LEO in APAC 2026: Five Markets, the GEO Transition, and the Regulatory Logic Behind the Constellation Race

There is a familiar way to tell the Asia-Pacific satellite story in 2026. Starlink is expanding. Amazon Leo is coming. China is building rival constellations. The connectivity gap is finally being closed.

That version is not wrong. It is just too smooth to be useful.

What is actually unfolding across APAC is messier and much more interesting. The region is not one LEO market. It is not even one satellite-access problem. It is a collection of national bargains about sovereignty, ground infrastructure, incumbent operators, disaster resilience, enterprise demand, and political comfort with foreign networks. Put that in plain language and the conclusion is simple: the same constellation can look like a breakthrough in one market, an overpriced niche in another, and a strategic risk in a third.

In conversations in Singapore over the past year, one thing has come up more often than satellite counts or launch cadence. Regional buyers want to know who can get a gateway licensed, who can support an enterprise contract across several jurisdictions, and who will still pick up the phone when service quality drops after the first wave of excitement. That is a more grounded question than most public market commentary asks, and at the moment it is probably the right one.

Last year, in LEO Constellations and ASEAN Connectivity: Coverage Is Only the First Layer, I argued that Southeast Asia should not read LEO as a simple last-mile broadband story. That argument still holds. If anything, it now looks understated. The more useful frame in 2026 is that APAC contains several LEO markets at once, each with different economics and different political logic, while China's parallel constellation push is slowly turning satellite connectivity into a governance question as much as a telecom one.

The core claim of this essay is straightforward: in APAC, regulation is not a barrier sitting outside the market. Regulation is the market. The operator that secures the right local structure for gateways, landing rights, reseller relationships, and institutional legitimacy often wins long before the technology argument is settled.

The Market Did Not Begin With LEO

A lot of current writing jumps straight into low Earth orbit as if the old architecture were merely obsolete background. That skips the part that explains why this transition matters.

For decades, geostationary orbit was the default commercial layer for satellite communications. GEO worked because it made selling capacity look simple. A small number of satellites could cover very large areas, and operators could sit upstream selling transponder capacity in C-, Ku-, and Ka-band to telecom operators, broadcasters, and enterprise integrators. The downstream customer relationship belonged to somebody else. The GEO operator collected rent on orbital real estate.

That model built real regional businesses. Malaysia's MEASAT remained important across Malaysia, Southeast Asia, and parts of South Asia. Japan's Sky Perfect JSAT became the dominant commercial satellite operator in the region's most advanced economy. Indonesia relied on PSN and Telkom Satellite to help serve an enormous archipelago. Kacific carved out a role in Pacific islands and rural Southeast Asia. Thaicom kept a position in Thailand. None of these businesses were imaginary. The point is not that LEO made them disappear overnight. It is that LEO has made the old GEO growth story harder to defend.

The usual explanation begins and ends with latency, but it is worth stating more plainly than that. GEO was designed for a world in which broadcast distribution and wholesale backhaul were more important than real-time interaction. Once cloud software, video calling, digital payments, remote work, and modern enterprise applications became normal, GEO's built-in delay stopped being a tolerable inconvenience and started becoming a structural handicap. The problem was not that GEO was bad. The problem was that the internet changed what users expected from a connection.

Starlink's low Earth orbit network, operating around 550 km above Earth, pushed that expectation gap into the open. The difference between roughly 20-40 ms and roughly 600 ms is the difference between a service that feels like an extension of terrestrial broadband and one that constantly reminds the user it is operating through a different physics stack. In APAC, where satellite links are often expected to fill in for missing terrestrial infrastructure rather than simply support television or wide-area backhaul, that difference matters a great deal.

But there was always more than one problem with GEO. The wholesale pricing model meant the final consumer offer was usually too expensive for broad middle-income adoption once intermediary margins were added. Coverage geometry imposed real constraints at high latitudes and over awkward geographies. And fixed orbital positions limited how supply could follow changing demand. Those limitations are visible in every market where satellite broadband remained technically available but commercially marginal.

This is why the industry's current pivot is not "GEO dies, LEO wins." It is a layered reorganization. GEO is being repositioned, not erased. Broadcast, non-latency-sensitive backhaul, redundancy layers for government and military customers, and markets where LEO cannot yet secure local infrastructure will all keep GEO relevant. The question is not whether GEO survives. It is what kind of company survives with it.

