D2D Has Already Crossed: Why the Satellite-to-Cell Convergence Is a Regulatory Problem, Not a Technology One

Key Insight

Question. Has Direct-to-Device satellite connectivity already crossed from a Horizon 2 emerging opportunity into a Horizon 1 deployed capability — and if so, is the prevailing “complementary gap-filler” framing still the right lens for the next five years?

Thesis. D2D has already completed its basic-service convergence: as of the February 2025 T-Mobile/Starlink commercial beta, unmodified consumer handsets connect to LEO satellites on terrestrial LTE spectrum, and the category is in early rapid growth rather than late emergence. The remaining uncertainties are not about whether the technology works but about three non-technical constraints — the regulatory ceiling on spectrum scope, the commercial viability of subscriber monetization, and the asymmetry between satellite operators and their mobile network partners. On current deployment cadence, D2D will become the default failover layer of the global communications grid inside the 2028-2031 horizon, with a single vertically integrated operator setting the pace and a policy apparatus designed for an earlier era deciding whether that operator ends up dominant, contested, or fractionally bifurcated against a parallel Chinese stack.

State of the Art

As of early 2026, D2D satellite-to-cell service is commercially live in the United States and in at least eight partner-MNO markets. SpaceX’s Starlink direct-to-cell constellation reached roughly 650 satellites by September 2025, up from zero in early 2024 and 400+ by the start of 2025, with another 1,000+ projected by end-2026 and an FCC request pending for a further 15,000. T-Mobile has published the first commercial pricing benchmark at USD 15-20 per month, operating on 5 MHz of Band 25 (1900 MHz) under the FCC’s Supplemental Coverage from Space (SCS) framework adopted in 2024. AST SpaceMobile, SpaceX’s principal rival architecturally, has five BlueBird satellites in orbit against an estimated 45-60 needed for contiguous US coverage, backed by AT&T, Verizon, and Vodafone. Lynk Global operates a third Western ecosystem in partnership with SES and emerging-market MNOs such as MTN, while China’s Huawei+Tiantong-1 pairing delivers a GEO-based equivalent on the Mate 60 Pro with LEO ambitions visible only as ITU filings. The installed base of satellite-capable smartphones is over 200 million units per year as of 2024, and the chipset has migrated from flagship differentiator to default feature.

The conventional placement across the three horizons, before the 2025 inflection, was clean: constellation deployment in H1, throughput scaling and ecosystem assembly in H2, the failover-layer and dual-use transformation in H3. That placement no longer holds.

Complication

What has changed is that the basic-service convergence — text, voice, and limited data on unmodified handsets using terrestrial spectrum from orbit — moved from prototype to commercial service in twelve months. That is faster than any three-horizons narrative is comfortable accommodating, and it has relocated the binding constraints of the category. The interesting questions are no longer “when will D2D work” or “who will build the constellation first” but “how high is the regulatory ceiling on spectrum scope”, “who captures the asymmetric platform value created by the convergence”, and “whether the H1→H2 transition is one the first mover can complete unilaterally through deployment cadence or whether it requires consensus the first mover cannot manufacture”. The Deloitte view that D2D will remain complementary to terrestrial LTE “for the foreseeable future” and the SpaceX view that its V3 generation will deliver LTE-parity consumer experience are, on close reading, a disagreement about which inflection happens first inside the next horizon — not about whether convergence is real. Both may be right, sequentially.

The Argument

The deployment lead is a phase-difference on the S-curve, not a head-start

The single most consequential fact in the current landscape is that SpaceX is not ahead of its rivals in the sense that runners are ahead in a race — it is on a different phase of the S-curve entirely. A 400-versus-5 satellite gap in early 2025, widening to 650-versus-5 by September 2025, is not a head-start to be closed through faster execution. It is a structural phase-difference: SpaceX’s Starlink D2D constellation is already in late emergence bleeding into early rapid growth, with launch cadence and vertical manufacturing throughput no other operator can match, while AST SpaceMobile and Lynk are in early emergence and the Chinese LEO curve is still pre-emergence, visible only in paper filings. Because the category-level curve is dominated by one actor’s curve plus noise, the trajectory of the whole category is effectively the trajectory of a single operator’s execution plan — a circumstance that makes the trajectory both more legible and more fragile than typical multi-actor technology evolutions.

