In late April 2026, the Federal Communications Commission unanimously adopted a new satellite spectrum-sharing framework, replacing a set of technical limits that had been written into international radio regulations when the internet was still carried on dial-up lines. The announcement drew little notice outside specialist circles. That is a mistake. The old rules — built around a metric called Equivalent Power Flux Density, or EPFD — were quietly acting as a ceiling on how much bandwidth every LEO constellation on orbit could actually deliver to customers. Removing that ceiling is one of the most consequential regulatory acts in commercial space since the FCC first authorized non-geostationary orbit systems in bulk.

What EPFD Actually Did

The EPFD limits originated in the ITU’s Radio Regulations as a protective mechanism for geostationary (GSO) satellites. The logic was straightforward for its era: a GEO satellite sits at a fixed point in the sky, serving television broadcasters, government agencies, and telecom carriers under long-term contracts. If a new constellation of LEO satellites was allowed to spray radio energy indiscriminately across the Ku and Ka bands, the aggregate signal landing on a GEO receiver could make its link unusable.

So regulators set a hard cap: no matter what, a non-geostationary (NGSO) system’s cumulative power flux density at any GSO receiver had to stay below a defined threshold. The threshold was expressed as a statistical limit — the NGSO system could not exceed it for more than a certain percentage of time. The framework was finalized at World Radiocommunication Conferences in the late 1990s and carried largely unchanged into the 2000s and 2010s.

The problem is that EPFD limits are blunt. They do not measure whether a GSO receiver is actually being harmed — they cap power levels across wide angular regions regardless of whether any GSO satellite is present in a given direction, or whether that GEO operator is even currently transmitting. The practical result, as constellations scaled up, was that LEO operators had to throttle their beams, widen avoidance angles around the geostationary arc, or cut back transmit power across large portions of their coverage footprints. A Starlink terminal in rural Montana might be looking at clear sky with no GEO satellite within thirty degrees — and still be subject to a constraint designed to protect a GEO link that does not exist in that geometry.

The FCC’s New Framework

The April 2026 rules, published formally in May and effective July 13, replace EPFD with what the FCC calls performance-based coordination. The operative question shifts from “how much power are you radiating?” to “are you actually degrading a real GSO service link?”

The new backstop limits are specific. For the affected bands — 10.7–12.7 GHz, 17.3–18.6 GHz, and 19.7–20.2 GHz — NGSO systems must not cause long-term throughput degradation exceeding 3% on adaptive-coded-modulation GSO links, and must not cause short-term link unavailability increases exceeding 0.1%. For non-adaptive links such as video distribution feeds, an interference-to-noise ratio threshold of -10.5 dB applies, with compliance required 80% of the time. A minimum avoidance angle of 3 degrees around the GSO arc remains in place as a rough guard.

The benchmark is 328 real GSO reference links drawn from ITU and FCC databases — actual operating satellites with actual service areas, not hypothetical worst-case geometries. NGSO operators must coordinate against those references in good faith before the backstops kick in.

Crucially, the FCC imposed no grandfathering period. The new rules apply immediately to existing licensees, pending applicants, and future applicants alike.

Why This Matters for Capacity

The magnitude of the efficiency gain is hard to state precisely — it depends heavily on geography, constellation architecture, and which GEO systems are co-located in a given coverage area. But the directional case is clear. Amazon, in comments filed during the FCC rulemaking process, estimated that replacing EPFD-based constraints with harm-based criteria could increase the usable capacity available to a given area by something on the order of 700 percent. Even heavily discounted, that figure points to the scale of spectrum efficiency that was being left on the table.

The FCC’s own economic analysis projected $1.6 to $19.9 billion in economic gains from the reform over the 2026–2030 period, against compliance costs measured in the low hundreds of thousands of dollars — a ratio that reflects how regulatory drag, rather than technology or physics, was the actual constraint.

