11,000
Satellites filed
Largest ITU filing outside US and China
$100M+
LOIs signed
In first 30 days from network operators
~$250k
Cost per satellite
vs ~$5M industry standard (20× under)
Feb '27
First launch
Booked and paid — self-funded by founder
Backed by
Thesis
Exosat is the third leg of the global satcom stack — a neutral, Singapore-based LEO constellation for every country and enterprise that will not stake critical infrastructure on Starlink or Chinese state operators.[1][2]
Ed Ge is manufacturing communications satellites at ~$250k instead of ~$5M using automotive-grade parts through East Asian supply chains, has filed for an 11,000-satellite constellation with the ITU (the largest outside the US and China), and — within 30 days of incorporation — booked a first launch, signed $100M+ in LOIs, and built the first satellite in his apartment.
If Starlink is the American internet in space, Exosat is what everyone else builds.
- 01
A third power in orbit is a $1T+ opportunity. Ukraine, Taiwan, Indonesia, India, and much of the Middle East are actively looking to exit or avoid Starlink and Chinese state constellations. No credible third option exists. The Ukraine kill-switch, India's rollout freeze, Indonesian telco lobbying, and Chunghwa Telecom's OneWeb exit are all datapoints from the last twelve months.[3][4][5][7]
- 02
10× cost advantage, 80% of the market. Sub-500km orbits + East Asian automotive-grade supply chain = ~$250k satellites at 3–5 year lifespan, versus the industry-standard ~$5M. Exosat is not attacking the Starlink monopoly directly — it is going after everything Starlink and Chinese state operators cannot or will not politically serve.
- 03
Launch diversification outside SpaceX. SpaceX for launch one, ISRO for two, then LandSpace — Exosat will be the only Western-aligned space company to ever fly on a Chinese rocket. This is the only realistic path to deploying 10,000+ satellites by 2032 without being SpaceX-gated, and the exact gate that has slipped Kuiper by two years.[8]
- 04
Ed Ge is the founder for this specific problem. Third-time space founder over six straight years — Stratodyne (stratospheric balloons, 2020) as a student at Missouri, then Aethero (space compute, 2023–2026) — millions in ARR from Blue Origin and defense primes, two satellites launched, radiation-hardening research with the Aerospace Corporation and AFRL, 8-figure acquisition offer turned down. Then paid for Exosat's first launch out of his own savings. First YC space company based in Asia.[9][10][12]
Problem
Starlink is now global critical infrastructure — and it's entirely gated by the US government.
Starlink has 8,000+ satellites in orbit, direct-to-cell rolling out, and 24/7 broadband to any device. It is the single most important piece of communications infrastructure to come online since the fiber-optic backbone. It is also gated end-to-end by the US executive branch — a fact that has stopped being theoretical.
Elon flipped the kill switch in Ukraine. India put Starlink rollout on hold pending a sovereignty review.[3] Indonesian telecoms are lobbying the government to restrict Starlink.[4] Chunghwa Telecom in Taiwan is trying to exit its OneWeb contract, but has publicly said it will not replace the capacity with Starlink for national-critical use.[7] Ukrainian drone operators privately admit they'd switch off Starlink for a neutral operator the day one existed.[5]
The obvious alternative — Chinese state constellations Guowang and Qianfan — is a non-starter for the rest of the world. Non-Chinese telcos cannot land spectrum. Non-Chinese enterprises will not route traffic through Beijing. Both problems compound: Chinese consumer OEMs (BYD, Xiaomi, Honor) ship the majority of unit volume outside China, use Chinese satellites domestically, and have no trusted option abroad. AST won't sell to them. Starlink won't sell to them.[12]
The rest of the field is broken. OneWeb is chronically delayed, expensive, and losing customers under contract. Kuiper (Amazon/Blue Origin) slipped ~14 months on Blue Origin pad access alone and had to amend its FCC deployment milestones by two years.[8] AST SpaceMobile is direct-to-cell only, US-domiciled, and still ramping. Iridium is legacy narrowband. Between them, there is no neutral, sovereign, high-bandwidth, affordable satcom option for the ~6 billion people outside the US and China.
