Google and Cathay Pacific Expand AI Contrail Trials: 40% Estimated Warming Reduction, Predictive Flight Routing, and Aviation Climate Tech

Google and Cathay Pacific are expanding an operational trial that uses predictive AI to reduce the climate impact of aircraft contrails. The first phase covered more than 80 Cathay flights and produced an estimated 40% reduction in contrail warming impact, according to the companies' September 7, 2026 announcement.

The project began in late 2025 and moves Google's contrail forecasting work into a different operational environment: long-haul flights centered on Asia, including ultra-long-haul missions where route length, changing weather systems, fuel constraints and air-traffic coordination make altitude changes more complex than on a narrow test corridor.

The second phase is larger and is intended to extend the trials across Cathay's Asian and transpacific network. That changes the question from whether AI can identify contrail-forming regions to whether airlines can make avoidance a repeatable part of normal flight operations at network scale.

··········

⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣

⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣

CATHAY'S FIRST PHASE MOVED AI CONTRAIL AVOIDANCE INTO LIVE LONG-HAUL OPERATIONS.

The initial Cathay deployment tested whether predictive contrail forecasts could be translated into practical altitude decisions without changing the aircraft, engine or fuel.

Persistent contrails form only under specific atmospheric conditions. A flight can cross a cold, ice-supersaturated layer in which exhaust particles and water vapor produce ice-crystal clouds that remain behind the aircraft for minutes or hours. Because those clouds can trap outgoing infrared radiation, especially when solar reflection is weak or absent, their net climate effect can be strongly warming.

The operational strategy is therefore selective rather than universal. The system attempts to identify the comparatively small regions where a flight is likely to create a high-impact persistent contrail. Pilots can then consider a modest altitude change that moves the aircraft above or below that layer, subject to safety, air-traffic-control clearance, aircraft performance and fuel constraints.

Cathay's first phase used Google's AI-based prediction tools on more than 80 flights. The companies reported an estimated 40% reduction in the warming impact attributed to contrails during the trial. That figure is a climate-impact estimate rather than a direct claim that 40% fewer visible contrails formed, so it should not be treated as interchangeable with formation-rate metrics from earlier Google trials.

........

CATHAY / GOOGLE TRIAL ITEM	REPORTED VALUE	OPERATIONAL MEANING
Project start	Late 2025	Moves predictive contrail avoidance from research toward routine airline operations.
First-phase scale	More than 80 Cathay flights	Provides live-flight evidence across Cathay operations rather than a simulation-only result.
Estimated warming reduction	~40%	Measures modeled climate impact reduction, not simply the number of visible contrails.
Technology approach	AI-based contrail forecasts	Identifies atmospheric zones where persistent warming contrails are more likely to form.
Pilot action	Altitude adjustment when operationally feasible	Avoidance is implemented through flight-level changes rather than new aircraft hardware.
Regional significance	First Google commercial airline partner in Asia-Pacific for this technology	Tests deployment in a dense and operationally varied aviation region.
Ultra-long-haul significance	First airline reported to test contrail avoidance on ultra-long-haul flights	Adds fuel, weather, reserve and airspace constraints to the optimization problem.
Second phase	Asian and transpacific network expansion	Tests whether the approach can scale beyond a limited first-phase sample.

........

Cathay's role is notable because its network includes long sectors across multiple weather regimes and air-traffic-control regions. A recommendation that is operationally easy on a short flight can become more complicated on a transpacific mission, where an altitude change can affect step-climb planning, fuel reserves, turbulence exposure and downstream routing.

The trial therefore tests more than prediction accuracy. It tests whether forecast information arrives early enough, is precise enough to influence a flight plan, can be communicated in a form pilots and dispatchers can use, and can be executed without creating a larger fuel or safety penalty elsewhere in the flight.

··········

⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁤

⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁤

THE AI SYSTEM PREDICTS WHERE SMALL ALTITUDE CHANGES CAN AVOID HIGH-WARMING CONTRAILS.

Google's broader contrail research combines atmospheric forecasting with satellite and flight data to estimate where persistent contrails are likely and to verify what happened after aircraft passed through those regions.

This is a different optimization problem from conventional route planning. Standard flight planning already balances winds, fuel, restricted airspace, weather, aircraft performance and air-traffic constraints. Contrail avoidance adds another variable: the expected non-CO2 climate effect of flying through a specific atmospheric layer at a specific time.

The AI layer is useful because the relevant regions are dynamic. Ice-supersaturated layers can be thin, spatially irregular and short-lived. A blanket instruction to fly lower or higher would waste fuel and may simply move the aircraft into another problematic layer. A predictive system instead tries to target the limited segments where an altitude change has a high expected climate benefit.

Google's earlier work with American Airlines illustrates the data loop. Forecasts were used to identify likely contrail regions; pilots adjusted altitude on selected flights; satellite imagery was then analyzed to determine whether contrails actually formed. That feedback is essential because the forecast is probabilistic and the atmosphere cannot be measured perfectly at every point along a route.

The climate objective is also more nuanced than reducing every contrail. Daytime contrails can reflect incoming sunlight while also trapping outgoing heat, whereas nighttime contrails do not provide the same solar-reflection effect and can be more strongly warming. Persistence, location, time of day, optical thickness and background cloud conditions all influence net radiative forcing.

This is why warming-weighted optimization can be more valuable than a simple visual count. Avoiding one long-lived, strongly warming contrail can matter more than preventing several short-lived trails with limited net forcing. The 40% Cathay result is therefore best interpreted as an estimate of avoided climate impact under the methodology used in the trial, not as a universal 40% reduction that every airline could reproduce.

