Tsuga raises $35M to reinvent observability in the AI era
The essentials. Tsuga, a Paris-based startup founded by two Datadog alumni, announced on June 23, 2026 a $35 million Series A led by Singular. Its promise: AI-native observability deployed directly in the customer's cloud, where telemetry never leaves the company's perimeter. The round comes six months after a stealth exit that was already profitable in revenue terms.
Key takeaways
- $35M Series A led by Singular, with General Catalyst, DST Global Partners, QuantumLight, Picus Capital, and Databricks Ventures (Tsuga, press release, 2026).
- Founders: Gabriel-James Safar and Sébastien Deprez, who sold their previous startup Madumbo to Datadog in 2019.
- Six months after emerging from stealth: several million dollars in ARR and average contract values in the "six figures".
- Customers cited: Le Monde, Camunda, Buk, Black Forest Labs. Headcount of around 40, targeting ~100 by early 2027.
- Core model: telemetry stays in the customer's cloud, with no per-byte ingestion tax.
What Tsuga does
Tsuga builds an "AI-native" observability platform. Observability is the set of tools that lets an engineering team understand what's happening in its production applications: logs, traces, metrics. A market historically dominated by Datadog, New Relic, or Grafana.
Tsuga's difference comes down to an inverted model. Traditional platforms pull the customer's telemetry into their own cloud, then bill by the byte ingested. Tsuga deploys inside the customer's environment — AWS, Azure, Google Cloud, or regional sovereign clouds — and the data never leaves it.
The product natively integrates what AI has added to the stack: agent traces, visibility into prompts and tokens, call graphs. In other words, it's built for a world where part of the code is written and deployed by models, not just by humans. Pricing is advertised as transparent, per gigabyte.
The positioning targets companies whose data volumes are exploding and for whom governance — keeping sensitive data within their own perimeter — has become a deciding factor, not a detail.
Funding details
- Amount: $35 million.
- Announcement date: June 23, 2026.
- Round: Series A.
- Lead investor: Singular (existing investor).
- Co-investors: General Catalyst, DST Global Partners, QuantumLight, Picus Capital, Databricks Ventures.
- History: a $10M seed round in December 2025 with General Catalyst, after coming out of stealth in late 2025.
- Headquarters: Paris, France.
- Founders: Gabriel-James Safar (CEO) and Sébastien Deprez, both Datadog alumni following the acquisition of Madumbo in 2019.
- Use of funds: product development, engineering hires, commercial expansion.
Why this raise matters
Beyond the amount, three takeaways stand out.
First, the founders' profile. Selling your company to Datadog, spending several years there in key roles — product management and engineering on a major product — then leaving to take on your former employer head-on: that's not a pitch, it's intimate knowledge of the problem. Investors aren't funding an idea, they're funding people who have seen from the inside where the dominant model creaks.
Next, traction before the raise. Several million in ARR six months after coming out of stealth, with six-figure contracts and clients like Le Monde or Camunda — that changes the nature of the round. This isn't a classic early-stage bet: it's a Series A accelerating a machine that's already running. That's what explains the presence of such a dense syndicate — Singular leading the round, but also DST Global and Databricks Ventures.
Finally, the real strategic message lies in the architecture, not the amount. By refusing to route data through its own cloud, Tsuga eliminates the marginal ingestion cost that underpins the entire rest of the market. Henri Tilloy of Singular puts it bluntly: "by staying entirely out of the data path, Tsuga removes every structural disadvantage the rest of the market is built on." It's less a new feature than a business model shift — and that's usually where category shifts happen.
One open question remains: a "deployed at the customer" model is more complex to operate and scale than centralized SaaS. The profitability of this promise at scale will be the real test over the coming quarters.
What this raise reveals about the observability market in 2026
Beyond Tsuga, the round sheds light on three underlying dynamics in the sector.
AI is sending volumes — and bills — soaring
When part of the code is generated and deployed by agents, the volume of telemetry to monitor scales up. The per-ingested-byte billing model, inherited from an era of "manageable" volumes, becomes a budget pain point. Several players respond with aggressive sampling — at the cost of visibility gaps. The cost of observability has become a C-suite topic, no longer just a technical team concern.
Data governance becomes a purchase criterion
Keeping telemetry within your own perimeter is no longer a paranoid CIO's whim. Between regulatory requirements, sovereign clouds, and the sensitivity of application data, "bring your own cloud" is becoming a selling point. Tsuga's bet is that this criterion will carry increasing weight in B2B RFPs.
Cap tables are getting denser around AI infrastructure topics
Seeing Singular, General Catalyst, DST Global, and Databricks Ventures on the same Paris round says a lot about funds' appetite for AI-era infrastructure layers. Observability joins storage, orchestration, and security on the list of must-have building blocks. For the French ecosystem, it's a signal: B2B deeptech topics now attract international syndicates as early as Series A.
FAQ
Who are Tsuga's founders?
Tsuga was founded in Paris by Gabriel-James Safar (CEO) and Sébastien Deprez. The two had previously founded Madumbo, sold to Datadog in 2019, where Safar led product management and Deprez led an engineering team on a major product before launching Tsuga in 2024.
How much has Tsuga raised in total?
Tsuga raised $35 million in Series A funding in June 2026, led by Singular. This follows a $10 million seed round closed in December 2025 with General Catalyst, for a cumulative total of $45 million since coming out of stealth in late 2025.
What does Tsuga actually do?
Tsuga develops an AI-native observability platform that deploys in the customer's own cloud (AWS, Azure, Google Cloud). Telemetry — logs, application traces, AI agent traces, prompts, and tokens — never leaves the customer's environment, with no per-byte ingestion tax.
Who are Tsuga's customers?
Six months after coming out of stealth, Tsuga reports several million dollars in ARR with average contract values in the "six figures". Its customers include Le Monde Group, Camunda, Buk, and Black Forest Labs. Headcount is around 40, with a target of ~100 by early 2027.
How does Tsuga differ from Datadog?
Datadog pulls telemetry into its own cloud and bills by the byte ingested. Tsuga flips the model: the platform runs in the customer's environment, the data never leaves it, and pricing is transparent per gigabyte. Eliminating the ingestion cost is its core selling point.