The best answer so far is multi-orbit. Eutelsat is the clearest example, operating 31 GEO satellites while also running the OneWeb LEO constellation at commercial scale. Australia's NBN Co arrangement with Amazon's Project Kuiper, now Amazon Leo, is the clearest market signal: a national broadband infrastructure operator preparing to replace its GEO service layer with LEO over time. MEASAT's decision to buy another GEO satellite while also signing a February 2025 MoU with SPACESAIL for LEO broadband, direct-to-device, satellite IoT, and Earth observation says the same thing in a more cautious register. GEO for continuity, LEO for optionality.

LEO Changed The Economics Before It Changed The Narrative

The popular argument for LEO is still latency. The more important argument is business structure.

Large LEO constellations were not historically impossible because nobody understood the orbital mechanics. They were uneconomic. Short satellite lifespans, atmospheric drag, replenishment needs, and the requirement for hundreds or thousands of spacecraft made the concept capital-hungry and operationally ugly. What changed was not the physics but the cost curve. Reusable launch, especially through Falcon 9, brought launch costs down sharply. Satellite manufacturing became more standardized. Terminal technology improved enough to reduce end-user hardware costs. At that point the economics shifted from interesting concept to plausible commercial system.

The market numbers are impressive, though they should be handled with more skepticism than they usually are. Forecasts pointing to Asia Pacific as the fastest-growing LEO region, with 17%+ CAGR through 2033, or to global LEO broadband growing from roughly $7.66 billion in 2025 to around $95.4 billion by 2034, are directionally useful. So are estimates that APAC satellite broadband subscribers could reach 38 million by 2030 from about 2.1 million in 2025. But not all forecasts are equally disciplined, and it is easy to stack them into a fake certainty. The more honest reading is that the growth direction is real, while the exact slope is still heavily dependent on regulation, ground infrastructure, and whether enterprise demand arrives on schedule.

The more durable shift sits elsewhere. GEO operators were wholesalers. Starlink is vertically integrated across launch, satellite manufacturing, terminals, network operations, and a growing share of the customer relationship. That is the real break. A constellation operator that also behaves like a retailer can price by segment, change offers quickly, and capture value that used to be spread across several intermediary layers. That is why incumbent GEO operators are not really defending themselves against a lower orbit. They are defending themselves against a new value chain.

Starlink Set The Pace, But The APAC Story Is More Specific Than The Headlines Suggest

It is still important to be clear about the scale. Starlink has expanded to 155 countries, reached more than 10 million subscribers, and accounted for 97.1% of all satellite Speedtest samples globally in Q3 2025 according to Ookla's 2025 global satellite broadband performance report. By late 2025, SpaceX had put more than 9,000 satellites in orbit. Manufacturing cadence was running around 4,000 satellites annually. In 2026, Falcon 9's dedicated Starlink mission rhythm was on a level no rival could currently match.

And yet the APAC story is not simply "Starlink arrives, everyone else lags." Through December 2025, the region accounted for roughly 430,000 direct Starlink subscribers. Indonesia was the largest APAC market by subscriber count, with Australia next. Those are meaningful numbers, but they also show how regulatory friction slows the conversion of technological lead into regional scale.

The part that is still underappreciated is the role of gateways. Gateway deployment is often a better leading indicator than subscriber growth because it tells you whether a market entry was merely licensed or actually operationalized. This is one reason performance in APAC diverges so sharply. Ookla found that New Zealand posted Starlink latency of about 35 ms in Q4 2025, the lowest globally, while Australia reached median download speeds of 162.47 Mbps. Markets relying on more distant gateway arrangements, such as East Timor and the Maldives, were materially worse. Bangladesh also benefited from local infrastructure. That divergence is not a quirk. It is the direct result of how local regulatory agreements shape ground build-out.

Indonesia illustrates the point better than any abstract model could. Device growth ran hot, up 33.9% in 2025, largely on enterprise and government demand, while median performance slipped from 45.16 Mbps to 40.69 Mbps. It is tempting to read that as generic network strain. The more useful interpretation is that demand scaled faster than the regulatory arrangement allowed supporting infrastructure to scale. In APAC, poor regulatory bargains and disappointing service outcomes often turn out to be the same story told from different angles.

The same logic applies to direct-to-cell. By Q3 2025, more than 600 Starlink satellites had been designed specifically for DTC services. That matters because DTC changes the addressable market. It turns satellite connectivity from a fixed broadband substitute into a mobile coverage extension layer. In archipelagic, mountainous, and remote-island geographies, that is commercially more disruptive than residential satellite internet.

The Globe Telecom live pilot in the Philippines on March 17, 2026 made that disruption concrete. Standard LTE handsets connected to Starlink's DTC service without new hardware in Rizal, Batangas, and Bataan. Users completed real tasks, including GCash transfers. That is more important than a standard product demo because it shows DTC surviving outside the lab and under actual Southeast Asian operating conditions.

