![]()
On July 24, 2025, Starlink went dark for roughly 150 minutes; the flight time from Teterboro to Palm Beach. Reuters reported the outage hit tens of thousands of users, with more than 61,000 incident notices logged on Downdetector. SpaceX called it an internal software failure in key network systems. The most modern satellite network on Earth got taken offline by the same thing that ruins your office printer.
Somewhere over the Atlantic, a Gulfstream cabin went quiet. A Citation principal over Kansas lost a Zoom call. And in a thousand flight departments at once, somebody remembered why their predecessors never retired the L-band.
Six months later, SpaceX filed its S-1, and a remarkable number of people I respect have decided to read that document as a green light.
Static Fire
Let me be fair to Elon before I become unfair.
Starlink earned the attention. Cabin connectivity used to be one of those things everyone hated but accepted, like crew catering invoices or the printer in the FBO business center. The user experience was geostationary: slow, expensive, lagged, complete with the standard cabin apology that it should get better after we cross the coast. Low Earth orbit cut latency to about 25 milliseconds against geostationary's 600. Independent in-flight testing puts the single-user median throughput around 64 Mbps down. Aviation went from apologizing about the Wi-Fi to handing the principal a working office at FL450. The cabin became the conference room. Most people in this industry spent their careers waiting for that.
The disruption is real. What worries me is the conclusion aviation buyers have drawn from it.
General Corporate Purposes
The S-1's language about how SpaceX will spend the money it is about to raise: "general corporate purposes."
Three words. They mean: we will spend it wherever we feel like. Reuters reported SpaceX generated $3.26 billion in Q1 2026 revenue against a $4.28 billion loss and a $41.31 billion accumulated deficit. The AI division alone burned $2.47 billion last quarter. Capex came in at $7.72 billion. Starlink throws off the profit, and the rest of the company spends it faster than it arrives.
A flight department reading this prospectus as a green light for cheaper service or faster certification on its tail number is reading it wrong. The document does not promise you a dollar. It barely acknowledges you exist. When a dealer tells the principal the IPO will mean better support and quicker STC turnaround on the King Air, ask which page. There is no page.
The Starship Asterisk
The most important connectivity risk in business aviation is sitting on a launchpad in Texas.
SpaceX's future capacity, including the capacity aviation is buying into, depends on Starship lowering launch costs at scale. Reuters has reported the IPO case is tied directly to Starship execution. The latest test strengthened the case. It did not erase the asterisk. Capital does not bend orbital mechanics. The FCC's January 2026 approval of 7,500 additional Gen2 satellites runs deployment milestones to 2031, and no IPO check accelerates that schedule.
Whatever Starlink looks like at demo today is not what it will look like at FL450 in three to five years. The version aviation is signing up for shares a constellation with American Airlines (500-plus narrowbodies starting Q1 2027, per Reuters), United, Southwest, Alaska, Singapore, Emirates, and the 500 Delta aircraft now contracted to Amazon's Project Kuiper for 2028. Add the maritime, defense, rural, RV, and direct-to-cell businesses Starlink is also pushing into.
Demand + shared spectrum + moving aircraft = a capacity problem eventually.
Your principal's beam over the North Atlantic in three years is going to be a lot more crowded than the dealer's PowerPoint suggests.
Case in Point
Blue Origin may have just provided the clearest reminder of the risk hiding underneath every satellite broadband story. On May 28, 2026, its New Glenn rocket exploded during an engine test ahead of a planned launch carrying Amazon's Kuiper satellites, following an earlier mission setback in April. A failed launch will most likely delay an entire constellation rollout. Project Kuiper only becomes a Starlink competitor when satellites reach orbit at scale.
If launch failures push deployment back by a year, the effects ripple through everything downstream. The same risk exists for Starlink. Airlines can debate terminals and pricing, but none of it matters if the launch cadence needed to grow and maintain the network breaks down. Capital can solve many problems but orbital mechanics still gets a vote.
Beam Sharing
Nobody in this industry will sell you a service-level agreement on a Starlink retrofit, because there isn't one to sell.
There is no public aircraft-specific SLA on any route. The published performance numbers — 25 ms median latency, 200 Mbps peak — are network-level averages. They include the dish on a roof in Akron and the dish on a sailboat off Sardinia. A marketing average is not a contract.
Independent in-flight testing found single-user medians of 64 Mbps down and 24 Mbps up, varying by route, altitude, ground-station path, and inter-satellite routing. Those are real numbers. They are also weather, not climate. The flight department signing the STC needs to be clear-eyed about which one the dealer is actually selling.
