\n<\/p>\n
Formula 1 data use and consumption has exploded in the past two years. (Photo by Mario Renzi – Formula 1\/Formula 1 via Getty Images)<\/span><\/p>\n Formula 1 via Getty Images<\/small><\/div>\n<\/div>\n<\/figure>\n Imagine streaming 130 million songs in a single weekend. Or binge-watching 162,500 hours of 4K Netflix. Or reading the entire Library of Congress, 32 times over, in just three days. That\u2019s basically the equivalent of how much data Formula 1 moves every single race weekend.<\/p>\n Two years ago, Formula 1 streamed roughly 500 terabytes of data between its trackside technical center and its UK broadcast hub on race weekends. Today that number is up to 650 terabytes: a 30% jump in just 24 months, and a signal that maybe the world\u2019s most data-intensive sport is racing toward a future where artificial intelligence, edge computing, and real-time telemetry are essentially omnipresent. <\/p>\n I recently had the opportunity to chat with Chris Roberts, who leads F1\u2019s IT team, and Milo Speranzo of Lenovo, F1\u2019s global technology partner, to talk about where all those extra bytes are coming from, and where they’re going.<\/p>\n \u201cWe are not nice people,\u201d Roberts said jokingly when I asked him about the challenge of putting together the hardware and software that helps F1 manage and broadcast races. Speranzo laughed and disagreed.<\/p>\n But it\u2019s a formidable data challenge.<\/p>\n The race cars that rocket around a Formula 1 track at over 200 miles per hour might look like the world\u2019s fastest racing machines on wheels, and they are. But under the surface, they\u2019re really data-spawning monsters. Each of the 22 cars on a 2026 Formula 1 grid carries 300 sensors that generate more than 1 million data points every single second. Add it up over a full race weekend, and F1\u2019s race cars will create roughly eight terabytes of telemetry alone, giving Mercedes, Red Bull, McLaren and every other team massive amounts of data on engine performance, suspension behavior, gearbox efficiency, fuel flow, G-forces, driver inputs and more.<\/p>\n And car telemetry is only one source of data at a Formula 1 race.<\/p>\n F1 also deploys 28 ultra-high-definition track cameras, between four to six cameras embedded in kerbs, bridges and barriers, plus a gyro-stabilized helicopter cam, mobile cameras throughout the pit lane, over 100 on-car cameras across the field, plus drone cameras. A non-live 360-degree camera on every car captures another 480 GB of footage per race and up to 150 microphones are deployed around each circuit to capture the sounds as well as the images of the race.<\/p>\n All of that data funnels through a roving technical center that gets disassembled after every race and shipped to the next location. It\u2019s “the biggest, most complex transportable facility of its type in the world,” Roberts says, housing 750 pieces of equipment running 40 bespoke software systems running on Lenovo servers that provide 1.4 terahertz of CPU across 512 cores, 8.2 TB of RAM, and 100 TB of all-flash storage. <\/p>\n It might look like data growth in F1 is purely a function of more sensors and higher-resolution cameras. That\u2019s a big part of the story, sure, but another major driver is who\u2019s watching today.<\/p>\n Formula 1\u2019s global fanbase now stands at 827 million, Roberts says. That\u2019s a 12% year-over-year increase and a 63% jump compared to 2018. More telling is the demographic shift: 43% of fans are now under 35 and 42% are female. F1 now says it has 300 million fans in China and India: relatively new markets for motorsport. <\/p>\n Those new and younger fans don\u2019t want just a static broadcast feed, Roberts says. They want stats, multi-camera angles, driver tracker overlays, strategy breakdowns, social clips, behind-the-scenes content, on-board camera views, and instantly-available TikTok videos of Verstappen\u2019s latest overtake. In 2025, F1 produced over 10,000 videos for its social channels that racked up more than 18 billion video views, and that\u2019s going up in 2026.<\/p>\n “There\u2019s casual fans all the way through to some real avid fans, and we\u2019ve got to talk to all of them,” Roberts says. <\/p>\n That requirement to feed all of that fan base with what they want is what pushed F1 to overhaul its telemetry pipeline at the start of this season, processing it on-site rather than at its UK broadcast center, resulting in a .3 second reduction in latency.<\/p>\n Seems small, sure, but in a sport measured literally down to the 10,000th of a second, it\u2019s kind of a big deal. Maybe not enough to help Ferrari make the right strategy calls, of course.<\/p>\n All that data is helping F1 teams race faster, of course. But it\u2019s also helping Formula 1 enter the AI era.<\/p>\n Roberts said his team is replacing F1\u2019s laptops with AI-enabled versions and shifting workloads that used to live in the cloud down to on-prem infrastructure and edge devices. That\u2019s closer and faster: important for a mobile operation.<\/p>\n But it also has operational benefits. Roberts described a scenario where AI agents running on edge compute can autonomously diagnose and recover a network node at the far end of a circuit, reducing the need for his team to run or drive over and fix it manually.<\/p>\n \u201cIf you\u2019ve got an AI agent that can then autonomously do some fault finding, understand why it might be disconnected, perform some actions and then get itself reconnected … this is a game changer for any organization,\u201d he said.<\/p>\n Predictive maintenance is another lower-hanging fruit. Formula 1 is using AI to flag which of the 100+ network switches deployed around a racetrack might be approaching failure, months before it happens. Then the team can replace it proactively, reducing potential disconnects and downtime during a race weekend. Statistical analysis of telemetry is another that can help the broadcast crew. They\u2019re using small on-prem AI appliances \u2014 Lenovo\u2019s Mac Mini-like PGX device \u2014 that can run open-source LLMs locally, including custom models trained on a customer\u2019s proprietary data.<\/p>\n But neither Roberts nor Speranzo sees AI replacing the human element of the core of the sport: driving and racing.<\/p>\n “Drivers are human. If it\u2019s all replaced by AI that’s controlling everything, isn’t it a bit boring, a bit dull?” Roberts said. “The human element is what makes it engaging.” <\/p>\n The 2026 season has brought new regulations, new hybrid power units with more sustainable fuels, two new teams in Audi and Cadillac and Ford returning to F1 as a technical partner to Red Bull.<\/p>\n Alongside those and future updates will come more and more data, Roberts says.<\/p>\n He framed his job as building infrastructure that lets F1 “pivot 90 degrees in a matter of days.” <\/p>\n If the past two years are any guide \u2014 500 TB to 650 TB per weekend \u2013 the race of the future might well generate over a petabyte of data, 1,000 terabytes, every three days. <\/p>\n<\/div>\nThe data is growing because the audience is changing<\/h3>\n
AI is entering the equation<\/h3>\n
What\u2019s next: the petabyte race weekend?<\/h3>\n