Drew Cukor and Andretti Global on using Palantir to unify IndyCar race data and compete on milliseconds

Summary

Andretti Global's IndyCar program is deploying Palantir's AIP platform to solve a core operational problem: race-critical data scattered across incompatible systems. Telemetry, car setup databases, and IndyCar-controlled timing feeds all live in separate silos, forcing engineers to manually reconcile information during practice windows that close regardless of readiness. The goal is a unified data environment where faster synthesis of inputs translates directly into faster on-track iteration.

Zach Porter, senior simulation engineer on the IndyCar program, and Kyle Kirkwood, driver of the No. 27 Honda for Andretti Global, frame the stakes plainly. Competition now happens in the margins, hundredths and thousandths of seconds where the fan-car era rule-book exploits are no longer available. The regulatory environment in modern single-seater racing is tight enough that data infrastructure has become one of the few remaining levers teams can actually pull.

The earliest concrete win from the Palantir integration is automated sensor anomaly detection. IndyCar machinery operates under extreme mechanical stress, and sensor failures during a session can corrupt the entire data picture. Machine learning models now flag anomalies automatically, freeing systems engineers to diagnose and replace faulty components before the next on-track outing rather than discovering the problem after the fact.

Drew Cukor, representing TWG, the holding company whose portfolio spans insurance, asset management, investment banking, and motorsports, positions the Palantir build-out as enterprise-wide, not just a car engineering project. Andretti's HR, technology, and engineering functions are all candidates for integration. Cukor draws on 30 years in the Marines to frame the transformation as a shift away from decision-making on incomplete, fragmented systems, and acknowledges the hardest cost is organizational, not financial. Resistance from personnel comfortable with existing workflows is the primary friction.

The competitive pressure argument is explicit. If Andretti does not adopt these tools, competing teams will. The cost of inaction is measured in championship points, not just software licensing.

The team's target deployment milestone is St. Petersburg 2026, the traditional IndyCar season opener. The off-season, which began roughly three days before the AIPCon appearance, is being used to productionize use cases that were trialed during the 2025 season. Physical testing on track is sharply limited by cost, which concentrates development work in simulation, driver-in-the-loop rigs, wind tunnel time, and shaker rigs, making the quality and connectivity of the underlying data infrastructure more consequential, not less.

Kirkwood notes that IndyCar runs without power steering, generates over 5,000 to 6,000 pounds of downforce, and subjects drivers to four to five Gs sustained for up to two hours per race, physical demands that cannot be replicated in simulation. That gap between sim and reality makes clean, reliable telemetry from actual track sessions the irreplaceable input into any performance model.