One lesson from that pilot is technical. Another is political. The rollout template is becoming visible: local MNO partnership, high-level political sponsorship, pilot deployment in underserved provinces, then commercial expansion once spectrum coordination and licensing are settled. The next set of markets likely to move in that direction includes Indonesia, Vietnam, Bangladesh, and eventually India.

There is another reason to avoid reducing Starlink's APAC story to household broadband. The region's most valuable verticals are maritime and aviation. Starlink's enterprise revenue rose from about $584 million in 2024 to $1.38 billion in 2025 and an estimated $1.68 billion in 2026. Maritime revenue alone is forecast to reach $1.94 billion globally in 2026, while aviation is entering a major installation cycle. That matters in APAC because the region contains the Strait of Malacca, the South China Sea, Northeast Asia's export routes, and one of the world's densest concentrations of maritime demand. Korea's commercial shipping fleet is the fifth largest globally. Japan remains a meaningful in-flight connectivity market. Singapore sits at the center of one of the most valuable maritime enterprise clusters any LEO operator can pursue.

The Challengers Are Not Chasing The Same Prize

One reason public discussion gets sloppy is that non-Starlink operators are often grouped together as if they were all trying to run the same race. They are not.

Amazon Leo is pursuing an enterprise-first and infrastructure-partnership path. The NBN Co deal matters so much because it gives Amazon a government-backed beachhead in Australia without forcing a frontal attack on Starlink's consumer base. NBN plans to transition more than 300,000 premises in its satellite footprint to the new service over time, while keeping the Sky Muster GEO satellites running during the migration. The arrangement is analytically important because it is a national-scale GEO-to-LEO conversion with revenue visibility built in.

At the same time, Amazon still has a hardware problem. It rebranded Project Kuiper as Amazon Leo in November 2025 and entered enterprise beta in April 2026, but as of late 2025 it had fewer than 200 satellites in orbit against an FCC deployment deadline that looked increasingly unrealistic. Amazon has bought launch capacity aggressively, yet much of that capacity depends on vehicles that either are still maturing or have not flown at the cadence Kuiper once assumed. A deadline-extension request to the FCC still looks likely. So the right way to read Amazon Leo in APAC is not as an immediate Starlink killer. It is a serious institutional entrant with a strong Australian anchor and a still-unfinished deployment machine.

OneWeb is playing a different game altogether. With 654 satellites at roughly 1,200 km altitude, typical throughput around 50-195 Mbps, and latency in the 30-70 ms range, it is the only other fully operational global LEO network besides Starlink. But it remains deliberately focused on enterprise and government demand rather than direct consumer broadband. In practice, that means service-level agreements, managed installation, distribution partnerships, and the ability to combine LEO capacity with Eutelsat's GEO assets when customers want a layered architecture.

The APAC examples are revealing. In Singapore, Can Marine's multi-year agreement with Eutelsat OneWeb is a small story in public visibility but a meaningful one commercially. A maritime integrator based in the region is using OneWeb to support merchant shipping and offshore energy clients across APAC. In Korea, the Ministry of Science and ICT approval for Hanwha Systems and KT Sat to supply OneWeb services gives the constellation a natural opening into maritime, industrial, and government segments.

Eutelsat's capital decisions also show that it is not treating OneWeb as a side bet. The company expanded its Airbus order to 440 next-generation satellites, including 340 added in January 2026, with 5G integration and compatibility with Europe's IRIS2 framework. That matters in APAC because Japan, Korea, and Australia increasingly care about non-terrestrial network integration rather than satellite broadband as a standalone product. The balance sheet, however, still looks strained. Eutelsat's March 2026 refinancing underlined that point. Operationally, though, management pointed to roughly 80% topline growth in 2025. OneWeb is one of those cases where the commercial and financial stories are moving in different directions at the same time.

AST SpaceMobile is different again. It is not really trying to build a classic satellite broadband market. It is trying to become an invisible extension of mobile networks. Its large-format satellites are designed to connect with ordinary handsets using existing cellular spectrum. That moves the commercial relationship away from retail users and toward mobile operators that want broader geographic coverage without rebuilding their whole terrestrial economics.

The company is still pre-scale operationally, with five satellites as of late 2025, so it should not be romanticized. But the partnership architecture is substantial. AST's SEC disclosures and company updates point to agreements or expanded initiatives with Orange, Telefonica, CK Hutchison, Taiwan Mobile, Sunrise, and Vodafone, and management has framed the commercial ecosystem as involving 50+ MNO partners and nearly three billion subscribers. It also referenced more than $1.2 billion in contracted revenue commitments. For APAC, the Taiwan Mobile and CK Hutchison ties matter because they place AST inside markets where rural gaps are real but operators may prefer coverage extension over building a whole new orbital business line themselves.

