Figma recently revealed its cloud computing costs in its initial public offering (IPO) filing, revealing a daily expenditure of around $300,000 on Amazon Web Services (AWS). As a result, the company allocates approximately $100 million each year, which accounts for about 12% of its reported revenue of $821 million.
Figma, a widely used interface design tool, relies entirely on AWS for its “computing, storage capabilities, bandwidth, and other services.” The S-1 filing further details a renewed hosting agreement with AWS, signed on May 31, 2025, which commits Figma to a minimum spend of $545 million in cloud hosting services over the next five years. However, the filing doesn’t break down how these costs are allocated across specific services, such as storage, compute, or bandwidth.
In addition to a substantial financial investment, Figma’s filing emphasizes the risks associated with extensive cloud integration. The company acknowledges its complete reliance on AWS’s performance, which makes it vulnerable to potential outages. Furthermore, AWS has the authority to change and interpret its terms of service and other policies, including during contract renewals, which could negatively affect Figma’s business operations. If AWS were to cancel the contract entirely, it would create significant challenges for Figma, as its cloud service infrastructure is specifically designed to operate within the AWS ecosystem.
Per Borgen, a CEO at Scrimba, posted on LinkedIn:
For comparison, at Scrimba, we spend far less than 1% of our revenue on infrastructure. This is because we skipped the cloud and run on dedicated servers instead. But it’s not just about cost. Figma’s infra is tightly coupled to AWS. If Amazon changes terms or pulls the plug, they’re in trouble. And it’s all by their own admission. So the question isn’t just “Is it expensive?” It’s also “How much control are you giving up?”
The deep entanglement of Figma with a cloud provider goes far beyond simply using virtual machines. As one insightful commenter on Hacker News, nevon, explained:
A common misconception about moving off the cloud is viewing it solely as a VM provider. In reality, the cloud integrates deeply into your systems. Your permissions rely on cloud identities, firewalls use security group references, and cross-region connectivity depends on cloud networking. You manage secrets with cloud tools, monitor metrics through cloud observability, and often utilize whitelisted IP ranges provided by the provider. Additionally, database upgrades, VM images, and auditing are tied to the cloud services. Even your applications may depend on cloud-managed containers and event buses, while disaster recovery plans hinge on the provider’s backup and failover capabilities.
The explanation of nevon clarifies why migrating from a cloud platform, even for a large company like Figma, is not a quick or simple task.
Figma’s revelation adds to the ongoing industry discussion about the escalating costs of cloud computing as companies scale. This has led some organizations exploring the repatriation of specific workloads or data from the public cloud to on-premises or co-located infrastructure. A prominent example is 37signals, whose CTO, David Heinemeier Hansson, has been a vocal advocate for exiting the cloud. They initiated their repatriation efforts in 2022, when their annual cloud bill exceeded $3.2 million. By October 2024, 37signals estimated a $2 million savings that year by moving away from cloud services. Their latest phase involves exiting AWS’s S3 storage service, which Heinemeier Hansson anticipates will save the company around $1.3 million annually.
Ultimately, Figma’s substantial cloud bill serves as a stark reminder of the core dilemma facing modern tech companies: the trade-off between the agility and convenience offered by hyperscale cloud providers versus the escalating costs and the inherent risks of deep vendor lock-in.
