average subscriber base that is less than three months old, we have no idea how sticky this service will be. SIRI better hope that it’s REAL sticky.
My concerns about subscriber acquisition, average revenue per subscriber, and customer churn notwithstanding, let’s grant SIRI the 231,000 subscribers at full $155 yearly value. Let’s see how long it would take SIRI to reach the 2 million subscriber mark at various customer acquisition rates (given that they had 1,769,000 to go at 12/31/03.)
At 231,000 per year 7.65 years Q3 2011
At 300,000 per year 5.90 years Q4 2009
At 400,000 per year 4.42 years Q2 2008
At 500,000 per year 3.54 years Q3 2007
At 600,000 per year 2.94 years Q4 2006
At 800,000 per year 2.21 years Q1 2006
Even assuming that everything works out exactly as planned, the company has to add an average of 200,000 subscribers per quarter, EVERY QUARTER, for the next 9 quarters to break even, and that still won’t happen until Q1 ’06. At my more realistic (but still quite aggressive) 600,000 annual subscriber estimate, 2 million subscribers won’t happen before Q4 2006.
COSTS AND EXPENSES
Sirius has a massive cost structure – 2003 cash operating expenses (not including depreciation and non-cash compensation of $95 million and $11.5 million, respectively) were $345 million. The company’s Q4 and full year 2003 financial results are mind-blowing. Q4 revenues were $5 million; expenses were $130 million. Full year revenues were $12.9 million; operating expenses (including depreciation and non-cash stock compensation) were $450 million.
I have heard SIRI bulls go on about the scalability of the satellite radio business model. But anybody who reads the 10-Ks immediately will come to the conclusion that the majority of expenses aren’t fixed costs. Satellite and transmission costs may be somewhat fixed, but everything else increases with the subscriber base. Certainly one would hope that revenues would ramp faster than expenses, but my point is that SIRI’s expenses have increased quarter by quarter despite some efforts at cost cutting prior to the re-structuring. I conservatively estimate CASH operating costs at about $395 million annually, which does not include D&A, non-cash compensation, interest expense, or any unusual expense. Here’s the breakdown on the major costs involved:
Programming and Content
Programming and content expenses include license fees paid to third parties for music and non-music programming, costs associated with the production of music and non-music programming, costs of on-air talent, royalties for music broadcast and the costs of programming personnel.
Programming and content expenses increased from $9.8 million in 2001 to $22.7 million in 2002 and were $30.4 million for 2003. These costs will likely keep going up each and every year.
Satellite and Transmission Expenses
Satellite and transmission expenses consist primarily of personnel costs, in-orbit satellite insurance expense and costs associated with the operation and maintenance of our satellite tracking, telemetry and control system, terrestrial repeater network and national broadcast studio. The company expects that a significant portion of its satellite and transmission costs will remain relatively constant, and that increases or decreases in satellite and transmission costs will be due, in large part, to increased or decreased costs of insuring our in-orbit satellites.
Satellite and transmission expenses increased to $39.3 million for the year ended December 31, 2002 from $31 million for the year ended December 31, 2001. For 2003, satellite and transmission expenses decreased slightly to $32.6. Note, however, that the reason given for the reduction was “reduced insurance coverage.” For the year ended December 31, 2002 satellite and transmission expenses also included a loss of $5 million on the disposal of equipment in connection with the optimization of our terrestrial repeater network. By the way, that reduction in insurance coverage is not an insignificant issue – this company needs insurance on its satellites. Here’s why:
Sirius has three orbiting and one spare Loral FS-1300 satellites, which are designed to have a useful life of about 15 years from time of launch. According to the 2002 10-K, Sirius maintained in-orbit insurance policies covering in-orbit losses totaling $110 million per satellite, which is an amount sufficient to cover the launch of the company’s replacement satellite, but not sufficient to purchase a new spare satellite. Most importantly, the coverage for in-orbit failure was only for the first two years of operation for each satellite. Sirius notes that if it were required to launch its spare satellite due to an in-orbit failure of one of its three operating units, its operations would be interrupted or impaired for at least six months. If two or more satellites should fail, operations could be suspended for at least 16 months. In either event, particularly if it were to occur in the next two years, the game’s over for Sirius. The company notes that while it expects to get the full 15 years of operation from its satellites (and some period thereafter with deteriorating performance), the useful life may vary from the estimate. It should be noted that Sirius would be in trouble if the useful lives of this equipment turned out to be shorter than 15 years, and also it should be noted that Sirius will require massive amounts of capital, likely in the half billion dollar range, to replace all three satellites and the spare over the next 13 years.
