Hard tech and biotech startups: why they're easier than you think

Executive overview

Hard tech companies face high technical risk — it may not be clear if the product can even be built. Yet founders systematically underestimate them, assuming they're harder to start than ordinary software companies.

The counterintuitive truth: the early steps of a hard tech company are uniquely hard, but everything else — press, recruitment, investor interest — is far easier than for an ordinary app. The bottleneck is making enough credible early progress to unlock successive funding rounds.

Staged de-risking, not a single moonshot raise, is the defining skill of a hard tech founder.

What makes a company "hard tech"

• Two criteria: (1) will take significant time and money to build the first product; (2) even with those resources, success is not guaranteed
• Distinct from market risk — the question is whether you can build it, not whether people will want it
• Applies to physical and non-physical products alike
• YC accepts hard tech applications at ~10x the rate of other categories

Why hard tech is easier than it looks

• Launching a mobile app is trivial; turning it into a big company is very hard
• Hard tech is the reverse: building is hard, but recruiting, press, and fundraising are easier
• Talented people are drawn to ambitious problems — a recruiting advantage, not a handicap
• Investor demand for ambitious ideas is at an all-time high; more capital is available than most founders realise

The heavy MVP problem: how seven YC companies solved it

Most hard tech companies need millions to build their first product. The solution: make maximum credible progress before spending that money.

1. Boom (supersonic jet) — assembled top advisors, built computer simulations, made a plastic model, used it to get signed airline interest, then raised
2. Solugen (synthetic biology hydrogen peroxide) — proved the industrial process with a single beaker; scaled progressively to a full plant
3. AirX (medical device) — sidestepped years of FDA approval by launching a basic version using an already-approved device plus software
4. Notable Labs (cancer drugs) — generated revenue and data by providing tumour-screening services to pharma companies before developing their own drugs
5. Astronis (telecom satellites) — built and launched a non-commercial test satellite in under three months for under $50k to prove orbital capability
6. Ginkgo Bioworks (genetic engineering) — closed contracts with large companies to create organisms before making them; used those contracts to raise the capital needed
7. Cruise (self-driving cars) — built an MVP self-driving car in a garage in three months during YC; acquired by GM for $1B

Common pattern: sell it, simulate it, or prove the core mechanism before attempting the full build.

Proving demand before you have a product

• Pre-sales are the strongest signal — Kickstarter-style campaigns work if regulation permits
• Pre-sales are illegal for FDA-regulated products; use LOIs instead
• Letter of Intent (LOI): a non-binding contract to buy when the product is ready
• Non-binding means low commitment for the customer, but signing one signals genuine urgency — polite interest won't produce a signature
• A specific LOI (price, quantity, timeline, specs) is far more valuable than a vague one
• A signed LOI doubles as a product roadmap: it tells you exactly what to build

LOIs as a sales rehearsal

• Getting an LOI forces you to navigate the real enterprise sales process early
• Reveals who the actual decision-maker is versus who the end user is
• Surfaces competing stakeholder incentives before they derail a real sale
• Difficulty getting an LOI is itself useful signal — if customers won't sign, the pain point may not be critical enough

Fundraising strategy for hard tech

• Avoid the single-big-ask plan ("I need $50M, I'll pitch until someone gives it to me") — no investor will fund an unproven idea at that scale
• Staged fundraising: split the total into five or more discrete rounds, each unlocking the next milestone
• Example ladder: no capital → $200k → $1M → $4M → $15M Series A, each tied to specific proof points
• The skill is making each step as small as possible — smaller steps mean more achievable milestones
• The most important rule: never attempt a raise before you've earned it with concrete progress
• Hard tech companies that apply YC's general advice (product, users, iteration) find it mostly applies — the differences are fewer than expected