The Horizon Summary inherited from the structural analysis anchors this claim without dictating it:

The phase-difference reading tightens what the table alone cannot convey. SpaceX is not just stronger in H1 than rivals; it has exited the emergence phase that AST and Lynk are still in, which means competitive responses that assume a symmetric race — build more satellites faster — are category errors. The defensible counter-positioning is architectural (AST’s larger satellites and owned MSS spectrum), geographic (Lynk via SES in emerging markets), or temporal (wait for the next generational leap). A head-on deployment race is ruled out by the phase-difference, not by capital. The USD 17 billion EchoStar spectrum acquisition by SpaceX — a quantity that only makes sense if the acquirer believes spectrum portfolio scope is the ratchet that converts the deployment lead into an infrastructure-value monopoly — confirms that the leader is reading its own position as one where spectrum lock-in, not more satellites, is the next binding constraint. A competitor burning capex to match the constellation while the leader burns capex to widen the spectrum ratchet is fighting the last war.

Convergence is done; the fight is over asymmetric value capture

The second trajectory force is that the cellular-satellite convergence — the literal collapse of two formerly distinct stacks into one operational service — is already done at the base-service level and is asymmetric in who captures the value it creates. The February 2025 T-Mobile/Starlink commercial beta demonstrated all six required integration interfaces: orbital use of terrestrial bands, the LTE air interface reaching LEO and back without handset modification, authentication and roaming into the MNO core network, cross-tier interference management at scale, multi-jurisdictional authorization via the FCC SCS framework and bilateral extensions, and government activation for emergency use. Every prerequisite the convergence required is in place except two: a durable cross-tier interference governance regime beyond the current dispute-by-dispute FCC adjudications, and viable commercial subscriber monetization. Both are non-technical constraints. The slowest boat is the regulatory boat, not the radio.

This is where the convergence lens and the platform lens reinforce each other. The convergence produces emergent properties — universal coverage with no specialized terminal, capex-lite telecom infrastructure inversion, dual-use civilian-military comms as a default capability, disaster-resilient connectivity as a routine commercial service — and platform economics govern how the value those properties create is captured. The MNO partnership architecture that made the convergence possible in H1 is structurally transitional: capex-lite is a generous framing for the MNO side when D2D is complementary, but becomes a disintermediation risk once V3-class throughput approaches LTE parity and the MNO is no longer contributing infrastructure but licensing spectrum to a partner that has become a potential substitute. The renegotiation inside the 2028-2031 window — whether the alliance institutionalizes with revenue-sharing or shatters into vertical competition — is not a scheduling question; it is the question around which the entire H2 value-capture contest organizes.

Ecosystem economics also sharpen where the asymmetry points. Three Western D2D ecosystems exist on paper — Starlink+T-Mobile alliance, AST+AT&T/Verizon/Vodafone, SES+Lynk+(planned)Omnispace — but only one has the deployment cadence for cross-side network effects to compound. The opposition coalition of legacy MSS operators (Viasat, Globalstar, Iridium, SES, Eutelsat) filing against SpaceX’s 15,000-satellite expansion is the diagnostic signal: it is the predictable reaction to a keystone with dominator characteristics rather than to a keystone optimizing ecosystem health. The less visible but more dangerous envelopment threat is at the operating-system layer, where an Apple or Google decision to treat D2D as a transparent system feature — an extension of what Apple has already begun with Globalstar’s emergency SOS — would disintermediate the MNO partnership entirely and turn every satellite operator in the stack into a raw infrastructure provider. Either outcome collapses the platform asymmetry into simpler dynamics than today’s three-cornered picture suggests, and both point to concentrated value capture at or above the constellation tier.

