Scaling global game infrastructure using AWS Local Zones with Riot Games

Prior to adopting AWS Local Zones, Riot Games ran multiple titles on AWS, using a combination of AWS Regions and AWS Outposts—a family of fully managed services delivering AWS infrastructure and services to virtually any on-premises or edge location—to host game servers. Using AWS Outposts, the company can deploy AWS compute in its own data center spaces, filling gaps in locations where it needs low-latency access for players. Riot Games wanted the flexibility to provision additional capacity for peak player demand while paying only for the periods of active capacity use.

As AWS expanded AWS Local Zones in more locations, Riot Games saw an opportunity to modernize its infrastructure while maintaining its global footprint. AWS Local Zones offer the same elastic, on-demand compute as AWS Regions but also placed core AWS services closer to metropolitan areas worldwide. With this approach, the company could scale capacity based on player activity without managing physical facilities.

Riot Games developed a standardized global deployment model: Prefer AWS Regions when available, use AWS Local Zones to reduce latency when AWS Regions aren’t nearby, and rely on AWS Outposts where neither option exists today. As AWS Local Zones launched in new locations, the company migrated workloads from AWS Outposts—gradually shifting toward a more flexible model where AWS managed both the infrastructure and data center facilities.

Riot Games now runs latency-sensitive game servers across 17 AWS Local Zones worldwide. These servers host real-time gameplay for titles like VALORANT and League of Legends, where consistent, low-latency connections are essential—not only to maintain competitive integrity but also to give matchmaking systems better options for equalizing the experience across all players in a game.

The company deploys both game servers and supporting services using Amazon Elastic Kubernetes Service (Amazon EKS), which is used to build, run, and scale production-ready Kubernetes applications. These applications run as managed nodes on Amazon Elastic Compute Cloud (Amazon EC2) instances, which provide secure and resizable compute capacity for the underlying workloads. Because AWS Local Zones and AWS Regions share the same APIs and tools, Riot Games can use identical deployment processes across all locations.

“As a game team, we don’t have to think about whether our game is running in an AWS Local Zone, an AWS Outpost, or an AWS Region—it’s all the same to us,” says David Press, tech lead for League of Legends at Riot Games. “The fact that the infrastructure presents a very similar interface, along with consistent billing, tools, and services, is a huge win. For highly mature services running at scale, that consistency matters just as much. This empowers us to focus on our game rather than our infrastructure.”

To connect these distributed deployments to its Riot Direct network backbone, Riot Games uses AWS Direct Connect, a service that provides a dedicated network connection to AWS. Over the process of rolling out the global deployment model, Riot Games has worked closely with AWS support teams to plan capacity and coordinate launches.

“Our AWS technical account managers and solutions architects are working alongside us as we launch new games and in new AWS Local Zones,” says Press. “This makes us feel like collaborators rather than just customers.”

By adopting AWS Local Zones, Riot Games has achieved consistent sub-35 ms latency for VALORANT players and sub-70 ms latency for League of Legends players in supported metropolitan areas. For example, in Türkiye—a high-growth country—Riot Games uses the AWS Local Zone in Istanbul to bring low-latency gameplay to the country's thriving gaming community. This has unlocked significant new capacity for a rapidly growing player base.

By using the elastic, on-demand model of AWS Local Zones, Riot Games has reduced annual infrastructure spending by millions of dollars. With automatic scaling, the company no longer pays for idle capacity during off-peak hours, and has access to newer Amazon EC2 instance types that deliver improved price performance. As AWS expands its global footprint, Riot Games can enter new locations faster without building out its own data center presence.

For the company’s game developers, standardized deployments across AWS Local Zones and AWS Regions have streamlined operations. Teams can focus on creating games rather than managing infrastructure. “We want to deliver the best possible player experience and solve real player problems, and that means investing in infrastructure,” says Press.