Verizon’s agreement with T-Mobile earlier this week for a swap of AWS holdings to give both sides more contiguous spectrum within the AWS-1 band, raises a host of interesting possibilities for future transactions in the wireless space. UBS is speculating today that AT&T will dispose of its remaining AWS holdings, but it seems more plausible that AT&T could actually swap its remaining AWS spectrum (most of which is in the A-block or C-block and would naturally pair up with Verizon’s holdings in Los Angeles, Chicago, Dallas, Miami, Minneapolis and numerous other markets) for Verizon’s 700MHz B block holdings.
Although AT&T has more AWS spectrum (~1B MHzPOPs) than Verizon has in the 700MHz B block (just under 600M MHzPOPs), as I’ve noted before Verizon overpaid in the 700MHz auction to prevent AT&T from buying all of the B block, and so Verizon spent $2B on its B block spectrum whereas AT&T’s remaining AWS holdings only cost between $500M and $600M (and using the SpectrumCo transaction as a benchmark, would only be valued at around $700M today). By exiting its 700MHz B block holdings via a swap, Verizon would avoid having to take a loss on this expensive spectrum purchase, but it would be interesting to see whether money actually changed hands in such a transaction with AT&T (given that both companies have an incentive to talk up the value of spectrum, not least to make spectrum acquisitions more expensive for their rivals).
Of course, if such a swap proves infeasible (perhaps because of the significant discrepancy in price per MHzPOP) then it still seems unlikely that AT&T would announce a sale of AWS spectrum without some quid pro quo acquisition, either of Verizon’s B block, or more likely in conjunction with the purchase of DISH’s holdings (after terrestrial use of the AWS-4 block is approved by the FCC). In either case, I assume that a deal wouldn’t come until after the November 2012 presidential election.
Looking further afield, it now also seems less certain that the FCC will free up the PCS H block (which would require DISH’s uplinks to move up by 5MHz to 2005-2025MHz) when it gives DISH the go-ahead for terrestrial use of the DBSD/TerreStar AWS-4 spectrum band. As a result, Sprint might also find itself looking at swaps with smaller players such as MetroPCS and Leap to gain wider channels for its PCS G block LTE network (by using existing F block spectrum). In those circumstances, Leap and MetroPCS could potentially be paid money for a spectrum swap without necessarily giving up any capacity, assuming that Sprint gave them alternative PCS holdings in the lower part of the band instead, and this could prove interesting given the financial pressures that smaller carriers are under.
That pretty much sums up the situation with inflight cellphone calls in the rest of the world, after OnAir’s CEO indicated that only 10% of OnAir’s inflight GSM traffic is now coming from voice calls. Just how disastrous a statistic that is can be discerned from the fact that Inmarsat’s total wholesale aero passenger connectivity revenue is only around $2M per year, from 160 planes equipped with cellular connectivity, or around $12K per plane per year (in fact the number might even be lower if Inmarsat is including the far larger number of long haul planes equipped with traditional seatback phones in its $2M total).
Grossing up to retail revenues, that means passengers are spending perhaps $30K to $40K per plane per year on cellular services, and so if only 10% of OnAir’s “traffic” (which I assume is measured in revenue terms) is derived from voice calls, then that is about $10 per plane per day in voice usage. Put another way, I estimate that on average there may be as few as 2 voice calls per plane per day!
For (a truly scary) comparison, take a look at OnAir’s estimates back in 2007, that Ryanair passengers would spend EUR300K per plane per year on voice calling, or 100 times more than the current level of usage.
Of course that shouldn’t really be a surprise, because the original providers of inflight phones in the US (Verizon Airfone and Claircom) both went out of business and Inmarsat’s revenues from seatback phones were about 2% of their original projection in the early 1990s. More recently, Ryanair also stopped providing inflight cellular services. Obviously the lack of privacy and the level of background noise make it pretty hard to conduct business on a cellphone call inflight, while the cost has generally been prohibitive for leisure users, and neither of those factors has changed in any meaningful way with the introduction of cellphone instead of seatback connectivity. As a result, its no wonder that business travelers much prefer email and SMS to voice calls, and some airlines have even decided to ban voice calling themselves, despite it being legal outside the US.
It therefore seems pretty ironic that we’ve had letterwriting campaigns to the FAA and FCC, a ban proposed in Congress and even a lobbying group set up by OnAir and AeroMobile, trying to argue that voice calls on planes are a bad or a good thing (depending on your point of view), when the reality is that almost no-one actually wants to do it anyway. However, just as with the title of this article, once people become convinced that there is no middle ground to the debate, logic tends to fly out of the window. In those circumstances, its no wonder that members of Congress are eager to get involved.
UPDATE (9/5): The FAA’s own consultation document on legalizing cellphone use onboard aircraft has now been published and gives some interesting specifics on the usage levels seen in other countries. In particular, the Brazilian regulator indicated that TAM was only seeing about 0.3 voice calls per flight leg, and the consensus of most respondent countries was “that there was relatively low use of cell phone voice communication on airplanes”.