Regulatory governance, not technology, is the binding ceiling on H3

The third trajectory force is that the plateau of the D2D category is set by rules, not by physics. The theoretical ceiling on D2D is a composite of four constraints, and all four are regulatory or political rather than technical: the total MHz that can be authorized in the 1.6-2.7 GHz LTE bands for SCS use, the EPFD and orbital-congestion rules the CSIS description of “from when MySpace was popular” captures accurately, the degree of CEPT and national harmonization in jurisdictions outside the United States, and the political willingness of governments to permanently designate a commercial dual-use capability as essential infrastructure. Each of these is a ceiling the market cannot raise by deploying more satellites or engineering better handsets. Each is also a ceiling that the first mover cannot in-house — SpaceX cannot vertically integrate the ITU, CEPT, or the national spectrum regulator of a partner country. This is a significant limit on the “leader burns capex to widen the spectrum ratchet” strategy that worked at the FCC.

Two dynamics make this ceiling the binding H3 constraint rather than a recurring H2 friction. First, regulatory first-mover lock-in: the ITU rewards filers who establish positions before rivals arrive, so decisions taken in H1 have outsized H3 consequences — the concentration that emerges from the current FCC adjudications has limited contestability because the framework that created it does not offer much mechanism to revisit it. Second, multi-jurisdictional fragmentation: the 27-plus European national regimes under CEPT mean that European D2D deployment lags the United States by years unless harmonization is forced via the WRC successor process, and the jurisdictions that resolve their rules last become the place where the rival ecosystem best aligned with their domestic politics gains a first-mover position. The binding pace in H3 is therefore set by whichever regulatory process is slowest globally, and the answer for the next horizon is CEPT harmonization, followed by the quieter question of whether the ITU EPFD regime is modernized or inherited intact into the decade where 15,000+ D2D satellites are sharing the LEO environment with everything else.

Against this backdrop, the dual-use dimension stops being a separate H3 scenario and becomes a regulatory forcing function. Ukraine’s Delta battlefield management system integration, the Helene and Milton and LA wildfire STAs, and the Wireless Emergency Alerts framework have already established that consumer D2D networks have wartime and disaster utility — and that governmental emergency response depends on a single commercial operator’s decisions. That dependency loop is politically combustible and pulls in the opposite direction to the commercial concentration the platform dynamics are producing: the more essential the capability becomes, the harder it is to leave its governance to first-mover ITU lock-in. Whether this tension resolves through universal-service designation, explicit government-customer commercial contracts, or a contested reopening of the spectrum and EPFD frameworks is the single most important H3 variable, and none of the technology trajectories inside the category determine which path it takes.

Implications

The Thesis is that D2D has already crossed at the base-service level and will reach default-failover-layer status inside the 2028-2031 window with an asymmetric platform structure set by regulatory ceilings. What flows from that for decision-makers is a reordering of what to attend to now and where the leverage actually sits.

Immediately, for decision-makers positioned inside the Starlink/T-Mobile alliance or its rivals, the priority is to lock in spectrum scope before the incumbent MSS opposition coalition succeeds in narrowing it — each MHz authorized in the LTE bands between 1.6 and 2.7 GHz is a one-way ratchet. Defending the FCC SCS framework as a one-regulator template and pushing it bilaterally into partner-MNO jurisdictions before CEPT crystallizes a fragmented alternative matters more in the next 24 months than another constellation increment. Emergency-use precedents from Helene, Milton, LA, and Ukraine should be converted into permanent essential-infrastructure designations that cannot be quietly unwound; the political window for that conversion is open now and may not be open later. Credible commercial subscriber economics — not just capex — need to be published inside 12-18 months, or the “monetization unproven” critique will start extracting a political price that the technical success has so far masked. And the OS layer deserves more attention than any additional MNO partnership signing: Apple and Google’s decisions about D2D as a system feature, rather than a per-MNO premium, shape the entire next-horizon value capture contest.

