Deeptech EuropeDeeptech Europe

Sec 03 / Method

Find the constraint. Move it. Repeat.

No stage-gate playbook. Seven principles that decide where the weight goes in a hard-technology company.

Reference sequence

  1. 01The physics works.The yield doesn't.
  2. 02The yield works.The supplier can't scale.
  3. 03Supply works.Qualification takes 18 months.
  4. 04Qualification works.The first market is wrong.
  5. 05Demand arrives.Production becomes the problem.

Operating principles

01

What has to move for everything else to move?

Find the rate-limiting step

Do not optimise everything. Find what is limiting the system now, and put the weight there.

  • physics
  • architecture
  • yield
  • a missing engineer
  • a test rig
  • tooling
  • certification
  • a supplier
  • a customer
  • production capacity
  • working capital
02

Kill the fatal assumptions early.

Kill the fatal assumptions early

Deeptech companies are stacks of assumptions. Attack the unknown capable of killing the company first. Then the next one.

  • Will it scale?
  • Will it survive real conditions?
  • Can tolerance be held?
  • Does performance collapse at production geometry?
  • Can the material be sourced?
  • Can it meet the power envelope?
  • Will it pass qualification?
03

The production system is part of the product.

Build the product and the production system together

A lab prototype and a manufacturable product are different objects. Production thinking belongs in engineering early.

  • BOM
  • yield
  • tooling
  • assembly
  • test
  • tolerance
  • supplier capability
  • serviceability
  • unit economics
  • lead times
04

Shorten the distance between question and answer.

Collapse the learning loop

Not 'move faster'. Reduce the time between assumption and evidence.

  • Can a 12-week learning cycle become 12 days?
  • Can simulation kill three design branches before machining?
  • Can the customer test an ugly prototype now?
  • Can engineering and manufacturing run concurrently?
05

Reality is the highest-bandwidth test rig.

Test where the thing has to live

A bench test proves something. It does not prove the product. Real environment. Real operator. Real integration. Real failure modes.

  • factory floor
  • substation
  • hospital
  • vehicle
  • data centre
  • offshore platform
  • production line
  • field deployment
06

Know what you cannot afford to depend on.

Own the critical path

Do not vertically integrate everything. Understand which external dependency sits on the critical path — and treat it as an engineering decision, not a procurement one.

  • performance
  • iteration speed
  • quality
  • IP
  • supply security
  • unit economics
  • customer experience
07

Don't optimise parts. Engineer the system.

Engineer the system

Hardware, software, power, thermals, manufacturing and economics are one object. Optimised independently, they fail together.

  • hardware
  • software
  • power
  • thermals
  • manufacturing
  • economics

Reality is the highest-bandwidth test rig.

Design → Test → Failure → Revision

Build something difficult.

Build with us How we build