APAC Is Not One Market

The easiest way to make this article feel artificial would be to give every sub-market the same amount of space. Real markets do not deserve equal treatment just because they all appear on the same map.

Oceania is the maturity benchmark. Australia and New Zealand already show what happens when a market gets relatively clear regulation and enough supporting gateway density. New Zealand's 35 ms latency and Australia's 162.47 Mbps median downloads in Q4 2025, both highlighted by Ookla, are not just performance trivia. They show what LEO service looks like when the ground layer is allowed to keep pace. Australia also shows what maturity does commercially: Starlink had more than 350,000 active customers there as of April 2025, and now Amazon Leo is entering through NBN rather than through open-market chaos. Oceania is moving from first-wave adoption into second-wave market design.

Northeast Asia is the enterprise theatre. Japan and Korea are not where LEO becomes a mass household substitute for terrestrial broadband. Their terrestrial networks are too good for that argument to be persuasive at scale. The opportunity is in maritime, aviation, industrial, and public-sector resilience. Korea's shipping position makes maritime connectivity the obvious first wedge for Starlink and OneWeb. Japan's DTC direction through KDDI and NTT DOCOMO matters because it tests whether satellite-mobile integration can become a serious commercial service in a highly saturated terrestrial market. If that model works in Japan, it becomes much easier to pitch elsewhere.

Southeast Asia is where the market is largest on paper and most fragmented in practice. The Philippines remains the clearest pioneer. The DTC pilot with Globe mattered not only because it worked, but because it showed that Manila is increasingly willing to treat satellite connectivity as public infrastructure. Indonesia is the biggest prize and the hardest bilateral negotiation. More than 17,000 islands, a rural population in the hundreds of millions, and a licensing environment that deliberately filters foreign operators create a market where every incremental regulatory concession matters. Vietnam's February 2026 opening under a five-year pilot with no foreign ownership limits was one of the region's most important surprise signals. Malaysia is strategically fascinating because MEASAT is hedging across GEO continuity, possible Qianfan distribution, and Starlink-adjacent logic at the same time. Thailand's Sirindhorn gateway with National Telecom gives OneWeb a regional support node that can matter even where Thailand's own consumer market is not the immediate focus. Singapore, meanwhile, remains less important as a household market than as an operating base, contracting hub, and maritime coordination point.

Then there is India, which still has the biggest unrealized upside in the region. One billion-plus internet users, rural broadband penetration that remains shallow by national scale, and explicit policy recognition that satellite broadband matters would normally be enough to trigger a stampede. Instead, India is forcing everyone to confront the difference between demand and executable access. Starlink became the third licensed satcom operator there in mid-2025, joining Eutelsat OneWeb and the Jio-SES venture, and its Gen 1 constellation was approved by IN-SPACe. But that still leaves trial spectrum, data-localization requirements, local security approvals for each gateway, lawful interception obligations, and indigenization expectations for at least part of the ground infrastructure. The debate around TRAI's recommended 4% AGR-based spectrum fee matters precisely because India can either become the largest APAC LEO growth engine or remain slowed by administrative drag.

The India story also contains one of the clearest commercial tells in the whole region: both Bharti Airtel and Reliance Jio have chosen to distribute Starlink rather than wait for a domestically controlled LEO answer to appear on a commercially meaningful timeline. That does not eliminate competition, but it does reveal how the incumbent telecom logic has shifted.

The final piece is the China wildcard. It is tempting in Western analysis to treat Chinese constellations either as a distant strategic curiosity or as an all-purpose future threat. Neither reading is especially good.

Guowang is state infrastructure first. The constellation, operated by China Satellite Network Group with direct state backing, is planned at 13,000 satellites, with roughly 163 in orbit as of March 2026 and targets of 310 in 2026, 900 in 2027, and 3,600 annually from 2028. Those targets imply a national industrial effort, not just a commercial rollout. Guowang is better understood as a sovereign control project designed to reduce dependence on foreign communications networks and secure China's own orbital and frequency position. Its larger satellite class, around 16,600 kg, also keeps the dual-use dimension squarely in view.

Qianfan is the more commercially relevant Chinese constellation for APAC. Operated by Shanghai Spacesail Technologies and marketed internationally as Sailspace, it is the first Chinese LEO system actively pursuing wholesale telecom partnerships abroad. It had more than 180 satellites in orbit by April 2026 after a difficult 2025 that included tumbling satellites, upper-stage debris issues, and a six-month launch suspension before launches resumed in October. That setback matters because it reminds us that deployment ambition and deployment competence are not the same thing.