The system also does not replace pilot judgment or air-traffic management. Forecasts become another planning input. Flight crews and dispatchers still operate within safety rules, aircraft limitations, traffic separation requirements and weather constraints, and an avoidance maneuver may be rejected when its operational cost or risk is too high.

··········

FUEL, VERIFICATION, AND METRIC DESIGN DETERMINE WHETHER THE CLIMATE BENEFIT SCALES.

The strongest evidence for contrail avoidance now comes from several Google-linked trials, but their headline percentages measure different endpoints and should be compared carefully.

In Google's 2023 trial with American Airlines, 70 test flights using AI-based predictions reduced observed contrail formation by 54%. The flights that attempted avoidance burned about 2% more fuel. Google noted that only a small fraction of flights may need to be changed to target a large share of contrail warming, estimating that the total fuel impact could be around 0.3% across an airline's operations under the assumptions used in that work.

A larger 2026 study integrated Google's forecasts directly into flight-planning software for 2,400 scheduled transatlantic flights. For the flights that successfully flew the contrail-avoidance plans, Google reported a 62% reduction in contrail formation rate relative to the control group. That study was important because it reduced the manual coordination required in the earlier trial and moved the technique closer to an automated planning workflow.

Cathay's reported 40% figure uses a different denominator: estimated warming impact. That can be a more climate-relevant metric, but it means the result cannot be ranked directly against the 54% or 62% formation-rate reductions without the underlying methodology, atmospheric conditions and selection criteria.

........

PROGRAM / EVIDENCE	REPORTED RESULT	WHAT THE NUMBER ACTUALLY MEASURES
Google + American Airlines, 2023	54% reduction across 70 test flights	Observed contrail formation reduction in a small operational trial.
Fuel effect in 2023 trial	~2% extra fuel on flights attempting avoidance	Local cost of altitude changes on the flights that were modified.
Estimated airline-wide fuel effect	Potentially ~0.3%	Google estimate based on targeting a limited share of flights rather than rerouting every flight.
Google + American integrated study, 2026	62% lower contrail formation rate	Result for flights that successfully flew the avoidance plans in a 2,400-flight transatlantic trial.
Cathay + Google first phase, 2026	~40% estimated reduction in contrail warming impact	Climate-effect estimate, not directly comparable with formation-rate percentages.
Operation Blue Skies, 2026	North Atlantic airspace-scale trial	Tests coordination with government, aviation and flight-planning partners at broader system level.
Cathay second phase	Results not yet reported	Expansion across Asian and transpacific operations will test scalability and operational consistency.

........

Fuel is the central trade-off because avoiding a contrail may require flying temporarily away from the most fuel-efficient altitude. Burning additional fuel increases CO2 emissions, so the climate case depends on the avoided non-CO2 warming being larger than the incremental CO2 and other operational costs.

Verification is equally important. A forecast can recommend an altitude change, but airlines and regulators need evidence that the avoided region would actually have produced a persistent warming contrail. Satellite-based detection, weather reanalysis, flight-track data and counterfactual modeling are therefore part of the measurement problem, not peripheral research tools.

Scaling also requires standardized accounting. Airlines will need common definitions for avoided contrail formation, avoided energy forcing and CO2-equivalent climate benefit if the technology is ever incorporated into sustainability reporting, operational incentives or air-traffic management systems. Without comparable metrics, a high headline percentage can describe a very different outcome from another trial's high percentage.

··········

THE SECOND PHASE TESTS WHETHER CONTRAIL AVOIDANCE CAN BECOME ROUTINE FLIGHT INFRASTRUCTURE.

Cathay's network expansion will show whether the technology can survive the variability of real airline operations: different aircraft, seasons, routes, airspaces, traffic levels, weather systems and dispatch constraints.

A scalable system would ideally make contrail forecasts available inside the same planning environment used for winds, fuel and weather, rank candidate altitude changes by expected climate benefit, calculate the fuel and time penalty, and present only the interventions with a favorable trade-off. That reduces workload and makes it possible to target the relatively small number of flight segments responsible for a disproportionate share of persistent warming contrails.

Ultra-long-haul flying is a demanding test case because cruise altitude changes are already tightly connected to aircraft weight. A heavily fueled aircraft may not be able to climb early in the flight, while later step climbs are used to regain efficiency as fuel burns off. Contrail avoidance has to fit around those performance constraints rather than assume that every recommended flight level is available.

Air-traffic coordination is another limit. One airline can optimize its own flight plan, but busy routes and oceanic tracks are shared systems. If many carriers begin requesting climate-motivated altitude changes, navigation service providers will need rules and tools to decide when the requests can be accommodated without reducing capacity or safety margins.

Google's August 2026 Operation Blue Skies initiative with the UK government and aviation partners points toward that next layer: contrail avoidance at airspace scale rather than one airline at a time. Cathay's Asia-Pacific expansion approaches the same problem from the carrier side, testing whether the prediction-and-avoidance loop can be embedded into a geographically broad commercial network.

Contrail avoidance is not a substitute for lower-carbon fuels, more efficient aircraft or long-term propulsion changes. Its attraction is that it can potentially reduce a substantial non-CO2 warming effect with today's aircraft and today's fuel, using software, forecasting and selective changes to flight operations.

The second Cathay phase will therefore be most informative if it reports not only another headline reduction percentage but also the share of flights receiving recommendations, the share of recommendations actually flown, additional fuel burn, route and altitude changes, verification confidence and the distribution of climate benefit across flights. Those variables determine whether predictive contrail avoidance becomes a narrow research tool or a durable layer of aviation infrastructure.

·····

FOLLOW US FOR MORE.

·····

·····

DATA STUDIOS

·····

[datastudios.org]