Dust has raised $40 million in Series B funding, led by Sequoia Capital and Abstract Ventures, with participation from Snowflake Ventures and Datadog. The round brings the company's total funding to over $60 million.
More notable than the funding amount is the label Dust has given itself: "multiplayer AI." The concept transforms enterprise AI from a single-player experience into a collaborative workspace.
The problem with single-player AI
Employees use ChatGPT, Claude, or Copilot individually, feeding prompts into private chat windows where results disappear into personal history. This has been the default enterprise AI model for the past two years.
Dust CEO Gabriel Hubert identifies a fundamental flaw: "Experience doesn't accumulate, context isn't shared, and no one can pick up where the AI left off."
"This is what we call multiplayer AI, and this is what we're building." —Gabriel Hubert, Dust CEO
In practice, this means AI and employees work together in a shared workspace where notifications, projects, and governance are fully integrated. Results from one agent are visible and reusable by the team and other agents.
This approach aligns with Salesforce's Agentforce and Microsoft's Agent 365, but Dust differentiates by not being tied to any single SaaS platform. It operates at the workspace layer, not as an assistant within a specific application.
What Sequoia saw: zero churn and 70% weekly active usage
Sequoia's partner stated: "Zero churn and 70 per cent weekly active usage tell you this isn't experimental anymore."
In enterprise SaaS, normal churn is 5-10% per year, and 70% WAU is typical for core productivity tools like early Slack or Notion. This indicates Dust is not a trial tool but one that users rely on daily and don't cancel.
Compared to other AI agent startups, which often see 30-40% churn and single-digit WAU, Dust's metrics explain Sequoia's willingness to lead the round.
Founding team background
Hubert and co-founder Stanislas Polu are not newcomers. They met at Stanford in 2007, co-founded TOTEMS (acquired by Stripe in 2014), and worked at Stripe for five years. Polu later became a research engineer at OpenAI, working on theorem proving (Lean's GPT-f). Hubert served as CPO at Alan, a French health insurance company.
This combination brings both foundational AI research and enterprise product experience.
Where the money will go
Dust outlined three areas:
- Self-learning agents: AI that improves automatically without manual prompt tuning.
- Collaboration primitives: Treating AI as an equal collaborator, not a tool.
- Governance and orchestration infrastructure: Enterprise-grade permissions, auditing, and agent orchestration.
The third point is key to why Snowflake and Datadog invested. As agent counts grow from dozens to thousands, infrastructure for data access control, change tracking, and rollback becomes critical. Snowflake manages data, Datadog handles observability—both are essential foundations for Dust's multiplayer AI.
A French company betting on European AI sovereignty
Dust is based in Paris. Over the past six months, Sequoia has increased its European bets, including Mistral and Cohere. European enterprises are increasingly reluctant to send internal data to US clouds from OpenAI or Anthropic, making "sovereign AI" a compelling pitch. Dust's workspace layer can integrate European models like Mistral or US models like Claude and GPT, letting customers choose.
With this round, Dust's next challenge is not just competing with Microsoft and Salesforce for enterprise customers, but also answering a bigger question: as half of a company's work is done by AI, will the workspace product itself be reinvented?
Zero churn and 70% weekly active usage suggest customers are already voting with their feet. The next 12 months will show whether Dust can scale those numbers by an order of magnitude.
Sources: Dust raises $40M Series B to build the "multiplayer" operating system for enterprise AI (Tech.eu); CocoLoop; Exclusive: Agentic AI startup Dust raises $40M (Axios)