Tsuga secures $35 million to maintain AI-driven observability within the customer's own cloud.

      The Paris-based startup, launched by two former Datadog employees, aims to eliminate the per-byte pricing model as demand for telemetry surges due to AI workloads. Tsuga, a company developing observability software tailored for the AI era, has successfully secured a $35 million Series A funding round just six months after emerging from stealth mode and positioning itself against the very market its founders once contributed to building.

      The funding round is led by Singular, with General Catalyst participating again after backing Tsuga’s $10 million seed round in December 2025. New investors DST Global and Quantumlight have joined, along with participation from Picus and Databricks. This latest round brings the total funding for the company to approximately $45 million within six months, highlighting both investor enthusiasm for AI infrastructure and Tsuga’s potential.

      The founders, Gabriel-James Safar and Sebastien Deprez, are driven by a familiar frustration. After selling their previous company, Madumbo, to Datadog in 2019 and spending years in the observability sector, they concluded that the existing business model had become outdated. Traditional platforms collect a customer’s telemetry, store it in the vendor’s cloud, and charge more as the customer's infrastructure expands. The data produced by every agent loop and autonomous deployment in an AI system exceeds what these platforms were designed to manage, leading to escalating costs.

      Tsuga's approach is to reverse this model. Instead of gathering telemetry in its cloud, the company operates within the customer's environment, ensuring that data stays within the customer’s perimeter and eliminating per-byte ingestion fees. Engineers deployed on-site collaborate with client teams to optimize the system and reduce the amount of data processed and stored. Automated root-cause analysis is conducted on complete and unsampled data, and a bundled MCP server along with a command-line tool allows engineering teams to create their own agents within their security framework.

      Keeping data on-site also serves as a regulatory advantage—particularly relevant in Europe. The rationale that drives governments toward homegrown alternatives to American software extends to telemetry, which can reveal sensitive insights into a company's systems and AI models. For clients in regulated sectors, the assurance that observability data will not be sent to third-party clouds is a key selling point rather than just an additional feature.

      The competitive landscape is bustling, with observability emerging as a dynamic segment within enterprise software. Venture capital is being invested in AI-oriented monitoring solutions as buyers seek tools capable of managing agent-based systems. Tsuga is betting that its architecture will distinguish the next generation of solutions from older designs.

      The investment from Databricks also establishes a partnership. Tsuga is a Databricks partner, facilitating the direct routing of observability data into Databricks for further analysis, aligning with Databricks' strategic push into security and operational data. There is a shared belief that the data layer and observability layer are best integrated within the customer's own environment.

      According to Tsuga, since its launch in December 2025, it has quickly amassed several million dollars in contracted annual recurring revenue, with an average contract value reaching six figures, securing clients like Le Monde, Camunda, Buk, and Black Forest Labs. Tsuga claims that Le Monde utilized the platform to monitor its infrastructure during the French municipal elections, while Camunda and Buk employed it across multi-cloud setups with stringent data residency standards. However, these figures are company-reported and not audited, and the referenced clients primarily represent the European and AI-native market for which the residency argument is tailored.

      CEO Safar indicated that the new funding will be directed toward team expansion, development of the Skills library and agent-building toolchain, and scaling the embedded engineering model. The sustainability of this model as Tsuga expands remains uncertain: hands-on, embedded engineering can be costly to scale, and the present gamble is that customers will be willing to invest in it.