8,000+
Starlink sats in orbit
The de-facto US-gated backbone
14 mo
Kuiper launch delay
Amended FCC milestones by 2 years
0
Neutral third operators
At scale, with LEO physics, today
Why Now
The sovereignty rupture, the orbit thesis, non-SpaceX launch, and direct-to-cell — all converged.
Voices from the Exosat launch on Bookface. YC's own group partner called it out in three words.
YC starlink!
Christina Gilbert[1]
Group Partner · Y Combinator
The most serious space startup I've seen launched in ages.
Geffen Avraham[1]
Formerly Apolink (F24) · YC space alum
Congrats man, so proud!!! Ed is an absolute killer.
Skyler Chan[1]
Founder · GRU Space (W26) · ex-Tesla
Four things converged inside a single 24-month window.
Political shock. The Ukraine kill-switch, India's rollout freeze, and Indonesian telco lobbying converted "sovereign satcom" from a policy-paper phrase into a board-level question for every non-US, non-China government. Once one government asks it, none of the neighbors can afford not to.[3][4][5]
The orbit thesis. Sub-500km LEO orbits give better link budgets, lower latency, and a milder radiation environment. That last property is the load-bearing one — it is what unlocks automotive-grade parts and East Asian supply chains for the bill of materials that dominates satellite cost. Every prior LEO player either flew higher (radiation-hardened parts, ~$5M satellites) or hadn't yet redesigned around the shorter-lifetime, mass-manufactured model.
Non-SpaceX launch matured. ISRO's PSLV/SSLV rideshare cadence and LandSpace's Falcon-9-class reusable stack came online just as Blue Origin's Kuiper deployment fell 14+ months behind.[8] This makes SpaceX's launch monopoly the physical ceiling for every rival constellation — except the one company willing and structurally positioned to buy Indian and Chinese launch capacity at scale.
Direct-to-cell went from prototype to standard. AST SpaceMobile and Starlink Direct-to-Cell proved the physics of connecting an unmodified smartphone directly to a LEO satellite works. The remaining question is who controls the standard for the ~5 billion smartphones outside US and Chinese markets. That is the exact market Exosat is designed for.
For a lot of countries in the Middle East with the Iran war, Starlink access has been not guaranteed. And a lot of these countries are looking at it and they're like — well, the US government can turn us off at any given time. And that's national critical infrastructure. But at the same time, we don't trust the Chinese state corporation. Who else is there? The answer is not much.
How It Works
A four-shell, 11,000-satellite constellation — sequenced by product line.
The constellation is four shells across three orbit types.
Exosat's ITU filing covers four shells: equatorial, mid-inclination, and SSO dawn-dusk, with the dawn-dusk shell doing double duty as a distributed compute backbone. Each satellite is a ~$250k build at target scale, delivering 1.2 Gbps of network capacity and connected across the constellation by 100 Gbps optical inter-satellite links. At 100 satellites the constellation delivers up to 120 Gbps of aggregate network capacity.[1][2]
The satellites are built with automotive-grade parts through East Asian manufacturing partners, which is only possible because sub-500km orbits have a mild enough radiation environment to make radiation-hardened components unnecessary. The trade-off is a 3–5 year satellite lifespan instead of the traditional 10–15 — which is not a bug: shorter lifespans are what lets Exosat refresh generations rapidly and amortize R&D across many satellites, exactly the way Starlink does at higher cost.
Three product lines, sequenced by scale
1.2 Gbps
Per satellite capacity
Software-defined radios
100 Gbps
Inter-satellite laser links
Full-mesh optical backbone
4 shells
Filed with ITU
Equatorial · mid-inclination · SSO dawn-dusk × 2
30 days from incorporation
TractionEd self-funded the first launch contract out of his own savings and hand-built the first satellite in a fume-hood clean room inside his apartment. Bookface launch drew responses from YC group partner Christina Gilbert ("YC starlink!") and Apolink founder Geffen Avraham ("the most serious space startup I've seen launched in ages").[1]
The 10× Cost Advantage
The most controversial architectural decision in the entire company.
