AI Infrastructure Investments in Dutch Scale-Ups

Governance, IP, data and investor rights in AI compute, model and infrastructure-heavy companies

AI infrastructure investments raise different legal issues from ordinary software investments. Dutch scale-ups building or using AI infrastructure may rely on compute capacity, model access, training data, hardware supply, cloud vendors, open-source components, strategic partnerships and complex IP arrangements. Investors need to understand not only the product, but also the dependencies behind it.

For Dutch companies, the legal work sits in the connection between investment documentation, governance, commercial contracts, IP ownership, data rights and operational resilience.

This article is part of Viotta’s Insights on AI and Dutch transactions and Dutch VC and PE investment implementation.

AI infrastructure is dependency-heavy

An AI infrastructure company may depend on cloud providers, GPU access, model providers, data suppliers, research partners, open-source tools, hardware vendors or technical contractors. These dependencies affect value and risk.

Investors should ask whether the company owns what it claims to own, whether it has reliable access to compute, whether key vendor contracts are transferable or terminable, and whether the business can scale without legal bottlenecks.

This is not only technical diligence. It is legal and commercial diligence.

IP ownership and model rights

AI infrastructure companies often involve layered IP. There may be proprietary code, licensed models, open-source components, training pipelines, datasets, fine-tuned models, customer-specific outputs and know-how.

The investment documents should reflect the diligence findings. If IP ownership is not clean, investors may require remediation covenants, founder assignments, contractor confirmations or specific warranties.

For Dutch scale-ups, this is particularly relevant where academic institutions, freelancers, former employers or strategic partners contributed to development.

Data rights and customer restrictions

AI infrastructure may depend on access to data. The company must understand whether it can use, train on, process, store, combine or commercialise that data.

Customer contracts may restrict model training. Privacy rules may restrict personal data processing. Strategic partners may restrict secondary use. Data licences may not allow the intended product roadmap.

Investors should not treat data access as an operational detail. It can be central to the investment case.

Governance and reserved matters

If AI infrastructure is core to the business, investors may ask for governance rights around major vendor contracts, model licensing, compute commitments, data partnerships, regulated deployments or strategic IP transfers.

These can be included as reserved matters, reporting obligations or board approval items. The drafting should be targeted. Routine engineering decisions should not require investor consent, but material infrastructure commitments may.

For Dutch BV companies, these rights should be aligned with the shareholders’ agreement, board rules and articles where needed.

Commercial contract risk

AI infrastructure investments often involve significant commercial contract risk. Cloud commitments, reseller arrangements, enterprise customer agreements, data processing terms, SLAs, indemnities and liability caps can materially affect the company.

Investors should look at whether the company has promised more to customers than it can deliver, whether vendor contracts support customer obligations and whether liability exposure is aligned across the chain.

Mismatch between supplier terms and customer commitments can become a serious value risk.

Practical conclusion

AI infrastructure investments require legal review beyond standard VC documentation. Investors and founders should focus on dependencies: compute, vendors, IP, data rights, customer contracts, model access and governance.

For Dutch scale-ups, these issues should be translated into investment documentation, warranties, reserved matters, information rights and remediation actions before closing.

FAQ

Why are AI infrastructure investments legally different?
Because value often depends on external compute, vendor contracts, model rights, data access, IP ownership and customer commitments.

Should AI infrastructure issues be included in reserved matters?
Material commitments may justify approval rights, especially where they affect core technology, data, IP or financial exposure.

What should investors review first?
IP ownership, data rights, compute contracts, vendor dependencies, customer commitments, open-source use and governance controls.

About Dirk de Waard

Dirk de Waard is a Dutch corporate and M&A lawyer, partner at Venture Lawyers in Amsterdam, and advises founders, investors, scale-ups and portfolio companies on Dutch VC, PE, AI diligence, investment documentation, governance and commercial contracts.

Investing in Dutch AI infrastructure companies?

AI infrastructure investments require a clear view of technical dependencies and legal rights. Compute, data, IP, vendor contracts and customer commitments should be reflected in the investment documentation.

Dirk de Waard advises investors and Dutch scale-ups on AI infrastructure investments and Dutch transaction implementation. Contact dirk.dewaard@viottalaw.com to assess the legal structure of an AI infrastructure investment.