Figma Builds In-House Redis Proxy to Hit Six Nines Uptime

Figma's decision to build its own Redis proxy reveals something important about how modern engineering teams approach infrastructure resilience. The company recently detailed FigCache, an in-house solution that has delivered six nines of uptime across its caching layer since deploying in late 2025. This level of reliability—99.9999% availability—represents not just a technical achievement but a philosophical commitment to owning the critical systems that underpin user experience. For teams wrestling with distributed systems complexity, this story offers both inspiration and practical insight into how targeted infrastructure investment can dramatically improve reliability. As organizations increasingly depend on AI agents to transform data science work in 2026, the underlying systems supporting these innovations demand the same level of attention and sophistication that Figma has demonstrated.

The real story extends beyond the metrics. Figma's previous fragmented caching stack had become a liability, threatening site availability and creating operational friction. Rather than applying temporary fixes, the team chose to fundamentally rethink their approach—an investment that speaks to a mature engineering culture willing to tackle root causes. This reflects a broader principle: sustainable reliability comes from understanding your systems deeply enough to improve them, not simply relying on vendor solutions to mask underlying complexity. The distinction between patching problems and solving them is what separates organizations that scale confidently from those that merely survive.

What makes this announcement particularly noteworthy is the transparency with which Figma shared its journey. Publishing detailed accounts of engineering decisions—including the reasoning behind building rather than buying—helps the broader tech community learn from real-world experience. This openness aligns with a progressive view of industry collaboration where competitors share knowledge to raise overall standards. It also serves as a reminder that infrastructure decisions are rarely purely technical; they reflect organizational values around ownership, long-term thinking, and willingness to invest in foundations that may not show immediate returns but pay dividends over time.

For engineering leaders evaluating their own infrastructure strategies, Figma's experience offers a useful framework. The team identified a genuine pain point affecting site availability, assessed whether existing solutions could address it, and made a deliberate choice to build a custom solution when the answer was no. This is not a prescription for always building in-house—rather, it's an example of thoughtful evaluation leading to appropriate action. The result speaks for itself: six nines of uptime does not happen by accident. It happens when teams combine technical expertise with the organizational courage to make meaningful investments in reliability.

Looking ahead, the question becomes how this approach scales across an industry increasingly dependent on complex, distributed systems. As companies like Origin Lab raise millions to help video game companies sell data to world-model builders, the infrastructure supporting these emerging use cases will face similar reliability challenges. Figma's story suggests that the organizations best positioned to thrive will be those willing to move beyond generic solutions and build systems tailored to their specific needs—while sharing their learnings generously with the community that benefits from them.