Summary

• London-based DIG Ventures has closed Fund III at $120 million to back roughly 30 pre-seed and seed companies.
• The firm is concentrating on enterprise AI infrastructure, including the data, identity, compliance, and orchestration layers used by other software.
• As generative applications become easier to build, DIG is betting that more durable value will sit in the infrastructure businesses depend on underneath them.

The application layer of artificial intelligence is becoming faster and cheaper to build, while DIG Ventures is betting that the infrastructure underneath those applications will become more valuable as enterprises depend on it.

The London-based venture firm has closed a $120 million third fund aimed at pre-seed and seed companies building enterprise AI and cloud infrastructure. Current reporting says the fund expects to back around 30 businesses and plans to lead most of the rounds in which it invests.

DIG Ventures was founded in 2018 by Ross Mason, who previously founded enterprise integration company MuleSoft. The current partnership combines that operating background with investment and commercial experience from Rytis Vitkauskas and Melissa Klinger.

The thesis concentrates on what DIG describes as control points inside the enterprise technology stack, including data, identity, compliance, and orchestration. Those layers may be less visible than a customer-facing AI application, but other software can become dependent on them once they sit inside production workflows.

Cheaper applications change where investors look

Generative models have lowered the cost of producing interfaces, code, prototypes, and increasingly complete software products, allowing small teams to test ideas much faster than during earlier software cycles.

That acceleration also increases competition because features that once required large engineering teams can be reproduced quickly, while model providers continue absorbing capabilities that previously supported standalone software businesses.

Infrastructure operates under a different set of constraints. An enterprise still needs reliable ways to manage data, authenticate users and machines, control permissions, observe systems, connect applications, meet compliance requirements, and coordinate increasingly autonomous software.

Those functions can become more important as agents spread because a system capable of acting across several applications requires a clear answer to which identity it uses, what information it may reach, which tools it can invoke, and how its activity is monitored.

An application can be replaced relatively quickly if customers discover a better interface, while moving the identity, data, or orchestration layer beneath many applications can create a much larger operational problem. That difference is the basis of DIG’s argument that infrastructure can offer more durable positions than some software sitting directly on top of foundation models.

Europe still has to turn engineering into scale

The strategy also intersects with a persistent weakness in Europe’s technology economy, where the region produces strong technical talent and research but has created fewer globally dominant enterprise infrastructure companies than the United States.

Mason’s own history forms part of DIG’s pitch because MuleSoft was founded from Europe before expanding in the US, later listing publicly and being acquired by Salesforce for $6.5 billion. DIG presents enterprise sales, product development, fundraising, hiring, and Europe-to-US expansion as areas where its operating experience can support portfolio companies.

The fund’s current portfolio already includes businesses operating in layers such as observability, data infrastructure, decision systems, and AI orchestration, giving the new vehicle a more specific focus than a broad fund simply labelled around artificial intelligence.

Specialist infrastructure companies still face formidable competition from hyperscalers and large software platforms, because AWS, Microsoft, Google, Databricks, Snowflake, and model providers can add adjacent functions to products customers already use. Open-source software can also weaken a commercial advantage quickly when technical standards remain unsettled.

The investment case therefore depends on whether new AI systems create dependencies that are difficult for one large platform to solve completely. Multi-model environments, agent identity, evaluation, permissions, data movement, and regulatory control all create technical problems that may support independent suppliers if buyers want flexibility across vendors.

Fund III is about 20% larger than DIG’s second fund, according to current reporting, and investments are expected to concentrate in the early stages where technical teams may still be proving the product rather than scaling a mature sales operation.

That stage carries greater technical risk because infrastructure businesses can require substantial engineering before commercial traction becomes obvious, while their earliest customers are often specialist development teams rather than mainstream corporate buyers.

DIG’s $120 million fund is therefore a concentrated wager on the plumbing beneath enterprise AI rather than another attempt to predict which visible application will become the category winner. Its returns will depend on whether today’s emerging data, identity, compliance, and orchestration problems become lasting infrastructure markets after the current generation of models and applications has changed again.