Strategic Autonomy Needs a Third Rail in LEO

A country can reject the middle-power label and still depend on somebody else's permission to use its communications. Ukraine's October Starshield restrictions expose that gap. Low Earth orbit networks now carry military operations, public administration and commercial activity whose continuity has sovereign consequences. Countries with strategic ambitions need control over that dependence. A credible third rail would supply it through an independent operator, protected customer access and infrastructure that keeps essential missions running under political disagreement. The task is to build that capability before dependence hardens.

Report date: October 27, 2026\

Information cutoff: October 27, 2026, 00:55 SGT (UTC+8)

Cover: OneWeb satellites prepared for India’s LVM3 M2 mission, photographed at Sriharikota in October 2022. Archival image of constellation hardware; cropped from the original. Source: ISRO. Source: ISRO; ISRO Copyright Policy.

This essay extends Singapore's Quiet Criterion for Technology Partnerships and The Third Pole in Orbit: OneWeb's Soul, DayOne's Method. The first examined sovereign deployability; the second proposed a Singapore-headquartered orbital-compute business with multinational capital and independent operating control. October's events sharpen the service that such an enterprise would have to sell.

---

Singapore Space Agency is an independent research platform, not a government agency. The infrastructure and venture proposals below are the author's analysis.

---

1. The 90-Second Summary

LEO connectivity has become strategic infrastructure. Once essential missions depend on it, price and coverage cease to be sufficient purchasing criteria. A country needs to know who can change the service, which government can compel that change, and what remains usable afterwards.

On October 14, S. Jaishankar resisted describing India as a middle power. His broader warning concerned a world organised around chokepoints. That supplies the political question. Ukraine supplies the operational test: restrictions associated with terminal movement disrupted military missions even while other connectivity remained available.^[1]^[2]^[3]

Ukraine's defence ministry reported restored connections above 400 km/h on October 24. The initiating authority behind the restriction remains contested. The uncertainty exposes a decision chain the beneficiary could neither clearly identify nor independently control.^[4]^[5]

A third rail, as used here, is an independently governed supply option that distributes dependence across jurisdictions and protects customers' authorised service. National ownership is another route to autonomy. Commercial diversification buys useful resilience along the way. These arrangements have different costs and powers; a second subscription alone does not create a third rail.

Singapore should pursue the operating-company opportunity: aggregate regional demand, secure capacity and build customer protections into a business that can finance its own renewal. The earlier orbital-compute proposal takes that logic further. Its engineering and paying demand still need demonstration; the governance problem is already visible in communications.

2. Jaishankar Named the Problem: Chokepoints

The October 14 exchange came from a Bloomberg New Economy Forum panel in New Delhi with Rishi Sunak, Anita Anand and Kevin Rudd. Jaishankar called India a power in the middle and argued that the middle-power label constrained its potential. He then turned to independent policies and arrangements among countries pursuing their own interests.^[1]^[6]

The stronger bridge to satellites is his description of a “world of choke points.” Rare earths were an obvious example, he said, but finance, sanctions and control over the dollar also created leverage. Sunak argued for making a country indispensable through its capabilities. Together, those remarks describe both sides of dependence: the advantage held by the supplier and the exposure borne by the customer.^[2]^[6]

Communications belongs in that discussion. A network's ability to authorise terminals, set geographic boundaries and change movement limits can determine which national choices are practicable. Jaishankar did not propose a satellite policy on that stage. His account of chokepoints gives a useful way to understand what happened in Ukraine: access to an indispensable system can be conditional even after the equipment has been bought.

Source: Prime Minister’s Office, Government of India; GODL-India.

3. Ten Days Exposed the Chain of Control

Starshield is SpaceX's government and national-security offering, encompassing communications, Earth observation and hosted payloads. Its secure communications can draw on Starlink infrastructure. The product name does not imply that each Ukrainian terminal uses an entirely separate satellite fleet.^[7]^[8]

Source: Odapresste; CC BY-SA 4.0.

The effect was selective but serious. A connection available to a stationary command post could fail on a fast interceptor or supply drone. Preserving ordinary internet access did not preserve every military use.

