France-based Tsuga has raised a $35 million Series A to scale its AI-native observability platform, as enterprises face rising telemetry costs, shrinking budgets and more complex AI workloads.
The round was led by Singular, with participation from General Catalyst, DST Global and QuantumLight. Picus and Databricks Ventures also joined. Singular and General Catalyst previously backed Tsuga’s $10 million seed round in December 2025.
The funding will be used to expand Tsuga’s team to 100 people and accelerate the rollout of the platform.
Tsuga is targeting a structural problem in the observability market where customer telemetry is shipped into a vendor’s cloud, stored it there, and charged as data volumes rise.
That model was already under pressure before the AI boom, according to Gabriel-James Safar, Tsuga’s co-founder and CEO.
“In 2021, we already started to see that sharp increase in the size of IT systems,” Safar told Pathfounders. “The pace of infrastructure growth, and hence the need and cost of observability, grew very, very fast.”
AI has now made that pressure worse, he says. “IT systems are growing faster than ever,” Safar said. “At the same time, you have budgets that are shrinking because the goal is to fund these AI systems. So, how do you do more with less is essentially the challenge that people are facing.”
Tsuga’s answer is to deploy inside the customer’s own cloud account, across Azure, AWS, Google Cloud and regional sovereign clouds, rather than ingesting telemetry into its own cloud. The company claims this avoids duplication costs, reduces governance risk and removes the need for sampling.
Safar argues that incumbents such as Datadog and Dynatrace are not being beaten by a feature gap, but by an architectural and business-model problem. “These are good products, these are good companies, but they are stuck in this innovator’s dilemma,” he said.
He said large SaaS observability platforms are made up of thousands of services, making customer-side deployment slow and expensive. “What we created is a system that can deploy for a few thousand dollars a month into the customer system, and we can deploy in an hour,” he said.
Tsuga brings together traditional application and infrastructure traces with AI agent traces, prompt and token visibility, confidence metrics and agent call graphs. Safar said the company does not see AI observability as a separate category, but as part of the same operational stack.
“To us, these are just different types of services and IT systems that companies and enterprises run,” he said. “In the end, large metrics, traces, chain of thoughts… all that is just data that people need to understand how their systems work.”
The company says its strongest fit is with customers facing three pressures: governance, sovereignty and scale. Safar said telemetry often includes personal data or company secrets, making control over access critical. Sovereignty is also becoming more important as enterprises ask where their data is stored and whether foreign cloud dependencies create operational risk.
“Our thesis is that there will be an increase of attention and care from executives on: who am I working with, and am I packed enough technologically not to disappear if some foreign power uses a kill switch?” Safar said.
Tsuga is also pitching itself as “software and a service” or a model where forward-deployed engineers work with customers to tune their observability environments, reduce data volumes and improve setup.
“SaaS is dead,” Safar said. “We have software and a service.”
Six months after coming out of stealth, Tsuga claims it has millions in contracted ARR, six-figure average contract values, and customers including Le Monde, Camunda, Buk and Black Forest Labs.