Over the 2028-2031 build horizon, the investments that matter are V3-class throughput execution, Starship deployment cadence sustaining, multi-jurisdictional regulatory harmonization via active CEPT and ITU and bilateral engagement, and — decisively — the design of a revenue-sharing architecture with MNO partners that survives the moment when D2D becomes a substitute rather than a complement. Alliance cohesion through the asymmetry transition is not a diplomatic question; it is the question that determines whether H2 ends with a stable platform structure or with a renegotiation that fractures the Western market. In parallel, formalizing the FEMA/DOD/allied emergency-services relationships with explicit commercial contracts rather than ad-hoc emergency authorities converts the currently combustible dependency loop into something governed, and engaging 3GPP NTN R17/18 standards work positions the incumbent for the 6G NTN successor curve that may dissolve today’s D2D platform category into a transparent feature of the integrated network.

What to watch in H3 is a shorter list than the horizon is usually given. Whether the Chinese parallel ecosystem operationalizes its ITU filings, whether Apple or Google integrates D2D at the OS level by default, whether a major event triggers universal-service designation of consumer satellite connectivity, and whether the ITU modernizes its EPFD and first-mover rules. Each of these inflections matters more than any further constellation milestone inside the Western ecosystem. The first decision a reader responsible for communications resilience should make, now, is to stop treating D2D as a future capability to be procured separately and start treating it as an operational layer whose governance they are already party to by default — and to act accordingly in spectrum consultations, emergency contracting, and partnership architecture.

Limitations

The H3 projections rest on assumptions that are plausible but not guaranteed: that SpaceX deployment cadence sustains, that the FCC SCS framework is not reversed, that V3 throughput claims are technically achievable as filed, that interference disputes are resolved without crippling power constraints, and that the dual-use political precedent is not unwound. Confidence declines sharply beyond the 7-year mark. Data gaps constrain several of the claims above: 3GPP NTN specifications, WRC-23 national implementation, AST SpaceMobile commercial and financial timelines, Chinese LEO operational status, and any published D2D subscriber revenue data are absent or thin in the source corpus, which makes the monetization-side reasoning the most inferential portion of the Thesis. The three-horizons frame on which the structural analysis rests favors incremental transition narratives and understates the speed at which the base-service convergence actually moved — twelve months from prototype to commercial beta is a pace the framework does not comfortably accommodate. S-curve positioning is recognized more easily retrospectively than prospectively, and the “approaching inflection” claims for V3 throughput rest on the incumbent’s own forecasts. Platform-ecosystem analysis was developed for consumer internet markets and does not handle cleanly the government-as-customer-regulator-complementor triple role that is, for D2D, the single most consequential dynamic in the dual-use dimension. If one finding would collapse the Thesis, it is a demonstration that commercial subscriber monetization cannot close the capex-revenue gap inside the H2 window — the trajectory would remain technically real but lose the economic engine that converts it into the default failover layer this analysis projects.

Glossary

Business and technology

• D2D — Direct-to-Device; connectivity from LEO satellites to unmodified consumer smartphones using standard terrestrial cellular bands.
• MNO — Mobile Network Operator; a terrestrial carrier (T-Mobile, AT&T, Verizon, Vodafone, MTN, etc.) that holds a cellular spectrum licence.
• MSS — Mobile Satellite Service; licensed satellite voice and data service on dedicated mobile-satellite spectrum (distinct from SCS).
• LEO — Low Earth Orbit; roughly 500-2,000 km altitude; the domain of D2D, Starlink, AST SpaceMobile, and Lynk.
• GEO — Geostationary Earth Orbit; ~36,000 km; the domain of Huawei/Tiantong-1’s current D2D implementation.
• LTE — Long-Term Evolution; the 4G cellular air interface, reused in orbit under the SCS framework.
• NTN — Non-Terrestrial Networks; the 3GPP work item that extends cellular standards to satellite, HAPS, and airborne platforms.
• V3 — Starlink’s third-generation direct-to-cell satellite; the throughput step aimed at LTE-parity consumer experience.