The 2002 10-K also notes that Loral Space Systems has identified circuit failures in its satellites launched since 1997, including the Sirius satellites, but that “the circuit failures our satellites have experienced to date are not expected to limit the power of the broadcast signal or reduce the useful life of our satellites” while noting also that “if a substantial number of additional circuit failures were to occur, the estimated useful life of the satellites could be reduced. “
Interestingly, XMSR last year reduced the expected life of their two in-orbit satellites, which are Boeing 702 models, to 6.75 years from launch (running through Q1 2008), due to a progressive degradation problem with the solar array output. I don’t have any idea what that means, but I do know that XM has filed a claim with its insurance company. XMSR maintains coverage of $200 million per satellite for five years from the date of launch. XMSR is also planning on launching its spare satellite by year-end 2004 and a new satellite by year-end 2006. Bottom line: this is risky stuff – nobody knows how long these satellites will really work. While XMSR already has financed and is preparing to launch its spare satellite, it appears that SIRI is still assuming that it’s got another 13 years. This is all speculation, but my best guess is that SIRI will also end up lowering the expected life of its satellites within the next year or two, and will also have to come up with the $200 million or so per satellite to purchase and launch new ones by 2010. In any case, don’t expect satellite and transmission expenses to go down any in the future. Also, it appears that Sirius may be gambling by cutting back on insurance coverage in 2003 that, at least when viewed in the light of XMSR’s experience and coverage, may have already been woefully inadequate.
Customer Service and Billing
Customer service and billing costs include costs associated with the full time operation of a customer service center and subscriber management system. In April 2003, Sirius terminated its agreement with Sentraliant, the company that developed and operated our previous subscriber management system, which had declared bankruptcy, and paid Sentraliant $5M in termination fees, of which approximately $1M related to fees for operating the system through the date of termination.
The company expects that customer service and billing costs will increase as it acquires additional subscribers, but that customer service and billing costs on a per subscriber basis will be “significantly reduced” as our fixed operating costs are spread over a larger subscriber base. Customer service and billing costs were $6.6 million in 2001, $7.8 million in 2002, and $9.2 million in 2003 (which excludes a $14.5 million loss on the disposal of the old system.) The Q3 10-Q noted that “in May 2003, we began using a replacement subscriber management system operated by Integra Touch LLC, noting however that “portions of our new system have not been implemented including certain billing, credit card, and collections functions.” These would seem to be fairly critical functions, right? One can make a reasonable assumption that higher expenses will be forthcoming here.
Sales and Marketing Expenses & Subscriber Acquisition Costs
Sales and marketing expenses increased to $108.4 million for the year ended December 31, 2002 from $21.6 million for the year ended December 31, 2001. In Q4, the company adopted two line items for sales and marketing that breaks out “subscriber acquisition” costs. These figures were $35.4 million for sales and marketing AND $27.8 million for subscriber acquisition for Q4, or $63.2 million for the quarter. SIRI spent $196.1 million for the full year 2003 (vs. the $108.4 million combined figure for 2002.) Given the company’s stated expectation of reaching 1 million customers by year-end 2004, or almost 4 times the year-end 2003 figure, I’d bet that SIRI will continue to increase marketing spend.
General and Administration
As you might expect, Sirius doesn’t skimp on real estate or executive compensation. In March 1998, SIRI signed a lease for the 36th and 37th floors and portions of the roof at 1221 Avenue of the Americas, New York, New York, to house its headquarters and national broadcast studio and to use portions of the roof for satellite transmission equipment and an emergency generator. The term of the lease is 15 years and 10 months, with an option to renew for an additional five years at fair market value. The annual base rent is approximately $4.6 million, with specified increases and escalations based on operating expenses. The top six executives were paid $3.6 million in cash compensation in 2002, which doesn’t included restricted stock awards and options. Overall, G&A expenses were $30.7 million in 2002 and $28.5 million in 2001. 2003 full year was $36.2 million. Expect this number to go up.