The public explanations did not converge. Reuters' October 15 report, citing two US officials and a Western official, located the order inside the White House; a senior US official denied a presidential request. The Wall Street Journal subsequently attributed the decision to Musk, citing US officials and people familiar with the matter.^[14]^[15]

On October 21, Marco Rubio denied an administration directive or foreknowledge. His remarks distinguished private commercial contracts from Starshield's US military procurement relationship, even as he assigned responsibility to the company. That distinction leaves the central question open: what authority did the government retain, and what could the contractor change?^[16]

In his October 24 X post, Musk alleged that Ukraine had used civilian Starlink for prohibited offensive drone strikes and said a loophole had been closed. He continued to assign Starshield control to the US government. Ukrainian technical advisers disputed the commercial-use explanation. His statement about Starlink therefore does not settle responsibility for the Starshield restriction.^[17]

Poland's foreign minister Radosław Sikorski challenged Musk publicly as a supplier to European customers. The issue had spread beyond an individual customer's technical complaint to trust in an infrastructure provider.^[18] Russian refinery strikes formed part of the political backdrop, but Reuters reported that Starshield did not operate over Russia. Those strikes cannot simply be treated as demonstrated uses of the restricted service.^[8]

The actors occupied different positions. SpaceX supplied and operated infrastructure; the US government procured relevant military services; European partners supported access; Ukraine depended on it in combat. Paying for a service, contracting for it and operating it are distinct powers. The beneficiary may hold none of the decisive permissions. The congressional letters are valuable because they ask for that authority to be traced; they are requests for evidence, not completed findings.

Earlier episodes show why the answer cannot be to remove network controls. In February, Ukraine and SpaceX used verification and whitelisting to block unauthorised Russian terminals. Centralised exclusion helped Ukraine.^[19]^[20] The 2022 Crimea operation, publicly debated in 2023, concerned a separate dispute about coverage and authorised uses.^[21] October concerns restrictions on authorised Ukrainian military access.

Ukraine demonstrates the cost of opaque dependence. It does not establish that a multinational operator would perform better. A third rail has to earn that claim through its contracts, operating authority and performance under disruption.

4. Public Authority Also Depends on Connectivity

Military demand makes the consequences acute, but civilian institutions already depend on the same infrastructure. Before the Philippines' May 2025 elections, COMELEC deployed Starlink transmission kits in the Davao region. The equipment transmitted results; separate machines read ballots and counted votes.^[22]

Source: Wide Awake!; CC BY 4.0.

That is a sovereign function carried over a commercial network. Authentication, service availability and an alternative transmission route matter even when the provider has no role in counting. A delayed connection can obstruct administration and weaken confidence without changing a single ballot.

Public discourse involves a different mechanism. An access network governs whether and how users connect; a platform governs recommendations, moderation and distribution within its service. Musk's interests in both SpaceX and X put those forms of influence close together, but shared ownership is not evidence that satellite controls manipulated an election. Each mechanism needs its own evidence.

The policy purpose should be to preserve institutions' ability to function and citizens' ability to communicate. Domestic ownership without accountable restrictions can reproduce arbitrary shutdowns at home. Sovereignty requires legitimate authority over the infrastructure, including limits on how that authority is exercised.

5. The Alternatives Have Their Own Sovereigns

There are real alternatives and a large gap in scale. SpaceX reported more than 11,000 Starlink satellites in LEO on October 6. Eutelsat describes a OneWeb constellation of more than 650. Its deployment supplies an existing non-US option, but spacecraft counts alone cannot measure regional capacity, terminal support or high-speed military authorisation.^[23]^[24]

Source: ISRO; ISRO Copyright Policy.

Europe's IRIS² remains future capacity. The Commission's August implementation agreement specified 348 satellites, comprising 330 in higher LEO and 18 in MEO, with first launches in 2029. Its defence and security emphasis expresses Europe's demand for autonomy. An Asian customer would still need its own access rights.^[25]

OneWeb's ownership supplies no automatic neutrality either. Eutelsat's post-recapitalisation prospectus records the French state at 29.65%, Bharti Space at 17.88% and the UK government at 10.89%. Eutelsat tells the UK Parliament that OneWeb is UK-licensed and UK-operated under UK jurisdiction. A foreign customer's rights cannot be read from that shareholder register.^[26]^[24]

The choice extends beyond Europe. Amazon reported an Army demonstration in October, adding another US supplier rather than another home jurisdiction.^[27] China's Qianfan and Guowang programmes are expanding the potential supply base: MIIT reported Qianfan had launched 256 satellites by September 16, and Xinhua reported a further satellite-internet batch in October.^[28]^[29] Those milestones do not establish equivalent service for an overseas mission. Choosing a Chinese network also relocates sovereign dependence; it does not abolish it.

Ukraine's own interest in an alternative is concrete. Nokia and ICEYE signed an October 22 letter of intent with its defence ministry to explore sovereign satellite communications, with an intended architecture covering national ownership and control of satellites, ground systems and terminals. The announcement establishes a development proposal, without a deployed replacement or demonstrated delivery economics.^[30]

The market offers competing architectures, political relationships and stages of readiness. It does not offer an off-the-shelf neutral equivalent to every Starshield mission. Building an additional option remains a strategic task.

