Engineering Formula One: cloud and AI are changing the game

The era when advanced technologies, like the cloud, were confined to one or more industries is long gone now. At this age, tech is in every industry, and sports is one of them for sure. Tech is leaving its fingerprints in every sport, but no sport comes close to the level of technology used seen in F1.

While attending AWS Re:Invent 2025 last December, we had the chance to meet David King, Head of Digital Technology, and Ryan Kirk, Head of Cloud DevOps at Formula One. They spoke to us extensively about F1’s approach to tech implementation and how the cloud and AI analytics are revolutionizing the beloved sport.

How would you explain your roles within Formula One?

Ryan Kirk:

I’ve been with Formula One for ten and a half years and have been involved with it for nearly fifteen. I started in desktop support, moved into systems engineering, and today I lead our cloud engineering team. That includes our cloud platforms, primarily AWS;; our DevOps function across the business, on-prem Kubernetes; and some MLOps. Essentially, I oversee the platforms that power our operational technology backbone.

David King:

My team is responsible for all of our owned and operated digital platforms; anything consumer-facing that we haven’t licensed. That includes F1.com, the F1 App, F1 TV, and the Grid game. We manage the design, build, and delivery of change across those platforms, ensuring the digital experience matches the scale and complexity of the sport.

How do you design an architecture that supports your cross-border activities?

Ryan Kirk

Experience is foundational. We have engineers who have been with F1 for ten, fifteen, or even thirty years. They understand our operations inside out. As we onboard new technologies, we build on that operational knowledge rather than replacing it.

There are many variables to manage, but our primary goal remains constant: delivering the Formula One product that fans see on TV and across platforms. We combine institutional experience with careful adoption of new technologies, working closely with partners to ensure everything integrates seamlessly.

David King

If we’d had this conversation four or five years ago, the answer would have been very different. Historically, we traveled with significant infrastructure and did everything at the circuit. We don’t operate that way anymore.

We now have a fixed production hub in the UK where much of the final production, monitoring, and distribution preparation takes place. While we still maintain a substantial trackside presence, the final product is created centrally and distributed from there. That stable anchor point allows us to maintain direct connectivity and creates architectural consistency across a global calendar.

Data is central to every race. What are the main challenges around latency, accuracy, and reliability?

Ryan Kirk

Latency tolerances are something we understand deeply because we’ve been running this operation for decades. The transition to remote production tested those tolerances. Previously, everything was trackside. Now, with production centralized in the UK, we’ve had to understand how systems, applications, and even production teams operate under different latency profiles.

We’re hyper-aware of what is acceptable from both a technical and human perspective.

David King

Latency exists at multiple stages. It begins with data capture, including radio frequency telemetry from cars traveling at 200 miles per hour, then moves through distribution back to the UK, and finally into digital delivery to fans.

With F1 TV, for example, we introduced multi-view functionality, allowing fans to watch synchronized feeds from multiple camera angles. That required precise latency alignment across all streams. If one feed lags behind another, the experience breaks down.

Latency also applies to replay turnaround — how quickly we can surface a key moment without detracting from what’s happening live. So latency isn’t just network delay; it’s about storytelling timing.

How is AI being used across F1’s operations today?

Ryan Kirk

When AI became a major industry focus, we deliberately stepped back. Instead of rushing in, we asked: Where can this genuinely improve our operation?

One of our most impactful applications is a root cause analysis (RCA) tool. Our environment includes hundreds of applications and multiple engineering disciplines. Traditionally, diagnosing an issue could involve multiple teams and long communication chains.

Our AI-driven RCA tool has awareness of over 500 applications. When there’s a potential degradation, it analyzes logs, network data, system metrics, and application behavior in parallel. It correlates findings and generates a root cause conclusion. If escalation is required, the tool creates a detailed ticket with much of the investigation already completed. Without AI, this simply wouldn’t be possible at the same speed or scale.

David King

On the broadcast side, we have Track Pulse. It analyzes telemetry, positioning data, timing feeds, and FIA race control messages to identify unfolding narratives within a race.

Experienced operators can often anticipate what might happen next by reading timing screens. Track Pulse complements that by consistently analyzing vast volumes of data, up to a million data points per second per car, and surfacing stories that might otherwise be missed.

It supports both live production and post-production highlight creation. While similar outcomes might theoretically be achieved without AI, the volume and consistency would not be possible at this scale.

F1 teams themselves use AI extensively. How does your role differ from theirs?

David King

We are the commercial rights holder of the sport. The teams have different use cases. They access extensive car telemetry, much of it private, to optimize driver performance and car setup using their own machine learning systems.

We have access to certain data sets to support broadcast and fan engagement, but the teams operate their own performance-focused AI environments. Our AI initiatives center on media production, digital delivery, and operational efficiency.

We want meaningful personalization, not an uncontrolled recommendation engine.