Close scrutiny of the spectrum related legislation in last week’s jobs bill appears to give a pretty comprehensive outline of the deal that is to be expected between the FCC and DISH over its transfer of control and waiver request for the TerreStar and DBSD spectrum. In particular, the FCC is tasked with auctioning the H block spectrum (1915-20/1995-2000MHz) and extended AWS-3 block (2155-2180MHz) plus 15MHz in the 1675-1710MHz band (presumably the 1695-1710MHz block previously recommended by the NTIA, though note update below), as well as identifying an additional 15MHz of contiguous spectrum to be auctioned.
Assuming that this additional 15MHz of spectrum is intended to be contiguous with the other spectrum listed above (rather than being a single 15MHz contiguous block), then it appears that DISH will give up 2x5MHz of its 2x20MHz of DBSD/TerreStar spectrum (2000-05/2180-85MHz), and move its uplink frequencies up by a further 5MHz into the original J block uplink (2020-25MHz) in exchange for the waiver, and so that there is no windfall to be criticized as in the case of LightSquared.
This would leave DISH with 2x15MHz of spectrum at 2010-2025MHz (uplink) and 2185-2200MHz (downlink) and would implicitly create a new guardband in the 2000-2005MHz block (though the FCC would apparently still be supposed to auction this spectrum) and a new paired block at 2005-10/2180-85MHz to be auctioned (though this block would likely be repurchased by DISH or its partners in the ensuing auction because it would have relatively little value to other players). As Walt Piecyk at BTIG Research previously noted, Sprint should also then be well positioned to add the H-block spectrum to its G-block LTE network (assuming that the 1915-1920MHz uplink would not interfere with the PCS downlink spectrum at 1930MHz).
Alternatively if the FCC really does intend to auction a single contiguous 15MHz block then I would expect DISH to give up 10MHz of its uplink spectrum, and move up into the J block uplink, creating an asymmetric 10x20MHz allocation with 2015-2025MHz uplink and 2180-2200MHz downlink. That might allow an extra 10MHz downlink of downlink to be inserted above the H block (at 2000-2010MHz), with an implicit guardband left at 2010-15MHz. Given the much higher value of downlink spectrum in a carrier aggregation model, this trade would potentially allow the FCC to raise much more money from a subsequent auction (though it would also prevent DISH from buying back the 2x5MHz it had given up).
Based on an ex parte filing by DISH last week, the company is also prepared to accept conditions to “facilitate competitive carriers’ access to DISH’s planned nationwide wireless network”, and presumably with these concessions the FCC would permit a relatively relaxed buildout mandate as DISH has requested. An extended timeline would also help DISH because the re-auction of the spectrum they would give up would probably come ahead of the activation of their new network, and potentially therefore allow 2x20MHz operations from launch (assuming DISH repurchased the spectrum).
However, AT&T is fighting a rearguard action to impose the same buildout conditions as LightSquared and therefore remove DISH’s flexibility to ally with alternative partners, which it seems could include not just MetroPCS or DirecTV but perhaps even T-Mobile and/or America Movil. Of course the more credible DISH’s alternative plan is, then the higher the premium that AT&T would have to pay to take over DISH, but based on DISH’s filing there appears to be no doubt that the FCC is determined to ensure wholesale access to DISH’s LTE Advanced network remains available to other operators, whether or not DISH is taken over by AT&T. After approval of the TerreStar transfer of control by the Canadian regulator two weeks ago, it now appears that the FCC is expected to announce its ruling in the first half of March, so we should know the answer to all of these questions very soon.
UPDATE (2/22): With respect to the 15MHz of spectrum to be identified in the 1675-1710MHz band, this was originally expected to be an uplink band (to be paired asymmetrically with the AWS-3 downlinks), and would therefore most appropriately be located next to the AWS-1 uplinks at 1710-55MHz. However, I understand that there is pressure to use this 15MHz of spectrum for either carrier aggregation downlinks or TDD operation, which would not be feasible right next to the AWS-1 uplinks. As a result, it is plausible that the 15MHz of reallocated spectrum could be moved to the bottom of the band (1675-90MHz), adjacent to Crown Castle’s 1670-75MHz block, which is leased to LightSquared Inc.
Harbinger has previously stated that it plans to sell this spectrum to raise money and the potential windfall from the spectrum bill creating a contiguous 20MHz block of spectrum might increase its value significantly. Harbinger would presumably hope to pay off the term loan of roughly $300M which is due for repayment on July 1 (this is separate from the debt secured by LightSquared’s satellite assets), and that would require the 5MHz block to achieve a value of around $0.35-$0.40/MHzPOP once the costs of buying out the Crown Castle lease are taken into account. Though that may be feasible, with only a small 5MHz sliver of spectrum being sold, it seems unlikely that this sale could raise enough to provide a meaningful amount of additional capital or any return to Harbinger on their previous investment in the company. In addition, it is much harder to see any use for the 8MHz of 1.4GHz spectrum leased from TerreStar, and so it is unclear how long this lease will continue. However, for once it seems plausible that Harbinger might have finally found one small block of spectrum that the government can enable them to actually make a profit on. Its just a shame that so much more money has been wasted on unusable satellite spectrum in the meantime.