Crypto.com scales 4x faster with 20% better price-performance

Crypto.com supports more than 150 million users worldwide across cryptocurrency trading, payments, wallets, and digital asset services. As the platform grew, it needed to maintain fast, secure, and uninterrupted user experiences in a market that runs 24/7, 365 days a year with no maintenance windows.

Traffic could surge suddenly with market activity, requiring backend infrastructure that could add capacity quickly without affecting performance or availability. At the same time, a larger user base increased pressure on security and infrastructure costs, making it more important for Crypto.com to manage resources efficiently across its growing environment.

Because its services run continuously, Crypto.com also needed to update, scale, and protect critical workloads without planned downtime. The company needed a more flexible infrastructure foundation that could respond faster to changing demand while keeping operational effort and cost under control.

“Unpredictable traffic was one of our biggest infrastructure challenges,” says Black Tam, director of infrastructure at Crypto.com. “As our user base grew beyond 150 million, our existing infrastructure could no longer support that level of demand cost-effectively.”

Crypto.com has run key application workloads on AWS for several years, evolving its infrastructure as the platform scaled. The company first ran application workloads on Amazon Elastic Compute Cloud (Amazon EC2), then moved to Amazon Elastic Container Service (Amazon ECS) as its containerization needs grew. As market activity became more unpredictable, Crypto.com optimized its backend environment on Amazon Elastic Kubernetes Service (Amazon EKS) to gain more control over how quickly it could scale critical application services.

The move to Amazon EKS gave Crypto.com a more flexible containerized foundation for microservices supporting its core mobile application. Previously, teams had to rely on pre-provisioned capacity and Auto Scaling groups that could take several minutes to add new compute capacity during traffic spikes. By adopting Karpenter on Amazon EKS, Crypto.com automated node provisioning, helping the company scale based on real-time demand when trading activity surged, while reducing the operational effort required from site reliability and monitoring teams.

Crypto.com also migrated key backend workloads from Intel-based instances to AWS Graviton Processors. With support from AWS, the company assessed how its applications could run on Arm-based architecture and built confidence across application teams to move suitable microservices to AWS Graviton. Today, most of these microservices run on AWS Graviton, helping Crypto.com improve price-performance while keeping legacy workloads on Intel or AMD where needed.

“With our Amazon EKS environment, we have a stronger foundation for critical application services that need to keep running while we update, scale, and optimize our infrastructure,” says Tam.

Crypto.com can now respond faster to sudden shifts in crypto market activity. It reduced the time needed to launch each new compute node from 2–3 minutes to about 30 seconds, helping teams add capacity more quickly during periods of increased demand.

The migration also improved performance for latency-sensitive trading activities. Crypto.com achieved 5–10 milliseconds faster latency for a core application programming interface (API), while peak performance for the core API improved by about 35 percent.

Furthermore, Crypto.com improved price-performance by 20–30 percent after adopting AWS Graviton processors for key backend workloads, while reducing infrastructure costs by 10–20 percent. The company can now run workloads across different instance types based on resource availability and application needs, helping teams use compute resources more efficiently across its AWS environment.

“With this migration, we can respond faster when traffic spikes and provide lower latency to end users,” says Tam. “That helps us deliver a better user experience while managing infrastructure more efficiently.”