Dust, a Paris-based enterprise AI platform, announced a $40 million Series B led by Sequoia and Abstract, with participation from Snowflake Ventures and Datadog. Axios Pro broke the news on May 18, and the round brings Dust’s total funding to over $60 million. The deal is a clear vote that the next layer of enterprise AI spend goes to shared, team-level agent infrastructure rather than individual copilots.
Dust’s metrics are the reason the round closed at this size. The company has deployed 300,000 agents across 3,000 organizations and reached $20 million in ARR. Net revenue retention is 240 percent, and the company has reported zero churn in 2025, per Axios Pro and a corroborating Sifted feature. Those numbers belong in the rare category of enterprise SaaS metrics that justify a primary round at this stage without obvious caveats.
The founding context informs the thesis. Dust was co-founded in 2022 by two former Stripe engineers, including Stanislas Polu, an ex-OpenAI researcher. The company’s positioning argument, repeated in its founder commentary, is that most enterprise AI deployment today is “single-player”: one user, one copilot, one workflow. Dust’s product is built around the opposite assumption, that the durable enterprise value comes from agents shared across teams and connected to over 100 internal data sources. Sequoia’s investment thesis appears to be a direct bet on that frame.
The competitive context is crowded. Glean, Hebbia, Writer, and several large-vendor copilots all sell into adjacent enterprise budgets. Dust’s 240 percent net retention is the differentiating data point. That figure means the typical Dust customer is spending 2.4 times more this year than last year on the same product, without churn. Few enterprise software categories have ever sustained that expansion rate, and the question is whether multi-agent enterprise deployments are genuinely structurally expansive or whether the number reflects an early-adopter cohort that will normalize.
For founders building in enterprise AI, three operational reads stand out.
- 240 percent NRR is now a benchmark, not an outlier. Sequoia priced this round on the assumption that comparable agentic platforms can hit similar retention. Boards evaluating competing companies will reference it.
- Multi-agent is the new sales motion. Selling one agent into one workflow is becoming a wedge product; the renewal expansion comes from the second, third, and fifteenth agent that the customer adds. Pricing models built around per-seat single-agent assumptions will compress in this market.
- The data-source moat is real. Dust’s 100-plus integrations are an explicit part of Sequoia’s investment memo, per the founder’s own framing in the Sifted feature. Companies treating integrations as a roadmap nice-to-have should accelerate them.
The geographic note also matters. Dust is European-founded but the United States is now its fastest-growing market, and the round is explicitly earmarked for US R&D expansion. Sequoia’s lead reinforces a pattern visible in several recent rounds: European AI infrastructure companies are pricing as US-coast deals when the customer mix is American.
Snowflake Ventures and Datadog as strategic investors deserve attention separately. Snowflake’s participation aligns with its enterprise data-cloud strategy, which positions agentic AI as a workload running on top of Snowflake-resident data. Datadog’s participation signals that observability for multi-agent systems is becoming a vendor priority. Both strategics are likely customers of Dust as well as investors, which compresses the typical strategic-investor information lag.
For operators evaluating multi-agent platforms in the next contract cycle, the question to ask is not whether the platform supports multi-agent workflows. Most now do. The question is what the platform’s net retention looks like 12 months in, because that number reveals whether the platform is delivering compounding value or one-time wins.
Originally reported by Axios Pro on May 18, 2026.