6. A Contract Cannot Overrule a State

The most important objection to shared infrastructure is legal. A contract cannot make prohibited service lawful. Applicable mandatory rules can override private arrangements; the UNIDROIT Principles of International Commercial Contracts expressly preserve them. A governing-law clause must be read alongside the operator's licences, export restrictions, sanctions exposure and emergency powers in relevant jurisdictions.^[31]

Source: Rainer Lippert; Public domain.

Customer rights still matter. They can limit commercial discretion, establish responsibility and give the customer remedies against the party that promised service. Where a sponsoring government or reseller signs the contract, the actual user needs defined rights and a route to invoke them. An urgent dispute process may help secure relief; an eventual damages award cannot recover a lost interception. Legal recourse and operational continuity must be designed separately.

The architecture needs four layers:

These are requirements for a proposed service, not features established by the Ukraine contracts. For civilian emergencies, a provider could sell defined continuity and priority. Military service requires agreed end uses and boundaries. The documents should identify lawful restriction procedures, including emergency exceptions, rather than promise universal access to every belligerent.

The earlier third-pole venture essay offers a corporate structure for this task. It proposed a Singapore-headquartered operator with capped, non-controlling sovereign and strategic capital. Management would select suppliers through competition. Collective protections against a change of control would be narrow; each customer would hold protections over its own contracted capacity. A state investor would gain no unilateral right to interrupt another customer's service.^[32]

Those obligations must bind founders and management too. Within the lawful service scope, changes to contracted mission permissions should require documented approval beyond one person's discretion. Emergency security restrictions need defined authority, a review deadline and retained records. Day-to-day operations can remain fast without making customer access a personal decision.

That division remains the right starting point. It separates financing an infrastructure asset from directing its daily use. It also needs protection against its own failure modes: collective vetoes can paralyse decisions, political procurement can inflate costs, underfunded replenishment can hollow out capacity, and broader participation can expose sensitive missions. Professional operating authority, restricted information access and funded replacement obligations must survive the founding coalition.

No corporate design removes every government's leverage. The objective is to reduce the functions that one external actor can disable, preserve lawful alternatives and make residual exposure explicit. If all routes depend on the same authorisation or control system, contractual diversity has left the strategic bottleneck intact.

National ownership can supply deeper control where a country can sustain it. Shared infrastructure can spread the cost and bring customers into its governance. Diverse commercial services can reduce immediate dependence. For a third rail to become more than a procurement arrangement, however, an independent operator must turn protected access into a service that customers trust enough to fund.

7. Capacity Must Be Available Where the Crisis Happens

The economic objection is substantial. A smaller country cannot reproduce the leading operator's manufacturing, launches and user base by declaring connectivity strategic. The first purchase should be a defined set of essential functions, with evidence that a second path can carry them.

An illustrative civilian programme might cover 10,000 endpoints. If 25% were active simultaneously at 2 Mbps each, the requirement would be 10,000 × 25% × 2 Mbps = 5 Gbps. That low-load case could describe scheduled reporting with controlled priority. A regional emergency can correlate demand across sites; if all endpoints need the service together, the same assumption becomes 20 Gbps. The concurrency factor has to follow the mission plan.

Before seeking quotes, the buyer must specify locations, traffic types, peak demand, latency, acceptable interruption and restoration time. Aggregate global capacity does not guarantee an available beam over the affected region.

The following arithmetic prices only assumed subscriptions, equipment and programme operations. All inputs are author-selected US-dollar scenarios. It excludes specialised military connectivity and does not price assured continuity.

Installation, taxes, mobility approvals and financing are also excluded. The subtotal is neither a vendor offer nor a minimum market price. The decisive cost remains unpriced: usable capacity in the specified region during simultaneous demand. A procurement decision needs that commitment, its legal exceptions and a demonstration that it can be delivered.

An exercise should disable the primary provider, load the backup with representative crisis traffic and measure essential services. It should also remove a gateway or terrestrial backhaul route and test the degraded mode. Fibre, radio, GEO and MEO may carry different parts of the workload. Replacing a low-latency LEO link with another orbit may preserve messaging while losing a time-sensitive mission; the result must be recorded by function.

Source: NASA; Public domain.

