Snowflake Becomes USA Cycling's AI Data Partner for LA28 Olympics

From volunteer data science to a single platform

The partnership grows out of a problem USA Cycling's leadership describes openly. In Paris, the U.S. women's team pursuit squad won the country's first gold in the event, finishing six-tenths of a second ahead of New Zealand. Behind the result was Project 4:05, a two-year effort to get four riders around the track in under four minutes and five seconds by studying who should lead, for how long and in what order.

Brendan Quirk, CEO of USA Cycling, says the work relied on volunteer and deeply discounted data science help, with no unifying system to bring the data together. "That's not scalable," he says. According to Quirk, the organization needed a partner to take disparate data, "put it in one layer, put AI on top of it so we could work quickly."

Snowflake says the partnership covers road, mountain bike, track, BMX racing and BMX freestyle, as well as Paralympic road and track.

What USA Cycling plans to build

Today, according to the source, performance information sits across seven to eight separate tools. Alec Pasqualina, Director of Mountain Bike at USA Cycling, describes a "messy landscape of different things that we use all in different places." He says deeper analysis of sleep, nutrition and recovery is often the first thing squeezed out when schedules get busy.

USA Cycling plans to bring its performance data together on a single, secure foundation, aligned with the Athlete 360 initiative from the U.S. Olympic & Paralympic Committee (USOPC). Snowflake lists four uses:

• Trackside race analysis: In Team Pursuit and BMX Racing, athletes often have only minutes between rides. Race video will be paired with power, speed and timing data so coaches can show what to change, from pacing to equipment.
• Medal forecasting: The model first used for the Paris Games will expand to Olympic and Paralympic cycling and update as results come in, helping leaders direct funding, staff and support.
• One platform for performance decisions: Performance data, race results and training metrics will sit in one governed platform, and athletes, coaches and performance directors will be able to ask questions in natural language.
• BMX freestyle scoring: Using Snowflake Cortex AI, USA Cycling will link competition footage to judges' scores to show which tricks score most reliably and where a run leaves points on the table.

On natural language, Quirk says the staff are "Swiss Army knives" but "not programmers," so it is "an absolute need."

Athletes and the road to LA28

Riders quoted in the announcement hope for clearer answers from data they already collect. Riley Amos, who raced cross-country mountain bike for Team USA in Paris, says data can be "extremely powerful or honestly kind of overwhelming" and hopes the program helps "see through the clutter a little bit." In cross-country racing, the top five often finish within about a minute of each other after an hour and a half on course, according to the source.

USA Cycling now joins USA Bobsled/Skeleton in building on Snowflake. Quirk says he wants to share what the organization learns with the USOPC and other national governing bodies. He has set a goal of 10 Olympic medals and 12 Paralympic medals on home soil.

Snowflake describes itself as the Official Data Collaboration Provider for Team USA and the LA28 Games, and says it is committed to helping make them the most data-driven Games in history. The announcement was written by Denise Persson, Snowflake's Chief Marketing Officer, and details such as contract terms or timelines are not given.