Communications satellites are supposed to cost $5M and last 10–15 years. Exosat is targeting $250k and 3–5 years — deliberately.
Not radiation-hardened components. Automotive-grade parts.
Why it's defensible. Sub-500km orbits have a mild enough radiation environment that automotive-grade components — mass-produced at automotive industry scale, throughput, and price — meet the reliability requirement for a 3–5 year satellite lifespan. Every prior LEO comms operator either flew higher (needing radiation-hardened parts) or hadn't yet redesigned around a shorter-lifespan, mass-manufactured model. Exosat's orbit + supply-chain choice is what makes the cost curve possible; nothing about it is a shortcut.
Why it produces a better product. Short satellite lifespans compound generation-over-generation. Exosat refreshes hardware in ~4 years instead of ~15, which means the constellation is always running last-generation silicon rather than 10-year-old chips. Starlink already exploits this at its own cost point; Exosat exploits it at ~10× lower cost.
Why the customer economics work. At 100 satellites, Exosat's internal cost model shows $3.20/Mbps/month delivered — versus $50–$200/Mbps/month for GEO incumbents and $10–$30/Mbps/month for Starlink wholesale. At 1,000 satellites the target is $1/Mbps/month. That is not a competitive price — it is a pricing regime the rest of the industry cannot follow without dismantling its cost base.[12]
Cost per satellite at scale
Chart
Build cost drops from $223k (1 satellite) to $78k (100 satellites); lifetime operating cost drops in parallel. Exosat internal model, shared with Orange Collective, July 2026.
Source · Exosat internal financial model [12]
Delivered connectivity cost ($/Mbps/month)
Chart
GEO incumbents run $50–$200. Starlink wholesale is $10–$30. Exosat's internal model shows $3.20 at 100 sats and a $1 target at 1,000 sats — a step-change in the price of orbital bandwidth.
Source · Exosat internal model · public wholesale price bands
Our thesis is simple: we're building a satellite at 10% of the cost to service 80% of the market rather than going after everyone.
Neutrality as Design
Two hard policies. Encoded from day one. Both are what unlock the customer base.
Launch strategy is the same design decision — deliberately diversified.
Three launch providers, three geographies. First mission on SpaceX; second on ISRO; subsequent launches on LandSpace. Ed believes Exosat will be the only Western-aligned space company to ever fly on a Chinese rocket. Diversified launch means Exosat's deployment schedule is not gated by SpaceX pricing or SpaceX manifest priority — the same gate that has slipped Kuiper by two years.[8]
Customer geography. Sales into Latin America, Southeast Asia, East Asia excluding mainland China, the Middle East, Europe, and Africa. Every one of these regions has active pull for a non-US, non-China operator that a Singapore-domiciled Exosat can serve where no one else can. The Indonesian telco JV — Exosat launches a demo pair, telco finances the constellation buildout, 50/50 revenue share nationwide — is the archetype.
Captive markets no one else can address. Chinese consumer device makers — BYD, Xiaomi, Honor, and their peers — ship the majority of unit volume outside China. Inside China they use government satellites; outside they have no trusted option. Neither Starlink nor AST will sell to them. A Singapore-domiciled operator can, and gets a captive customer base of hundreds of millions of devices per year.[12]
Market
Three sequenced markets. All three sit outside the incumbent-served region.
Near term (2027–2028) — IoT NTN. Sparse constellation is enough to serve IoT customers in LatAm and Southeast Asia. First revenue path with a small orbital footprint. Store-and-forward packet delivery avoids the need for continuous coverage.
Mid term (2028–2030) — direct-to-cell and telco capacity leases. As shells fill in, Exosat sells cellular capacity to telcos in target regions and to Chinese consumer OEMs for their non-China unit volume. Signed Indonesian civil government JV is the archetype: demo pair from Exosat, buildout financed by the telco, 50/50 revenue share.
