Google is widening an AI-guided contrail-avoidance program with Cathay Pacific, the Hong Kong-based carrier, testing whether ultra-long-haul flights across Asia-Pacific can be steered around the narrow bands of cold, humid air that produce the aviation industry’s most damaging contrails. It is Google’s first partnership of this kind with a commercial carrier in Asia, following earlier pilots the company ran over the United States and the North Atlantic. The premise sounds narrow, but the stakes are not: a minority of flights create a majority of aviation’s warming effect, and finding them ahead of takeoff is a forecasting problem, not an engine redesign.
Contrails form when hot, humid exhaust from an aircraft’s engines meets sufficiently cold air at cruising altitude. Most evaporate within seconds or minutes. Under specific conditions, the ice crystals instead persist and spread into thin cloud layers that trap outgoing heat much like natural cirrus does. Climate researchers have linked roughly a third of aviation’s overall warming impact to this smaller category of long-lived contrails, which is why steering planes around them, separate from cutting fuel burn, has become its own research target.
The system Google built pairs AI forecasting models with satellite imagery and weather data to flag, hours ahead of departure, the pockets of sky where a flight would otherwise leave a persistent contrail. Dispatchers get that information before the aircraft leaves the gate. Cathay Pacific then routes the live forecast into the cockpit through its in-flight Wi-Fi and Electronic Flight Folder system, letting pilots make a small, planned altitude change of the sort crews already use to dodge turbulence, inside existing safety limits.
Google set an initial target of over 100 flights for this phase. More than 80 of them actually flew the avoidance routing. By Google’s own account, satellite-imagery analysis of those flights found roughly a 40 percent reduction in the warming effect of their contrails. That figure is Google’s internal estimate from its own pilot, not an outside audit, and it describes results from one partial phase on one airline’s network rather than a permanent change to how Cathay Pacific flies. Google also says a single route, between Hong Kong and Singapore, produced over half of the entire trial’s estimated benefit. That is evidence a handful of contrail-prone corridors can carry most of the upside, or, on a different network, most of the disappointment.
One tradeoff goes unaddressed in Google’s announcement: whether flying around contrail-forming air burns meaningfully more fuel, which would eat into the net climate gain. The post is silent on that accounting. Until Google or an independent group publishes it, the true benefit of the program stays unverified.
What does distinguish this effort from most AI climate pitches is that its central claim is a short-range weather prediction, not an optimization promise. A forecast of where contrails will persist is something outside researchers can eventually check against real flight and satellite data, which makes it an unusually testable use of AI compared with claims that hinge on a model quietly getting smarter at resource allocation.
Google and Cathay Pacific are now moving to a larger second phase and bringing in Contrails.org, a nonprofit focused on contrail science, as a research partner. Airlines watching this trial should press for two things before treating the Asia-Pacific numbers as a template: a public fuel-burn accounting, and evidence that the Hong Kong-Singapore corridor’s outsized share of the benefit holds on route networks with less contrail-prone airspace.
Google detailed the trial’s design and its own effectiveness estimates in a post on its research blog on September 7, 2026.
