For most of commercial aviation’s history, inflight internet was a grudging concession to passenger expectations: slow, expensive, prone to dropping over oceans. The geostationary satellites that powered it were an architectural mismatch for the task — orbiting 35,786 kilometers up, they introduced round-trip latencies of 500–600 milliseconds that made video calls stutter and real work nearly impossible. Airlines sold it anyway, because nothing better existed.

That era is ending. The low Earth orbit constellations built to sell residential broadband are now signing fleet-wide deals with carriers at a pace that would have seemed implausible three years ago. Aviation is shaping up as one of the most commercially consequential battlegrounds for LEO broadband — and the competition between SpaceX’s Starlink and Amazon’s rebranded Kuiper service, now called Amazon Leo, is becoming a genuine two-horse race.

SpaceX moved into commercial aviation earlier and more aggressively than most anticipated. United Airlines signed what it described as the industry’s largest Starlink agreement in late 2024, and by early 2026 had equipped more than 300 aircraft, with a stated goal of around 800 by year’s end and full fleet coverage by 2027. In June 2026, United flew the first Starlink-equipped widebody transatlantic route — a symbolic milestone signaling that LEO connectivity was no longer just for short domestic hops.

Alaska Airlines and Hawaiian Airlines, now operating as a combined fleet under Alaska Air Group, reached roughly 150 Starlink-equipped aircraft by mid-2026, ahead of schedule, and are offering free Wi-Fi to members of their Atmos Rewards program through a T-Mobile partnership. Frontier Airlines and Indigo Partners, which collectively represent a large portfolio of budget carriers, signed with Starlink in July 2026, with Frontier targeting a 2027 service launch covering more than a thousand aircraft.

The scale of these deals — hundreds of aircraft per agreement — reflects something more than ordinary supplier contracts. Airlines are making long infrastructure bets, and the majority are placing them with Starlink.

Amazon Leo Finds Its Anchor Tenant

Against that backdrop, Amazon’s March 2026 deal with Delta Air Lines stands out. Delta chose Amazon Leo over Starlink for a deployment covering 500 aircraft, with installations scheduled to begin in 2028. Each aircraft will receive an antenna capable of up to 1 Gbps download and 400 Mbps upload speeds. Wi-Fi is expected to remain complimentary for SkyMiles members through Delta’s existing Sync Wi-Fi program.

The 2028 start date is not an accident — Amazon Leo’s constellation is still reaching operational density, and Delta is, in effect, betting that coverage will be sufficient by then. That is a meaningful commercial vote of confidence for a service that has not yet served a single commercial airline passenger.

Why would Delta pass on Starlink, which is operational today? Delta has not said explicitly. But the competitive logic is readable. A carrier that wants to avoid full dependence on a single vendor — SpaceX specifically — has structural reasons to back an alternative. Amazon’s existing commercial relationship with Delta across cloud infrastructure and logistics almost certainly shaped the decision too.

The Economics of the Sky

Aviation is an attractive segment for LEO operators because revenue density is high. A single widebody aircraft might carry 300 passengers, each effectively subsidizing connectivity through fares or loyalty programs, moving continuously through high-value routes. But the economics are more complicated than they first appear.

Starlink’s hardware and service costs have been rising. By mid-2026 the company had roughly doubled monthly service fees from their launch-era pricing, and installation costs for business jet antennas were reported at up to $200,000 — substantially higher than before. For major airlines negotiating fleet-wide volume deals, the per-aircraft economics are almost certainly more favorable, but the direction of travel on pricing is not flattering, and private jet operators have begun pushing back publicly.

On the market side, the in-flight LEO satellite Wi-Fi segment was estimated at around $2.16 billion in 2025, with projections toward $5.66 billion by 2030 — a compound annual growth rate above 20%, according to The Business Research Company. Even discounting for typical market-research optimism, the trajectory is unambiguous. Legacy GEO-based inflight connectivity is on a path to obsolescence in all but the most price-sensitive niches.

A Proving Ground Unlike Any Other

Aviation deals matter to LEO operators beyond their direct revenue. A fleet of hundreds of aircraft crossing the North Atlantic and transpacific routes is a demanding, high-profile stress test of coverage and reliability. When connectivity drops mid-ocean, an airline’s operations team knows within minutes. The aviation market is a proving ground that residential broadband is not — route coverage must be genuinely global, not merely good in dense suburban areas.

For Starlink, accumulating airline contracts reinforces a visible lead: every passenger who works a transatlantic flight on fast Wi-Fi is implicitly experiencing the product. For Amazon Leo, the Delta deal functions as an anchor — a credible commercial customer that signals the service is real and will be delivered, even if installations are still two years out.

Eutelsat OneWeb, for its part, has been largely absent from the headline airline deal news of 2025–2026. That silence is notable for a network that once positioned inflight connectivity as a core market. The aviation sector may be where the gap between first- and second-tier constellation scale becomes most legible.

What the Passenger Actually Gets

Strip away the commercial strategy, and what is changing is the experience of being a passenger. For the first time, the internet available at 35,000 feet is converging on something close to what you get on the ground. That is not a marketing promise — it follows directly from orbital geometry. LEO satellites at 500 kilometers altitude introduce roughly 20–40 milliseconds of round-trip latency. GEO satellites at 35,786 kilometers introduce 500–600. That difference, compounded across the length of a transatlantic crossing, is the difference between a usable workday and an exercise in frustration.

Airlines understand this. They are spending fleet budgets accordingly. The question that remains is whether the supply side — two serious LEO operators with different timelines, pricing philosophies, and constellation maturity — can execute cleanly enough to meet what they are promising. The contracts are signed. The satellites are launching. The passengers are waiting.

Sources