Earlier today AT&T announced details of its new proposal with SiriusXM to resolve disagreements over how to deploy mobile broadband in the WCS band. AT&T is the largest holder of WCS spectrum, with about 4B MHzPOPs of spectrum, and NextWave is the second biggest holder. However, there is a difference in usability within the WCS spectrum between the C&D blocks (unpaired 5MHz blocks) that are immediately adjacent to the 2320-2345MHz satellite radio (DARS) band (and so have a much higher risk of causing interference with satellite radio receivers) and the A&B blocks (each a paired 2x5MHz channel) which are further away from the DARS band.
AT&T and Sirius’s proposal would sharpen this difference by prohibiting mobile use of the C&D blocks, while further liberalizing use of the A&B blocks. As a result, AT&T would then have between 10MHz and 20MHz of usable A&B block spectrum over roughly half the country. The result of this proposal would likely make NextWave’s life more difficult (because nearly half of its WCS holdings are in the C&D blocks), thereby potentially giving AT&T a chance to pick up additional A&B block WCS spectrum.
However, what is particularly intriguing about the details of AT&T’s submission is the proposal that it should be allowed more flexibility to deploy FDD technologies such as LTE, with downlinks in both halves of the WCS A and B blocks. In other words, AT&T would gain yet more downlink spectrum, in addition to the Qualcomm 700MHz spectrum that it acquired last year “to allow support of asymmetrical data bandwidth allocation”. Of course, the obvious unanswered question is where would the uplink spectrum to be paired (under the proposed FDD configuration) with both the Qualcomm and WCS A&B block spectrum come from?
What AT&T needs is a clean block of paired spectrum for an LTE Advanced deployment so that the Qualcomm and WCS spectrum can be used as carrier-aggregated downlinks. When it proposed the Qualcomm transaction, the scenario that AT&T advanced was to use the AWS band to serve this purpose, but of course that is now off the table with the collapse of the T-Mobile takeover. As I’ve noted before, the Qualcomm transaction alone therefore provides a pretty compelling reason for AT&T to be interested in buying DISH.
However, another fascinating possibility is that perhaps Moelis’s assertion last week that LightSquared’s spectrum would still be worth a considerable amount of money if used on an unpaired basis (i.e. as uplink only spectrum) might be grounded in something more than wishful thinking. Indeed Moelis cites the potential for at least some of LightSquared’s spectrum (the 1670-75MHz block leased from Crown Castle) to be paired with other spectrum blocks through carrier aggregation “similar to AT&T’s planned usage of Qualcomm’s 700MHz spectrum” and I’m told that this possibility has been explored with AT&T in recent months.
Of course, GPS interference concerns in LightSquared’s satellite band would still need to be resolved, and LightSquared would still need to pay for leasing spectrum from Inmarsat (according to Moelis’s figures the lease payment if the full L-band band is usable has now been increased to $145M p.a. from April 2014 under the revised agreement struck with Inmarsat in April). Even then, uplink spectrum is generally worth much less than downlink spectrum, both because there is a need for additional downlink spectrum due to traffic asymmetries and, as LightSquared found out to its cost, interference concerns can be more problematic in downlink spectrum.
As a result this severely undercuts Moelis’s argument that LightSquared should be able to attribute the same valuation to its spectrum whether it is used for uplinks or downlinks (not to mention the use of comparisons based on recent sales within the well established and widely deployed AWS-1 band). However, this possibility does at least raise the question of whether AT&T’s acquisition plans (which are intended to give it enough spectrum for the next five years) include options other than buying both DISH and Verizon’s 700MHz B block spectrum.
Without any hint of the PR blitz that I had expected, Intelsat has quietly updated its website to confirm my blog post in March, that it is about to order at least two new satellites, IS-29 and IS-33 to provide high capacity spot beam Ku-band service in the North Atlantic and Indian Ocean regions. These satellites will be in-service in 2015 and 2016 respectively, and are intended to “provide four to five times more capacity per satellite than our traditional fleet” with total throughput “in the range of 25-60 Gbps” (this appears to be a total not a per satellite figure – I would guess the throughput per satellite is around 12Gbps, roughly the same per satellite as Inmarsat’s GX, including its high capacity overlay beams).
UPDATE (June 7): Intelsat has now put out a press release and added more data to its website including a fact sheet, which states specifically that the throughput of 25-60 Gbps is per satellite. Obviously this figure is a wide range but it is clearly much greater than the Global Xpress per satellite capacity. I understand that one reason for Intelsat’s lack of publicity is the quiet period associated with its proposed IPO, but Intelsat definitely considers this a very important development and has been trumpeting it privately to distributors at its recent partner conference.
Intelsat is planning to integrate these new satellites into its existing maritime coverage as shown below (indeed there is less high capacity oceanic coverage than I expected, presumably because it will take time before Intelsat’s existing capacity fills up) and it appears that Intelsat will now be looking to compete head-to-head with Inmarsat’s Global Xpress as well as Viasat (both of which Intelsat appears to be referring to with its comment that “Unlike many new satellite operators, Intelsat is not constrained to Ka-band“)
What we haven’t yet seen are the details of Intelsat’s launch partners. It is clear that one partner is Panasonic, but the more important question is who Intelsat might have managed to secure in the maritime market. Inmarsat’s recent list of XpressLink distribution partners was notable for the absence of Vizada and most other major maritime VSAT providers, so if one or more of these maritime players now makes a substantial commitment to Intelsat, it will be another important sign that the transition to Ka-band in the maritime and aeronautical sectors is far from a foregone conclusion.