Where the rules get written

Over the last eighteen months the American regulator has quietly settled several questions the commentary is still treating as open, and it has settled them on the public record. What has already been decided is separable from what has only been proposed.

The decision that has already been taken

On 12 May 2026 the FCC’s Wireless Telecommunications Bureau and Space Bureau approved both EchoStar transactions, moving roughly 115 MHz of mid-band spectrum to SpaceX and AT&T in deals the Commission described as “collectively surpassing $40 billion in deal value” (news release; order GN Docket No. 25-302, DA 26-471). The two purchase prices in the parties’ own filings are $19.617B and $20.25B, which is $39.87B between them, so the Commission’s figure reconciles only once the $2.4B escrow is included. The reconciliation is not cosmetic: the headline is the number every summary repeats, and it is not the number in the filings.

Filed FCC news release and order of 12 May 2026: GN Docket No. 25-302, DA 26-471; the two purchase prices from the parties’ own filings

The headline is the money, but the consequential part of the decision is a single sentence sitting under a heading called Tech-Neutral Innovation. It allows SpaceX to use its new spectrum flexibly for terrestrial, space-based and hybrid network architectures, subject to what the Commission calls “first-of-its-kind performance obligations” requiring “meaningful, reliable connectivity to the public — whether D2D, terrestrial, or both — within years”. Neither half of that sentence is written to prefer one architecture over another.

The symmetry is the point. The permission covers three architectures without ranking them, and the obligation attached to it is drafted in the same terms, so the holder of this spectrum is not being told to supplement a terrestrial network and handed a terrestrial option as a courtesy. It is being given a position whose permitted use it determines for itself, measured against a test that does not specify how the signal reaches the handset. Almost everything still argued about satellite operators entering mobile sits on the other side of that sentence.

From a carrier’s seat, the question of whether a satellite operator is allowed to build a terrestrial mobile network on its own spectrum has been answered, and the answer is yes. The remaining questions are financial and physical, not regulatory.

The rest of the order is conditions rather than permissions. It requires EchoStar to fund the $2.4B escrow, accelerates AT&T’s 600 MHz build-out, and preserves Boost Mobile through a hybrid MVNO arrangement with AT&T: obligations on the parties to the transaction, set alongside the performance obligations rather than in place of them. The order settles who holds this particular spectrum; where the next tranche comes from had already been answered elsewhere, by Congress.

The auction machine restarted, and spectrum got more expensive

Congress restored the FCC’s auction authority in the reconciliation act signed on 4 July 2025, running it to 30 September 2034 and attaching a mandated pipeline of at least 800 MHz, including at least 100 MHz of upper C-band, at 3.98–4.2 GHz, to be auctioned by July 2027. The FCC’s own auction release credits that statute for both the restored authority and the 800 MHz pipeline. The pipeline band-by-band detail here comes from secondary summaries of the Act and should be checked against the enacted text before it is relied on.

The first auction in four years followed quickly. Auction 113 opened on 2 June 2026 and offered AWS-3 spectrum at 1695–1710, 1755–1780 and 2155–2180 MHz: 200 licences left unsold from the 2014 defaults, with the proceeds funding the “rip and replace” reimbursement programme (FCC, 2 Jun 2026). The opening release framed it as housekeeping. The published results tell a sharper story.

Gross winning bids as published by the FCC. Dollars per MHz-POP are our own calculation from block bandwidths and market populations.

Filed FCC Auction 113 published results-by-licence file, retrieved 17 Sep 2026

The results turn on two things, and only one of them was predictable. The first is the concentration: Verizon took 88.5% of the money, $3.16B in a single auction, which is a scale of commitment no other bidder came near. The second is that SpaceX bid at all, a company that had, until this point, appeared in the American terrestrial record only as a tenant.

It won exactly two licences. The first is the B1 block at 1700–1710 MHz, unpaired uplink, in Cincinnati–Hamilton, for $8,406,000; the second is the AWS-3 G block at 1755–1760 / 2155–2160 MHz (not the PCS G block discussed in the earlier parts), in the Gulf market, for $84,200. B1 sits inside the 1695–1710 MHz range SpaceX is separately acquiring from EchoStar. We cannot see intent from a bid sheet, but the band alignment is a matter of record, and so is what followed: the FCC market licence file of 13 September 2026 shows Space Exploration Holdings holding two market licences granted 26 August 2026, its first that are not leases (see part 4).