Still, the commercial intent is clear. The MEASAT-SPACESAIL MoU covers LEO broadband, direct-to-device, satellite IoT, and Earth observation across Malaysia, Southeast Asia, and South Asia. Songjiang's manufacturing center is already producing around 300 satellites annually and plans to expand to 500-600 by 2026. Qianfan's own deployment ambition, 324 satellites in 2026, another 324 in 2027, and 4,000 in each of 2028 and 2029, is aggressive enough that it should be treated seriously without being treated credulously.

The geopolitical frame around all this is no longer subtle. Singapore became the first ASEAN member to sign the Artemis Accords in 2022, followed by Thailand in 2024, then Malaysia and the Philippines in October 2025. Yet those same countries are still hedging commercially. Malaysia can sign Artemis while MEASAT talks to SPACESAIL. The Philippines can sign Artemis while simultaneously deepening its Starlink relationship. ASEAN is not choosing one side cleanly. It is trying to avoid single-provider dependence.

That is why I think the better analogy is not simply telecom competition. It is the 5G Huawei / non-Huawei split, only slower and more institutionally blurred. Satellite connectivity is becoming part of the region's infrastructure diplomacy.

What I Would Actually Watch Now

If the argument above is right, then the next phase will not be decided by whichever operator says the boldest thing about orbit.

I would watch India's final spectrum and compliance path more closely than almost any regional market forecast. I would watch whether Indonesia's licensing structure begins to allow gateway expansion that matches demand. I would watch whether Amazon Leo's hardware ramp catches up fast enough for Australia to become a true proof point rather than just a promising contract. I would watch whether Qianfan converts outreach and memoranda into real service capability in Southeast Asia. And I would watch direct-to-cell not as a technology spectacle but as a regulatory process, especially after the Philippines pilot.

Those are not equally weighted variables, and they do not all mature on the same timeline. That is exactly the point. APAC satellite connectivity is not a clean regional growth story. It is a sequence of uneven openings in which regulatory alignment, local infrastructure, and enterprise demand will decide much more than marketing narratives do.

Conclusion

The easy conclusion is that APAC is finally becoming a major LEO market. That is true, but it is not yet the most useful conclusion.

The more useful one is that APAC is becoming the region where the difference between technology leadership and market leadership is easiest to see. Starlink still has the biggest industrial lead. Amazon Leo has found a smart institutional entry point. OneWeb remains commercially relevant because it is playing a different game. AST SpaceMobile may yet matter because mobile operators do not all want to become satellite operators. China is building not one answer but several, and some of them are more commercially legible than outside observers still assume.

But the operating reality beneath all of that is less glamorous. Buyers in this region care about gateways, licensing, local support, service continuity, and whether the provider can work through local partners without creating a sovereignty headache. They care about whether maritime and aviation connectivity can be sold as real infrastructure, not a novelty product. They care about whether a foreign constellation can be politically tolerated when things get tense.

That is why the APAC LEO story in 2026 should not be written as a race to blanket coverage. It should be written as a fight over who gets to become the trusted infrastructure layer in very different national systems.

And at the moment, that fight is still wide open.

Sources

1. LEO Constellations and ASEAN Connectivity: Coverage Is Only the First Layer
2. Ookla 2025 Global Satellite Broadband Performance Report
3. NBN Co selects Amazon's Project Kuiper to bring LEO satellite broadband to Australia
4. MSIT approves cross-border supply agreements for Starlink and OneWeb in Korea
5. MEASAT partners with SPACESAIL to advance LEO satellite services
6. Can Marine to deploy Eutelsat OneWeb for Asia-Pacific maritime connectivity services
7. Thailand's NT and Eutelsat OneWeb gateway launch plans at Sirindhorn station
8. Eutelsat taps Airbus to build 340 additional OneWeb satellites
9. AST SpaceMobile SEC filing / business update on MNO partnerships
10. AST SpaceMobile 2025 results and partnership update
11. Globe Telecom newsroom, including Starlink Direct-to-Cell pilot update
12. Globe Telecom seals agreement with Starlink to launch Southeast Asia's first direct-to-cell service
13. State Department note on Malaysia and the Philippines signing the Artemis Accords
14. India licence update and TRAI spectrum-fee context
15. Industry summary of India's satellite internet market and operator positioning
16. Starlink performance and regulatory divergence in APAC market reporting
17. OneWeb maritime deployment coverage through Smart Maritime Network
18. Thailand gateway service coverage summary via The Exchange Asia