Long term (2030+) — broadband + orbital compute. Full LEO broadband, plus edge inference on-satellite for robotics, autonomous vehicles, and maritime. American Tower buying CoreSite and AT&T/Verizon partnering with AWS are early terrestrial signals that compute and connectivity are collapsing into a single stack — Exosat's compute-backbone shell is designed to intercept the same trend in orbit.[12]
Global satellite communications services market, 2024–2033E
Chart
Novaspace / Euroconsult forecasts the global commercial satcom services market growing from ~$18B (2024) to ~$32B (2033E), with LEO and direct-to-device as the dominant growth drivers. Ed's stated addressable opportunity outside the US and China — connectivity plus edge compute — is $1T+.
Source · Novaspace / Euroconsult · Satellite Communications & Broadcasting Markets Survey [13]
Compute and connectivity get merged into one market in the future. American Tower is buying CoreSite; AT&T and Verizon partner with AWS. As AI becomes more integrated, especially in the physical economy, people want to move the compute to the closest possible location. In maritime, in rural areas, the closest thing might be a satellite flying overhead.
Competitive landscape
Four categories. Exosat is the only company defined by the intersection of all four gaps.
Each competitor category has a structural limitation that Exosat's Singapore-domiciled, low-cost, diversified-launch design is specifically built to exploit.
Filed LEO constellations — Exosat is the third-largest globally
Chart
ITU/FCC filings for top LEO communications constellations. Starlink and China's Guowang/Qianfan sit above 10,000 satellites. Exosat's 11,000-satellite filing is the largest single-operator filing outside the US and China — an order of magnitude larger than every other Western non-US filing.
Source · FCC filings · ITU public filings · operator disclosures
Exosat's positioning
If Starlink is the American internet in space, Exosat is what everyone else builds. Singapore flag, sub-500km orbits, automotive-grade parts, diversified launch, IoT-first revenue path, direct-to-cell physics, orbital compute optionality. Every one of those choices is structurally hostile for the US and Chinese operators — and structurally aligned with the customer base Exosat is going after.
Founder deep dive
Not a first-time founder attempting a moonshot. Six straight years of space companies, and Exosat is the third.
Founder & team
Hiring pattern
Ed is recruiting from Starlink, Planet, and DJI with two explicit filters: willingness to relocate to Asia, and ability (or willingness to learn) Spanish or Mandarin — both are needed for the customer base. The Bookface launch surfaced multiple senior operator intros in the first 72 hours. Adding a strong technical co-founder is at the top of the list of things we are watching.[1]
Risks & mitigations
What we're watching
References
- [1]Exosat — Launch on Bookface (YC internal, S26)
- [2]Exosat — Company Website
- [3]Reuters — India puts Starlink rollout on hold as sovereignty review deepens (2025)
- [4]Financial Times — Indonesian telcos lobby government to restrict Starlink expansion (2025)
- [5]The Wall Street Journal — Ukraine's Drone Operators Weigh Alternatives to Starlink After Coverage Cutoffs (2025)
- [6]Y Combinator — Exosat (S26) company profile
- [7]Nikkei Asia — Chunghwa Telecom seeks exit from OneWeb capacity contract as delays mount (2025)
- [8]Bloomberg — Blue Origin's Kuiper Program Slips Two Years as Launch Pad Access Delays Deployment (2026)
- [9]TechCrunch — Aethero wants to become the space industry's Intel or Nvidia (2024)
- [10]TechCrunch — Aethero Raises $8.4M for Space Computers (2024)
- [11]Reuters — SpaceX Prices IPO at $2 Trillion Valuation in Largest-Ever Industry Debut (2026)
- [12]Orange Collective — Meeting notes with Edward Ge, July 1 2026 (internal)
- [13]Novaspace / Euroconsult — Satellite Communications & Broadcasting Markets Survey, 2024–2033 forecast
- [14]Orange Collective — Starcloud investment memo (data centers in space)
- [15]Fortune — 'Europe doesn't realize how dangerous it is': Telecoms CEO warns of U.S. dominance in satellites, AI (May 21, 2026)
- [16]Stratodyne — high-altitude balloon platform for persistent remote sensing (Ed Ge's prior company, 2020–2023)