Research & Development
R&D costs were $30 million in 2002, down from $47.8 million in 2001. 2003 figure was $24.5 million, so it looks like we’ve finally found one expense that might go down sometime soon.
You get my point here -- these costs aren't fixed, but are likely to increase over time.
Balance Sheets: I estimate SIRI to have net cash of $117 million after the effects of its NFL deal. (The convertible adds $250 in cash and $250 in debt, so I make no adjustment for this.) XMSR has approximately $187 million in net debt after including the proceeds of its January secondary offering.
VALUATION ASSESSMENT AND SHORT THESIS
The valuation numbers below show that SIRI is trading at a massive premium to XMSR. Based on Q4 run-rates, SIRI trades at an EV/ Revenue of 140 vs. 27 for XM. On an EV / subscriber basis, SIRI is valued at over $10,700 per subscriber, while XM is valued at $2,750 per subscriber. Here’s the table:
SIRI XMSR
Stock Price (2/13/04) $2.95 $21.80
Shrs Out 99 million 157 million
Market Cap $2.95 B $3.42 B
Est. EV $2.80 B $3.64B
Year-end Subscribers 261,061 1,360,228
Q4 New Subscribers 111,449 430,580
2003 New Subscribers 231,114 1,013,228
Run rate numbers:
Revenues ($ thous) $19,800 $133,980
EV / Q4 Run rate sales 141 27
EV / Subscriber $10,730 $2,680
In looking at these numbers, it strikes me that there are several obvious conclusions:
• XMSR is clearly on track to generate enough subscribers (somewhere between 2.5 and 3 million) to achieve operating cash flow breakeven by year-end 2004.
• SIRI has publicly guided for 1 million subscribers by year-end 2004 and operating cash flow breakeven by 2005. This guidance requires that SIRI acquire subscribers at a vastly higher rate than they have demonstrated so far.
• SIRI will burn through well over $300 million in 2004, XMSR will burn through at least $200 million, if you believe that each will meet their own projections for new subscribers (and this number assumes that subscriber acquisition is smooth over the year and not back-end loaded.)
• The market currently values a SIRI subscriber at 4X the value of an XM subscriber, despite XM’s clear leadership in the business.
• The market currently values SIRI at a a forward EV/Sales ratio of 40 times (based upon SIRI’s 2004 guidance of $70 million in revenues. Using a $200 million revenue estimate for XM, XMSR trades at an EV/Sales of 18.
Based upon these conclusions, we believe that it makes sense to short Sirius, OR short Sirius and go long XMSR as a pair trade. In addition, because SIRI has massive implied volatility, you can get to play your hand at writing super-cat insurance if you can take the risk and sell out-of-the-money call options on Sirius. Going out to Jan ’05 gets you $0.60, or more than 20% of the current stock price. Of course, these options expose you to significant risks if the stock ran up over $5, but clearly it is actually more conservative to take 20% of the stock price in cash today and commit to begin shorting the stock at $5 sometime between now and January 05 (should it run up that far) than to short the common stock at $2.97 today and receive no premium (size considerations being equal).
Risks: The biggest risk of the pair trade idea that we can think of would be if XM lost an orbiting satellite or failed to successfully launch a new satellite in time to replace its currently orbiting satellites before any performance degradation becomes noticeable to XM subscribers. If an XM satellite were lost, XM’s service would be interrupted, which would obviously be a huge boon to Sirius. It might not be possible for XM to recover from such a catastrophic event. If such an event happened to SIRI, the game would certainly be up for them.
Catalyst
SIRI is not likely to meet their own very aggressive projections for subscriber count of 1 million by year-end and 2 million by end of 2005.
A convergence of the valuation of SIRI relative to the far better XMSR.
Continued massive dilution in SIRI shares.
Any closing of the financing window to risky and speculative stocks would cause a cash crunch for SIRI.