Amazon Prime Video’s AI NASCAR Graphic Explained

Amazon Prime Video’s newest NASCAR stat incorporates tens of thousands of incoming data points in real time, using machine learning over the course of a race for more accurate outputs. But when “The Burn Bar” pops up on screen, senior coordinating producer Alex Strand doesn’t want viewers to spend time thinking about the math involved. Instead, the miles-per-gallon readout, along with a color-coded guide, is meant to instantly convey the story of a driver conserving fuel or maxing out. 

“That’s ultimately our North Star at Amazon,” Strand said. “How do we bring more data to viewers, but do it in a way that they can digest?”

The graphic package was 18 months in the making, stretching back to Strand’s first meetings with analysts Dale Earnhardt Jr. and Steve Letarte, as they watched a race and discussed what Amazon could add to the picture when it began streaming races this year. 

“We actually think fuel is a really, really cool part of racing,” Strand said. “It’s just no one’s really figured out how to tell that story yet.”

Letarte added that fuel strategy would be a particularly relevant talking point during Amazon’s stretch of races this spring, including Sunday’s showdown at Michigan International Speedway. 

Back in 2008, for instance, Earnhardt Jr. cut off his engine during a late caution at MIS, coasting to save every ounce of gas and outlasting his competition for the checkered flag. 

It’s exactly the kind of maneuver—nearly invisible to the untrained eye—that Prime Video’s booth hopes to spotlight, similar to the company’s work on pre-snap cat-and-mouse games during Thursday Night Football broadcasts.    

“This is awful because I’m an announcer, but I think the more a sport can be displayed, both visually by the director and graphically by the great graphics director and producer—that’s my goal,” Letarte said.

Amazon says it has developed a proprietary artificial intelligence model to track fuel mileage for every car on a lap-by-lap basis, primarily based on throttle and RPM measurements, while race teams are likely only viewing similar data for select vehicles. Amazon data scientists, computer vision experts, engineers and producers—including Prime Video producer for analytics and insights Sam Schwartzstein, a former football player who helps lead Thursday Night Football’s Prime Vision with Next Gen Stats efforts—contributed to the development process, which also relies on Amazon Web Services’ cloud computing power. 

So far, the Burn Bar has largely been broken out for specific drivers in key moments, though it could also be used to compare a car’s race to previous events, or even more significantly, to the rest of the field.

“That’s really where the story unfolds as the race unfolds,” Letarte said.

Viewers are now getting to see it for themselves.