Mission Critical
Starlink for aviation is cabin connectivity. It is not certified safety comms. The dealer will not say that clearly because the principal does not want to hear it.
Viasat sells two distinct product lines for a reason — Business Aviation on one side, Flight Operations and Safety on the other (SwiftBroadband-Safety, Iris, Classic Aero). Starlink lives entirely on the cabin side of that line. The July outage proved the point inside two and a half hours: when the shiny pipe drops, the ugly pipe earns its keep.
While bandwidth was getting the headlines, cybersecurity moved from an IT problem to an airworthiness problem. The FAA's 2024 proposed rulemaking would standardize cyber criteria inside aircraft type certification. EASA's AMC 20-42 already exists as the formal risk framework. Segmentation between the cabin Wi-Fi and the avionics bus, authentication, logging, update governance — these are airworthiness words now.
A 200-megabit pipe on a sloppy network architecture is an open door with a radome attached. Bandwidth without architecture is just speed wearing a fake mustache.
Burn Rate
The $2,000-a-month Aviation Jet 20GB tier reads beautifully on a CFO spreadsheet right up until the cabin behaves like a cabin: streaming, video calls, cloud sync, phones updating, kids watching sports over Greenland, three laptops pretending not to be syncing Dropbox, and one passenger who somehow downloads an entire operating system over the Atlantic. The overage meter starts at $100 per gigabyte. Spreadsheets bleed.
The monthly fee is the wrong number to anchor on. The right number is how much the cabin consumes in an average month, by tail and by route, which almost no flight department has ever measured. They have anecdotes, complaints, and invoice arguments instead. A CFO deciding connectivity spend without usage data is playing darts in a dark hangar.
The Other Constellation
Starlink did not kill the incumbents. It made them serious.
Honeywell is dangling promotions worth up to $300,000 to keep customers on JetWave. That is what defensive pricing looks like before an IPO road show has even started. Viasat just announced a fivefold capacity boost over the US for business aviation customers and is pushing into the Middle East, South America, and Hawaii. Gogo, which markets its 5G product as "satellite speeds at ATG prices," is genuinely cheaper than Starlink Unlimited at $5,500 a month if your operation never leaves North America. Gogo's Satcom Direct acquisition turned a domestic ATG company into a multi-orbit competitor in eighteen months. Galileo via OneWeb is the cleanest non-Starlink LEO option in the market. Amazon Kuiper just took Delta.
This is not going to be winner-take-all. The realistic outcome is multi-orbit, multi-bearer, and mission-specific — which is how aviation has always actually worked, even when the press wants a king.
Second Stage
Business aviation gets the headlines because the checks are bigger. General aviation may be the louder long-term story.
Starlink split its consumer aviation offering into Local and Global tiers, both bucketed by aircraft speed class. The owner-flown SR22 heading to a fishing camp in Labrador now has a coverage story that did not exist two years ago. The bizjet retrofit market has a natural ceiling somewhere in the low tens of thousands of aircraft. The US general aviation fleet is over 214,000. The economics are messier, the STC landscape is more fragmented, and a Cirrus owner is not going to absorb a $400,000 install the way a Global operator might.
If Starlink figures out affordable hardware, clean certification paths, and rational service tiers for light aircraft, the business jet market ends up being the appetizer.
Splashdown
I want to be clear before this lands.
I like Starlink. That is what makes the whole thing a little bit frustrating. I like that it embarrassed a lazy connectivity market. I like that it made airborne Wi-Fi feel normal. I like that principals, crews, and passengers can stop pretending 2009 internet is acceptable just because it happens to be moving at Mach .82. I am, for the record, not your dealer; I am a pilot who reads prospectuses on weekends, which is a personality flaw I am still working through. But I have friends in flight departments, and I do not want any of them to be the case study at the next NBAA panel.
So lean in where it fits. Lean in if you fly internationally and you have spent a decade saying yeah, sorry, the Wi-Fi doesn't really work over the water to a principal who is fifty-five and has stopped finding it charming. Lean in if the STC is sitting on a desk at Gulfstream Savannah or AeroMech or Textron. Bundle the retrofit into a heavy check or a cabin refurb that is already eating the calendar.
Then hold the line where the prospectus runs out. Build the cyber architecture. Read the SLA you do not actually have. Do not turn your aircraft into a beta test for your own optimism.
Elon filed a 277-page document about a constellation. It does not contain a procurement strategy for your airplane, an SLA for your routes, or a runbook for the next outage. The $400,000 retrofit is still yours. Sign it carefully.