Deep Tech and Hardware Startups
Startups built on scientific or engineering breakthroughs — biotech, robotics, semiconductors, energy, aerospace, advanced materials, and hardware. Different rules apply: longer timelines, higher capital requirements, and validation that can’t be done with a landing page.
How Deep Tech Differs from Software
| Dimension | Software Startup | Deep Tech / Hardware |
|---|---|---|
| MVP timeline | Days to weeks | Months to years |
| MVP cost | $0 - $50K | $500K - $50M+ |
| Iteration speed | Deploy daily | Physical prototyping cycles (weeks/months) |
| Capital needs | Low (bootstrappable) | High (almost always requires funding) |
| Moat | Network effects, distribution | IP, patents, proprietary science, manufacturing |
| PMF signal | Users, retention, revenue | Letters of intent, pilot programs, regulatory approval |
| Team | Mostly engineers | Scientists + engineers + regulatory + manufacturing |
| Failure mode | No market demand | Technical risk + no market demand |
| Exit timeline | 5-10 years | 7-15 years |
What Still Applies
Despite the differences, core frameworks remain valid:
product-market-fit — Still the Only Thing That Matters
The market must exist and want what you’re building. Andreessen’s insight applies with even more force: deep tech founders can spend years building something technically impressive that nobody needs. The market doesn’t care how elegant your science is.
customer-development — Even More Critical
Altman: “For hard tech/biotech, talk to customers first, build minimum viable solutions.” Blank’s “get out of the building” is essential because:
- Customers tell you which technical specifications actually matter
- Regulatory bodies tell you which approvals you’ll need
- Partners tell you about distribution channels you haven’t considered
- Everything you learn early saves years of misdirected R&D
ideation — Deep Tech Ideas Are Often Secrets
Thiel’s “secrets” framework maps perfectly: deep tech founders know something about physics, biology, or engineering that most people don’t. The question is whether that knowledge translates into a viable business, not just a viable paper.
do-things-that-dont-scale — Especially “Pull a Meraki”
PG’s “Pull a Meraki” category was written for hardware: manufacture the first units by hand. Pebble assembled their first watches themselves. This applies to deep tech: build the first prototype manually, learn from it, then figure out manufacturing.
What’s Different
Validation Requires Different MVPs
Software MVPs don’t translate:
| Software MVP | Deep Tech Equivalent |
|---|---|
| Landing page | Letter of intent from potential customer |
| Clickable prototype | Technical demonstration / proof of concept |
| Beta product | Pilot program with design partner |
| Revenue | Pre-orders, government contracts, or LOIs |
The minimum-viable-product is a minimum viable proof — not a minimum product.
Fundraising Is Structured Differently
| Stage | Software | Deep Tech |
|---|---|---|
| Pre-seed | $100K-$500K (friends, angels) | $500K-$2M (specialized angels, grants) |
| Seed | $1-3M | $2-10M |
| Series A | $5-15M (needs traction) | $10-30M (needs technical milestone) |
| Series B+ | Revenue-driven | Often still pre-revenue; milestone-driven |
Deep tech investors evaluate differently:
- Technical risk: Can you actually build this? (Show progress, not just theory)
- Team pedigree: PhD-level expertise matters more than in software
- IP position: Patents, trade secrets, regulatory moats
- Market timing: Is the technology ready? Are customers ready?
- Government funding: Grants (NSF, DARPA, EU Horizon) can fund early R&D without dilution
The “Valley of Death”
The most dangerous phase: between “the science works in the lab” and “we can manufacture at scale.” This gap kills more deep tech startups than market risk:
Lab proof → Prototype → Pilot → Manufacturing → Scale
↑ ↑
Grants/Angels fund this VC funds this
THE VALLEY OF DEATH
(often unfunded)
Solutions:
- Government grants and non-dilutive funding for the middle phase
- Strategic corporate partnerships (they fund pilots in exchange for first access)
- SPAC or specialized deep tech funds (Lux Capital, Breakthrough Energy, DCVC)
- Revenue from adjacent applications while developing the core technology
Regulatory Is a Feature, Not a Bug
For biotech, medical devices, energy, and aerospace:
- Regulatory approval is a moat — once you have FDA/EMA/FAA approval, competitors face years of catch-up
- Start the regulatory process early — it runs in parallel with product development, not after
- Hire regulatory expertise from day one, not when you “need” it
- The companies that treat regulatory as an afterthought are the ones that die in the Valley of Death
Team Composition Differs
| Role | Why It’s Critical in Deep Tech |
|---|---|
| CTO/CSO | Must have deep domain expertise (PhD-level); can’t hire this in later |
| Regulatory lead | Navigating FDA/EPA/FAA is a specialized skill; mistakes cost years |
| Manufacturing/ops | Bridging lab to production is its own expertise |
| Business lead | Translating technical capabilities into customer value |
| IP counsel | Patents filed early and correctly are the primary moat |
Deep Tech in the AI Era
AI changes deep tech startups too:
- AI accelerates scientific discovery (protein folding, materials science, drug discovery)
- Simulation reduces physical prototyping cycles
- AI-powered lab automation increases experiment throughput
- But: physical reality still can’t be compressed — atoms are slower than bits
See Also
- ideation
- product-market-fit
- fundraising
- minimum-viable-product
- competitive-strategy
- scaling
- legal-foundations
- ai-era-entrepreneurship