For rural and remote users — the populations that LEO constellations most directly serve, and who have the fewest alternatives — capacity constraints translate directly into slower speeds, higher prices, and the need to build more gateway infrastructure to compensate for artificially limited spectral efficiency.

What the ITU Hasn’t Done Yet

The FCC’s reform covers U.S.-licensed systems and applies within U.S. jurisdiction. The ITU framework — which governs how national regulators coordinate spectrum globally — has not moved at the same pace.

WRC-23, the World Radiocommunication Conference held in late 2023, declined to revise the international EPFD limits. Instead, the conference tasked ITU-R Study Group 4 with analyzing aggregate interference from multiple co-frequency NGSO systems, with no formal revision expected before WRC-31. A new FSS downlink allocation in the 17.3–17.7 GHz band was approved for NGSO systems in ITU Region 2 (the Americas), but with strict EPFD conditions attached — conditions that largely reflect the old framework.

This creates a split landscape. U.S.-licensed constellations gain operational flexibility under the FCC’s new rules. Their counterparts authorized by other administrations — or providing service in regions where operators must coordinate under traditional ITU EPFD obligations — remain constrained. Eutelsat OneWeb, licensed partly through the UK regulatory framework. Amazon’s international market access. The interplay between domestic FCC relief and unmodified ITU obligations will be a live question as operators expand coverage.

The Harder Problem: NGSO-to-NGSO Coordination

The FCC’s reform addresses how LEO constellations share spectrum with GEO incumbents. A parallel and in some ways thornier question is how multiple NGSO systems share spectrum with each other.

A Starlink satellite and a Kuiper satellite, both flying low orbits in overlapping coverage areas, both transmitting in the Ka downlink band: the interference geometry is dynamic, changing by the second as both constellations move. The ITU’s legacy rules were designed for a world with one or two NGSO systems. That world is gone.

The FCC replaced EPFD-based NGSO-to-NGSO rules in 2023 with thresholds tied to actual service degradation — link unavailability increase capped at 0.4% short-term, and aggregate interference causing no more than 3% throughput degradation long-term. Earlier-licensed systems retain priority protection for up to ten years, after which that protection sunsets. This time-bounded seniority system matters: it means today’s first movers have a regulatory advantage that is real but not permanent.

Meanwhile, researchers are identifying a deeper issue. The technical paper trail from ITU-R Study Group 4 and academic work published in 2026 points to aggregate interference — the cumulative effect of many NGSO systems all coexisting in the same bands — as an open modeling problem. No single system causes harm by itself; the concern is what happens when five or ten competing constellations all operate in the same Ku-band downlink window simultaneously. Static worst-case models break down. The emerging consensus is that multi-operator coordination requires real-time, data-driven tools: dynamic beam management, shared sensing infrastructure, and interference databases updated in near-real time.

What Comes Next

The FCC’s EPFD overhaul resolves an immediate and important constraint. It does not resolve the structural tension at the heart of spectrum management for LEO: a regulatory regime designed for a handful of GEO incumbents is being stretched to accommodate hundreds of competing spacecraft from a dozen operators, all moving, all frequency-hopping, all under pressure to maximize throughput.

The tools exist to do this better. Adaptive beamforming, AI-driven interference detection, and operational data-sharing arrangements like the one SpaceX and the National Radio Astronomy Observatory have demonstrated for protecting radio telescopes all point toward what a modern coordination infrastructure could look like. What does not yet exist is the multilateral framework that would mandate or incentivize their adoption across operators and across jurisdictions.

The FCC acted within its authority and within a domestic timeline. The ITU’s timeline runs to 2031. In the interval, spectrum management for LEO broadband will be increasingly governed by bilateral coordination agreements, operator-specific technical backstops, and whatever voluntary frameworks operators are willing to adopt when their commercial interests align. That is a workable arrangement for the next few years. Whether it scales to a world of ten mature constellations all competing in the same bands is an open question no one has answered yet.

Sources