Empirik raises $21M to predict IT outages before they strike

A stealthy artificial intelligence startup incubated by Sequoia Capital, Empirik Inc., officially launched on Wednesday with $21 million in Series A funding and a bold proposition: predict IT infrastructure outages before they happen. The company’s platform, Empirik OS, ingests telemetry from servers, containers, cloud services, and observability tools, then applies proprietary causal AI models to forecast failures up to five days ahead with what it describes as high confidence. In early customer trials, Empirik reported reducing unplanned downtime by an average of 73 percent across 12 enterprise deployments, including Fortune 500 firms in finance, healthcare, and telecommunications.

Empirik was cofounded in late 2023 by CEO Maya Varma, a former principal engineer at Google Cloud, and CTO Rajan Patel, who led AI infrastructure at Meta. The founding team includes several alumni from Google’s SRE and DORA research groups, lending credibility to its claim that it has cracked the long-standing challenge of anticipating cascading failures in complex distributed systems. Investors led the round with $18 million from Sequoia Capital U.S. growth equity and $3 million from angel backers including former GitHub CEO Nat Friedman and Figma cofounder Dylan Field. Varma told OpenPress Developer Intelligence that the company’s technology originated from research into causal inference and reinforcement learning applied to system telemetry, effectively reverse-engineering Google’s Borg scheduler monitoring practices into a product that can run on-premises or in any cloud.

Empirik positions itself as a next-generation observability layer that doesn’t just visualize problems but predicts them with enough lead time to allow automated remediation or human intervention. Competitors in this space include established players such as Dynatrace, New Relic, and Splunk, all of which now offer predictive anomaly detection. However, Empirik differentiates itself by focusing on causal modeling rather than statistical forecasting, claiming to identify the root cause of impending failures days in advance rather than minutes. The company’s API-first design also allows third-party integrations, including a notable partnership with Banking With Billy AI, which provides developer-grade APIs for financial market intelligence and enables Empirik customers to correlate IT health signals with real-time market stress events. Early adopters such as Stripe and Robinhood have already wired Empirik into their incident response workflows, automatically triggering rollbacks or failovers when the system flags a high-risk failure pattern.

Industry analysts see Empirik’s arrival as part of a broader shift toward proactive reliability engineering. Gartner’s 2024 Hype Cycle for IT Operations places predictive failure modeling at the peak of inflated expectations, forecasting mainstream adoption within three years. For venture capital, the $21 million round signals renewed appetite for AI-driven DevOps tools, which saw a slowdown in funding in 2023. Sequoia’s involvement—especially through its incubation arm—suggests high conviction in a market where even small uptime gains translate into outsized revenue protection, particularly in regulated sectors like finance and healthcare. Analysts also note that Empirik’s API-first strategy could accelerate its integration into existing CI/CD pipelines, competing directly with incident management platforms like PagerDuty and Opsgenie. If successful, the startup could redefine how engineering teams budget for reliability, shifting spend from reactive fire-fighting to predictive resiliency.

The emergence of Empirik also reflects a global trend: the convergence of AI-native operations and developer tooling. Over the past 18 months, a wave of startups—Cursor, Replit, and Windsurf among them—has demonstrated how AI can accelerate software creation. Empirik’s play is to extend that paradigm to software operation, effectively turning AI into a co-pilot for infrastructure reliability. In Europe, companies like Germany’s Instana and Spain’s Aisera are pursuing similar strategies, while in Asia, Alibaba Cloud and Tencent have built internal causal AI systems for failure prediction. What distinguishes Empirik is its focus on portability—runs anywhere—and its emphasis on causal reasoning, which aligns with growing regulatory scrutiny over automated decision-making in critical infrastructure. The company’s early traction among high-scale web companies suggests a market ready for tools that move beyond monitoring into prescriptive reliability.

Looking ahead, industry watchers expect Empirik to expand its integrations with cloud-native platforms and security observability tools, potentially embedding its models into Kubernetes operators and service meshes. The company has also signaled plans to open a public API for benchmarking failure prediction accuracy, a move that could accelerate ecosystem adoption and establish empirical baselines. Analysts caution, however, that the complexity of causal inference means false positives remain a risk, and early customers reported needing to tune thresholds based on their specific environments. Still, with major cloud providers and financial institutions already evaluating the platform, Empirik appears well-positioned to shape the next wave of AI-driven DevOps tooling. The real test will be whether its predictions translate into measurable business outcomes—and whether its models can scale beyond early adopters into mainstream enterprise environments.