On 1 July, the Dutch Ministry of Defence opened the MIND Tech Centre on the TU Delft campus. The acronym stands for Military Innovation by Doing. We like the plainness of that name. It points to a truth founders already know: capability improves when users, engineers and buyers work on the same problem at the same time.

Our thesis is straightforward: if the Netherlands wants useful defense technology faster, we need more places where founders can test with real users, learn from failed assumptions and return the next week with a better system.

The point

The center brings together Defence, TU Delft, startups, SMEs and the Province of South Holland. The fields named in the announcement are familiar to any dual-use founder: drones, sensors, space technology, IT, smart materials and digital systems. The value is not the list. The value is the loop between those fields and the people who may one day deploy them.

South Holland is committing €1 million, partly with European Union funding. That is not enough to build an ecosystem by itself. But it is enough to create a shared place where a founder can move from conversation to experiment faster than a normal procurement path allows.

Evidence we see in the field

A drone team can learn more from one afternoon with an operator than from three months of desk research. Battery swaps, weather, jamming, launch procedure, transport cases, repair tools — these details decide adoption. The founder is the protagonist because the founder has to absorb all of that and still make the system simpler.

• Operators reveal constraints that specifications miss

• Universities bring test talent and technical depth

• SMEs know how to build repeatable hardware

• Founders turn feedback into product decisions

Why speed changed

State Secretary Derk Boswijk linked the center to the pace of adaptation seen in Ukraine, especially around drones. That reference matters. Drone development has made clear that the unit of progress is no longer a five-year requirements cycle. It is a build-test-learn rhythm measured in weeks when the team is close enough to the user.

The hard part is not only building a better sensor or airframe. It is learning which version a user can carry, maintain, update and trust. That learning requires contact. A center like MIND gives that contact a physical address.

Example: dual-use is a translation problem

Many technologies that matter to defense started with civilian language: autonomy, mapping, computer vision, resilient communications, lightweight materials. The founder’s job is to translate without losing the product. Defense users need reliability under stress; civilian markets often teach cost, usability and scale. The best teams learn from both.

Example: procurement begins before procurement

For seed-stage companies, the formal procurement moment often arrives too late. The important work starts earlier: understanding who owns the problem, what proof is credible, which field test matters, and what cannot be said publicly. Places like Delft can help founders ask those questions before they have spent a year building the wrong thing.

Counterpoint

Innovation centers can become theatre. A ribbon, a demo day, a few panels, then little changes for the founder trying to get test access or a first paid pilot. The test for MIND will be practical: how many teams get closer to a real user, a real technical constraint or a real procurement path because the center exists?

• What we are doing at Tenzing: we look for founders who want the hard feedback early, not after the product has become expensive to change.

• We ask which operator has touched the system, which test failed, what timeline the customer actually works on, and where a Lead Partner can help remove one concrete blocker.

• We do not need certainty at seed. We need a founder who can learn quickly from the right room.

What this asks of founders

Delft matters if it becomes a place where founders leave with sharper constraints than they arrived with. The work is not to perform innovation. The work is to sit close enough to the user that the next version becomes more useful.