Picture a routine flight. An electric air taxi lifts off from a pad near JFK and turns for Manhattan, the kind of short urban hop the whole industry is built around. Somewhere over Jamaica Bay, it loses power. Not one motor. All of it. Where can it land?
For a helicopter, the honest answer on that route is not great but not hopeless: gliding down in autorotation, it can reach a safe landing site over roughly 9 percent of the corridor. For an eVTOL with wings, the answer is worse, roughly 5 percent. For a multirotor, the kind with many small rotors and no wing, it is close to nothing.
And here is the part that makes this an AirIndex finding rather than an engineering footnote. Even most of those few reachable pads cannot actually be used by an eVTOL as they exist today. Reachable is not usable. Put the two together and the usable emergency landing coverage for an eVTOL, on a real corridor, in the worst case, barely exists.
This piece measures both halves. The numbers are illustrative, built from public method and representative aircraft archetypes rather than any manufacturer's certified figures, and we say so plainly where it matters. The point is not a verdict on any aircraft. It is the shape of a gap.
Why eVTOL emergency landing is a different problem
The framing here owes a great deal to Rex Alexander of Five-Alpha, who has spent a career on vertical flight infrastructure. His point is that vertical flight has always been survivable because of three layers, and each of them breaks differently for eVTOL.
The first layer is alternates. Airline flying leans on more than 5,000 public use airports, a dense net of places to divert. Vertiports will be far sparser, especially in the first years, which means the net under an air taxi is mostly not there yet.
The second layer is the glide. A fixed wing aircraft glides. A helicopter autorotates, spinning its rotor on the way down to cushion the landing. Most multirotor eVTOLs can do neither, because their rotors are too small and too numerous to autorotate. Winged eVTOLs can glide in forward flight, but not in a hover, and a hover is exactly what takeoff and landing are. The FAA's own criteria ask for gliding or autorotation or an equivalent means, and that phrase, equivalent means, is not yet defined.
The third layer is warning. A fuel aircraft runs dry gradually. The engine sputters, gauges fall, there is a window to react. A battery can fail abruptly: a voltage cliff or a thermal event, with no sputter and no window. The FAA has a name for this, the fade out problem.
There is a fourth point underneath all three, and it is the one that should keep planners up at night. Today the ultimate fallback is a pilot. Federal rules let a pilot in command do whatever an emergency requires. An autonomous aircraft cannot improvise a forced landing. It plans against data. As Rex puts it, autonomous eVTOL is the unwritten chapter. The map it would plan against does not exist in one place yet.
Finding one: reach, an eVTOL can't glide out of a landing desert
Reach is not one glide number. It depends on the energy class of the aircraft, on the failure mode, and on altitude, so we modeled it that way rather than drawing a single circle. A fuel rotorcraft autorotates. An eVTOL losing charge gradually keeps the range its remaining state of charge allows, which shrinks as the route goes on, making this a long route problem more than a short hop one. An eVTOL in a total power loss descends nearly vertically and reaches almost nothing. An eVTOL in a battery thermal event has minutes of controllable flight and must land immediately.
Run the worst of these, total power loss, along the JFK to Manhattan corridor at 1500 feet, and the reach footprint sweeps the route looking for somewhere to come down. Over the bay and much of Brooklyn it finds nothing. Only on short final, close to the destination, does a registered pad come into range. There is a whole band of the route where a helicopter would still make a safe site and an eVTOL is already too far. That is the roughly 9 percent versus roughly 5 percent, and for a multirotor, close to zero.
One honest qualification. This is the total power loss case, the hardest one. A well designed eVTOL is built so that losing a single motor is survivable, and that is a real and important safety margin. The case modeled here is losing propulsion entirely, the case where the old escape hatch, autorotation, is gone.
Finding two: usability, reachable is not usable
Now the AirIndex specific half. Suppose a pad is in reach. Can the aircraft actually use it? Mostly not, today. An eVTOL needs a larger footprint than a helicopter pad provides. Following the FAA's vertiport design pattern, the final approach and takeoff area scales with the aircraft's controlling dimension, roughly twice it, which is a bigger square than most existing helipads were ever built to.