Local gateways are useful shore infrastructure, but their location does not confer authority over satellite scheduling, terminal authorisation or the operator's software. Laser-linked constellations can route traffic through other gateways. Local facilities should therefore be purchased for defined routing, resilience and security functions, with the operator's retained powers disclosed.^[33]

Commercial viability belongs in the same assessment. Eutelsat reported €297 million of LEO revenue for the year to June 2026, up 69.5% like for like. That is evidence of a growing alternative business, not a demonstration that its capacity matches every Starlink use.^[34] A third rail needs recurring customers and replenishment finance; a protected ownership structure cannot pay for replacement satellites.

8. Singapore Should Build an Operator Around Regional Demand

India has the market and industrial scale to pursue domestic capability while buying useful foreign services. On October 9, its communications minister named Starlink, Jio Satellite Communications and Eutelsat OneWeb as licensees, with commencement subject to security and regulatory requirements.^[35] Those requirements govern admission to the market. Critical users must separately secure continuity. Domestic participation, shared capacity and competition can all contribute without turning autonomy into a refusal of foreign technology.

Singapore's opening is an operating business that aggregates demand across a region too fragmented for each customer to finance its own network. Maritime operators, critical enterprises and public agencies need different services, but can support common infrastructure if their capacity and security obligations are kept distinct. Can Marine's February OneWeb agreement demonstrates an Asia-Pacific maritime distribution channel from Singapore.^[36] It offers a starting point for assembling customers, not national control over the constellation.

Source: Aatu Dorochenko / Aaaatu; CC BY-SA 4.0.

The first work should be specific:

1. Contract a regional workload. Anchor customers should identify the functions they will fund through a disruption: shore-to-vessel operations, emergency coordination, critical-site backhaul or government reporting. Reserve capacity by location and priority, with separate authorisations for military uses.
2. Operate the shore segment and prove the alternate path. Specify gateway diversity, independent backhaul and power, security responsibilities and incident access. Demonstrate service through loss of a provider or gateway. A compulsory local gateway alone would leave much of the foreign control chain untouched.
3. Finance against demand and service rights. Use paid pilots and anchor contracts to support procurement. Keep state investors out of routine mission decisions and supplier selection. Commit replenishment funding before expanding assets.

This is the entrepreneurial logic of the third-pole proposal: one accountable company takes demand to suppliers and takes responsibility for delivery to customers. Singapore supplies a jurisdiction, operating capability and commercial discipline; regional partners contribute customers, capital and competitively priced technology. It is a design worth pursuing, with its first proof in paid service and exercised continuity.

The company would initially depend on existing networks. That dependence should be disclosed and tested. Owned or jointly financed assets become justified where they remove a demonstrated bottleneck that purchased capacity cannot solve. A business that remains an ordinary reseller cannot sell away the upstream operator's veto merely by changing the invoice.

9. Orbital AI Adds Control Over Execution

Computing extends the dependence from carrying a task to executing it. A customer may retain its encryption keys yet lose access to the processor, scheduler or downlink. That is the governance issue behind the earlier third-pole orbital-compute proposal.

Source: Oleg Alexandrov; CC BY-SA 4.0.

The technical case must earn its own economics. Google's September Suncatcher update described experimental work on orbital TPUs; Planet subsequently announced the prototype's launch and initial contact. Radiation, power, heat rejection and interconnection remain engineering questions. A demonstration payload supplies no evidence of a commercially competitive orbital cloud.^[37]^[38]

If a customer develops a valuable orbital workload, its contract must cover tasking, priority, data and model portability, updates and exit. The venture should begin with paid, limited demonstrations and expand only against verified requirements. Ukraine establishes the consequence of dependent access in communications. Whether buyers will pay a sovereignty premium for orbital compute is a separate demand test. The proposal remains worth developing because the control problem can be designed for early, before a useful service becomes concentrated in systems governed elsewhere.

10. The Next Proof Is Operational

Congress has requested the Starshield instruction chain, contractual powers and notification record. The October 15 letter sought an initial briefing by November 1. Those records may settle responsibility for this episode; they will not transfer operational control to Ukraine.^[9]

The alternatives face their own tests. Nokia and ICEYE need a funded programme and a delivery specification. IRIS² needs to meet its revised deployment plan. A Singapore-led venture needs customers willing to sign for protected service, plus evidence that its governance and alternate paths work when a participant disagrees or an asset fails.

A country whose essential choices rely on LEO should invest in its control over that dependence. For states unable to sustain a complete national system, an independent, commercially viable third rail deserves deliberate demand, capital and institutional support. It must reduce the external veto over critical missions, while retaining the operating discipline to keep the network alive. That is an infrastructure business worth building.

All information comes from public sources, including official records, company disclosures and attributed reporting. The proposals and cost scenarios represent the author's independent views and are not investment, legal or procurement advice.