MuleSoft's founder raises €106M to back Europe's AI foundations: DIG Ventures bets the plumbing wins
DIG Ventures, the London venture firm founded by MuleSoft's Ross Mason, closed a €106 million ($120 million) Fund III on Thursday to back 30 pre-seed and seed startups building the data, identity, compliance, and orchestration layers underneath enterprise AI.
While the AI industry’s spotlight stays fixed on foundation models and billion-dollar training runs, the new money is moving one layer down. DIG Ventures announced on Thursday the close of its third fund at €106 million ($120 million), dedicated to the enterprise and cloud infrastructure companies “underpinning the future of AI” in Europe.
The firm started in 2018 as Ross Mason’s family office. Mason founded enterprise integration company MuleSoft, which Salesforce acquired for $6.5 billion. DIG has since grown into an operator-led fund whose pitch to founders is that it has done the job it now funds: building enterprise software, selling it globally, and getting it into the hands of US customers.
Founders funding the foundations#
Fund III’s backer list is notable for who’s writing the checks. Alongside institutional limited partners — Horsley Bridge, Sofina, Granite, and a leading US university endowment — the fund signed entrepreneurial LPs including the founders of Slack, Datadog, Nord Security, Cast AI, Supercell, and Dash0.
That’s the thesis rendered as a cap table: the people who built the last generation of developer and enterprise infrastructure are now financing the people who will build the AI-era plumbing. The fund says it will lead most of the rounds it enters and has already begun deploying capital.
The thesis#
Mason’s bet is explicit and runs against the model-hype grain: “As development accelerates and differentiation erodes, the truly scalable, defensible companies will be the ones building the foundations on which wider enterprise software depends. These are the companies whose advantages compound as the cost of building software falls.”
Fund III will target what the firm calls the critical AI control points: the data, identity, compliance, and orchestration layers. In practice, that’s the unglamorous middle of the stack — where agents get permissioned, data gets governed, and workloads get routed — where the agentic AI boom is currently creating its messiest problems.
Mason also named Europe’s structural gap: “In Europe, we are seeing huge waves of innovation on this frontier – the continent is not short of technical talent. But the challenge is turning technical advantage into global distribution.” The fund’s operator-led partner team — Mason, Melissa Klinger, formerly MuleSoft’s UK sales lead, and Rytis Vitkauskas, founder of YPlan and a former Lightspeed partner — is built to carry portfolio companies “from day zero through to their first US customers.”
A track record that raised the fund#
Fund III follows Fund II, the firm’s first institutional fund, which Carta benchmarking data ranks in the top decile globally for its vintage and size. The portfolio gives LPs something concrete to price: Dash0, the AI-native observability platform, became a unicorn after its €94 million Series B in March 2026; Taktile, an AI decisioning platform, raised a $110 million Series C led by Goldman Sachs; CUBE’s regtech is used by more than 30 global financial institutions; AI HR platform Jack & Jill raised a €34.68 million Series A earlier this month; and Nexos.ai is building AI orchestration from the Nord Security founders’ playbook.
Exits include insurance infrastructure company Flock (acquired by Admiral), data infrastructure platform Tower (acquired by MotherDuck), and workload identity company Cofide (acquired by Keyfactor). And the distribution story Mason sells has receipts: according to the firm’s internal data, 80% of Fund II companies went on to raise additional institutional capital within two years, and more than 90% entered the US market commercially within a year.
What to watch#
Thirty pre-seed and seed bets across data, identity, compliance, and orchestration is a small enough portfolio that each check is a signal of where the operator set sees the AI stack’s real bottlenecks. If DIG is right, the most valuable European AI startups of the next few years won’t train models at all — they’ll be the companies that make models deployable, governable, and billable inside real enterprises. The first Fund III portfolio companies should tell us exactly where Mason thinks the plumbing is leaking.