One caveat governs every use of that clearing price. Auction 113 sold a leftover inventory concentrated in dense markets, so its $2.658/MHz-POP is not comparable to a nationwide clearing price such as the C-band’s $0.852 or the $0.887 SpaceX agreed with EchoStar. Comparing them properly needs a population-weighted like-for-like by market, which we have not done. What the figure does establish on its own terms is that in 2026 a national carrier paid roughly $3 per MHz-POP for mid-band capacity in the markets it wanted.

What the midterms do, and what they do not

The mechanics are widely misdescribed. The Congressional Research Service’s standing account of the agency (CRS R45699) sets them out plainly: the FCC has five Commissioners on staggered five-year terms, with no more than three from one party; the President designates the Chair from among sitting Commissioners on day one of an administration, with no separate Senate vote; a Commissioner whose term has expired may serve as a holdover; and three are needed for a quorum. None of that turns on an election.

What does move is the line-up itself. A nomination to one of the Commission’s vacant seats was announced in August 2026 and had not been confirmed as of the Axios report of 10 August 2026 (Axios, 10 Aug 2026), which is the whole of what we can evidence. Verify the current line-up at fcc.gov on the day you rely on it. An August source cannot evidence a September seat count. Which seats are filled changes none of the decisions already taken, but anything built on top of them may turn on it.

Election day is 3 November 2026. All 435 House seats and 35 Senate seats are contested: 33 regular Class 2 seats plus two specials, per published election trackers rather than a primary source, and none of it changes the FCC’s composition, because the terms and the appointment mechanism run on their own statutory calendar. What the result can change is the chairmanship of Senate Commerce and House Energy & Commerce, the Senate’s appetite for confirming nominees to those two vacant seats, appropriations, whether further spectrum legislation advances beyond the 2025 Act, and the intensity of oversight on NTIA and BEAD. Those are real levers, but they act on the pace of policy rather than on the permissions already granted.

The strategic read is therefore about what is left to decide rather than about who decides it: we are not aware of any scheduled proceeding that would reopen the core direct-to-device permissions before 2028: the SCS framework is adopted, the EchoStar transfers are approved, and the auction pipeline is statutory.

The FCC’s composition turns over with the presidency, and the next presidential election is November 2028, which is also the period in which SpaceX’s performance obligations fall due and AST’s leases expire. That is a coincidence of dates, not a forecast.

The wildcard nobody is modelling

Separately, and with far less attention, the Commission circulated a Notice of Proposed Rulemaking for its 6 August 2026 open meeting: ET Docket No. 26-169, Unleashing Unlicensed Spectrum for Direct-to-Device, set out in the fact sheet the Commission published on 16 Jul 2026. It proposes adding Earth-to-space allocations in 902–928, 2400–2483.5 and 5725–5850 MHz, more than 225 MHz of unlicensed spectrum, at Part 15 power levels up to 36 dBm EIRP, and it asks whether power-flux-density, out-of-band and aggregate limits should apply across multiple satellite operators sharing a band.

If that becomes rule, a class of direct-to-device service arrives that needs no spectrum position at all: no auction, no lease, no landlord. It will not carry voice, and nothing in the proposal suggests it could. It may well carry the sensor and asset traffic that several of the business cases in this market quietly depend on, which is why a proceeding about unlicensed bands belongs in a strategic model that otherwise tracks only licensed ones.

The same NPRM puts numbers on why the Commission is interested. More than $40 billion has moved through the US space economy for D2D in eighteen months, on the Commission’s own count. It cites SpaceX/EchoStar at $19.6B, AST’s $550M L-band agreement, and Amazon’s acquisition of Globalstar and Rocket Lab’s of Iridium as “collectively totaling nearly $20 billion”, the same transactions the earlier parts read from the commercial side. The NPRM is a circulated draft rather than adopted rules, and the technical limits in it are proposals rather than Commission action; treat it as a signal of direction rather than settled law.

PESTLE with the evidence attached

Each row in the table below carries the document, docket or dataset it rests on, so that a reader can disagree with the reading and still check the fact.

One row in that table rests on a weaker source than the others. T-Mobile’s 0.0002–0.0003% network-usage figure reaches us through trade reporting, Sept 2026; we have not obtained a transcript. The rest of the column points at FCC documents, docket numbers and figures set out in the earlier parts, all of which can be pulled directly.