We measured this against real pads. Of 5,594 U.S. helipads with measured dimensions, set against a typical eVTOL with a 50 foot controlling dimension and therefore a required 100 foot landing area, only about 30 percent are large enough as they exist today. Roughly 70 percent are not. Drop to a smaller eVTOL, a 40 foot dimension and an 80 foot area, and about 56 percent qualify, which still leaves around 44 percent short. The median pad's limiting side is 48 feet. The footprint the next generation needs is simply larger than the one the last generation built.
And that is footprint alone, the most generous test. For a rooftop hospital pad the binding constraint is usually load bearing: the structure has to carry the weight across the whole landing area, and the FAA does not track that, so it cannot be read from any record. Add obstruction clearance around the approach. Layer those on and the truly usable share falls below the footprint number, not above it.
This is what we mean by approved is not operable. A route can be approved, an aircraft certified, a vertiport built, and the system still not be operable, because the places it would land in an emergency cannot actually take the aircraft.
The two findings compound
These are not independent worries to weigh separately. They multiply. Reach puts roughly 5 percent of the JFK route within gliding distance of a pad in a total power loss. Usability says only about 30 percent of pads can take the aircraft at all, and fewer once load bearing and obstruction are counted. A small fraction of a small fraction is the usable emergency landing coverage for an eVTOL on that corridor today. It barely exists.
The reason this matters is not that eVTOLs are unsafe. It is that the safety case for the helicopter era rested on a pilot looking out a window and a dense net of places to put down, and the next era inherits neither by default. The coverage has to be built, deliberately, and right now no one is measuring whether it is there.
So what's the plan
The useful response is not alarm. It is a map. Map which existing pads can support an eVTOL emergency landing, which ones need upgrading and by how much, and where the deserts are, before a corridor is declared operable rather than after.
That map is what the AirIndex substrate already produces. The corrected landing record gives the real inventory, including the pads the federal registry misses. The emergency reach model turns it into corridor coverage, with the landing deserts and diversion nodes called out. Per facility risk and dimension data say which pads can take the aircraft and which cannot. And where a specific gap needs closing, the remediation path shows what it would take to upgrade a pad into a usable diversion site.
For an autonomous aircraft this stops being a planning nicety and becomes the safety system itself. A machine cannot invent a forced landing site in the moment. It descends to one that was computed in advance and trusted. Building and verifying that map is the work.
The federal record is being called broken, on the record
This is not only our view. On June 11, 2026, a federal advisory committee established under the No Surprises Act and led by CMS, with the Department of Transportation, delivered a report to Congress that recommended funding to modernize the nation's helipad data, and a low altitude framework that explicitly names uncrewed aircraft and advanced air mobility congestion. The committee, working an entirely separate problem, arrived at the same conclusion: the landing record is not good enough for what is coming.
One distinction worth keeping clean. That report cites Five-Alpha's finding that on the order of 1,600 to 1,800 heliports are missing from the Airport Master Record, a count of all heliports. The findings in this piece are a different and narrower measurement, the eVTOL specific edge of the same gap: how far an eVTOL can reach and which pads it can use. The two are complementary, not the same number.
Method and limits
Reach is modeled by elip-coverage-1.3 with elip-energy-1.0, which treats emergency reach as a function of energy class, failure mode, and altitude rather than a fixed glide ratio. The eVTOL configurations are illustrative archetypes, a vectored thrust type, a lift and cruise type, and a multirotor, and their parameters are illustrative, not drawn from any manufacturer's certified performance. Usability is measured by conversion-feasibility-1.1, applying the FAA vertiport footprint ratio against existing final approach and takeoff dimensions for the 5,594 pads we have measured.
The honest limits. Per airframe emergency procedures are still emerging and manufacturer specific, so the reach figures describe archetypes, not real aircraft. Load bearing capacity cannot be verified from public records, so the usability number is an upper bound on what is truly usable, not a lower one. And the total power loss case modeled here is the hardest case by design, not the expected one. None of that changes the shape of the finding. On a real corridor, today, an eVTOL that loses power has very few places it can both reach and use, and that gap is measurable before anyone has to discover it the hard way.
The corridor coverage, per facility dimension data, and the diversion analysis behind this piece are available to institutional readers. If you are planning routes, certifying vertiports, underwriting the risk, or building the aircraft, and you want to know where the usable landing coverage actually is on a corridor you care about, ask us.