Regulation and standards

• FCC — Federal Communications Commission; the US telecom and spectrum regulator.
• SCS — Supplemental Coverage from Space; the 2024 FCC framework authorising orbital use of terrestrial spectrum by SCS-designated operators.
• ITU — International Telecommunication Union; UN body coordinating global spectrum and orbital filings.
• CEPT — European Conference of Postal and Telecommunications Administrations; the 48-country European body that harmonises spectrum decisions across 27+ national regulators.
• WRC — World Radiocommunication Conference; the quadrennial ITU treaty conference that updates the Radio Regulations.
• EPFD — Equivalent Power Flux Density; the ITU limit on signal power from non-GEO satellites into GEO systems, designed to protect incumbents.
• STA — Special Temporary Authority; FCC mechanism for time-limited spectrum use (invoked for Helene, Milton, and LA wildfire D2D activations).
• 3GPP — 3rd Generation Partnership Project; the standards body behind LTE, 5G, and NTN (Releases 17 and 18 are the NTN baselines).
• GSMA — GSM Association; the global industry body representing mobile network operators.

Strategic framework

• H1 / H2 / H3 — horizons of the Three-Horizons framework: H1 “Maintain & Defend” (0-2 years), H2 “Build & Scale” (2-5 years), H3 “Explore & Transform” (5+ years).
• S-curve — the technology lifecycle curve (emergence → rapid growth → maturity → decline) used to locate D2D operators against one another.

Defence and dual-use

• CJADC2 — Combined Joint All-Domain Command and Control; the US-led multi-domain battlefield command architecture, referenced here for Ukraine’s Delta integration as an operational dual-use precedent.

Primary Sources & Research

Federal Communications Commission (2024). FCC Adopts Rules for Supplemental Coverage from Space. FCC. https://www.fcc.gov/document/fcc-adopts-rules-supplemental-coverage-space

NASA SCaN (n.d.). Spectrum Management. NASA. https://www.nasa.gov/directorates/somd/space-communications-navigation-program/

European Space Agency (n.d.). Non-Terrestrial Networks for 5G/6G. ESA. https://www.esa.int/

NTIA (2025). BEAD Program Technology-Neutral Update. U.S. Department of Commerce. https://www.ntia.gov/

International Telecommunication Union (2023). World Radiocommunication Conference 2023 Final Acts. ITU. https://www.itu.int/

Deloitte Insights (2026). TMT Predictions 2026: Next-Gen Satellite Internet. Deloitte. https://www.deloitte.com/us/en/insights/industry/technology/technology-media-and-telecom-predictions/2026/next-gen-satellite-internet.html

Deloitte Insights (2023). Signals from Space: Satellite-to-Phone Connectivity. Deloitte. https://www2.deloitte.com/

CSIS — Pearl, M. and Swope, C. (2025). Modernizing Satellite Spectrum Rules Is Key to U.S. Space Leadership. Center for Strategic and International Studies. https://www.csis.org/analysis/modernizing-satellite-spectrum-rules-key-us-space-leadership

CSIS — Bondar, K. (2024). Does Ukraine Already Have Functional CJADC2 Technology?. Center for Strategic and International Studies. https://www.csis.org/analysis/does-ukraine-already-have-functional-cjadc2-technology

East Asia Forum — Taylor, M. (2026). Starlink, China and the Governance of Low Earth Orbit. East Asia Forum. https://eastasiaforum.org/2026/02/19/starlink-china-and-the-governance-of-low-earth-orbit/

East Asia Forum (2026). AI, Standards, and the US-China Tech Competition. East Asia Forum. https://eastasiaforum.org/

Observer Research Foundation (2025). India’s Satellite Broadband Authorization. ORF. https://www.orfonline.org/

Observer Research Foundation (2026). The Dual-Use Dilemma of Commercial Satellite Connectivity. ORF. https://www.orfonline.org/

GSMA (2024). The State of Mobile Internet Connectivity 2024. GSMA. https://www.gsma.com/somic/

Lee, N. et al. (2013). Power Control for D2D Underlaid Cellular Networks. arXiv. https://arxiv.org/abs/1305.6161