Blog · Personal ·
Hard Tech Is Hard, But the Studio Model Might Be the Right Vehicle
Hard tech founders applying SaaS logic face brutal feedback loops and capital intensity that punish speed. The studio model provides shared infrastructure and p
Hard Tech Is Hard, But the Studio Model Might Be the Right Vehicle for It
Ten years. That's how long it took Impinj to go from a semiconductor idea backed by Madrona to an actual IPO. Ten years of RFID chips not working the way they were supposed to, of manufacturing partners falling through, of investors who'd put money in expecting a five-year exit and instead got a decade-long lesson in patience. It worked out. Impinj is now the backbone of a huge chunk of the world's inventory tracking, and Madrona's bet looks, in hindsight, obvious. Nothing about it was obvious at the time. Ask anyone who lived through the middle years, the bit where the tech almost worked and the market wasn't ready and the burn rate didn't care about either of those facts.
I bring this up because founders keep applying SaaS logic to hard tech problems, and it's a bit like bringing a butter knife to a bar fight. Technically a tool. Wrong tool.
SaaS has a rhythm everyone's internalised now: raise a seed round, get to product-market fit in twelve to eighteen months, raise a Series A on the back of some encouraging MRR chart, repeat until you're either dead or a unicorn. Hard tech doesn't run on that clock. You can't A/B test your way to a working fusion reactor. You can't ship an MVP of a satellite. The feedback loops are long, sometimes brutally long, and the capital required just to find out if the physics works (before you even get to "will anyone buy this") is enormous. That's the bit SaaS investors and first-time hard tech founders both underestimate, usually at the same dinner.
Why hard tech eats solo founders alive
Here's the uncomfortable maths. A solo founder building software can get to a testable product with a laptop, a weekend, and mild sleep deprivation. A solo founder building, say, a mineral exploration AI system needs geologists, drone hardware, satellite data contracts, and enough runway to survive eighteen months of "inconclusive results" before the model starts finding anything real. That's not a founder problem. That's a systems problem. One person, however brilliant, can't be the fundraiser, the domain expert, the ops lead, and the person debugging sensor calibration at 2am. Something breaks. Usually it's the founder.
I think this is where a lot of well-meaning advice fails hard tech founders specifically. The standard line is "just hire A players and delegate." Fine advice for a marketplace app. Less useful when hiring a single senior geophysicist takes four months and a relocation package. The capital intensity and the long feedback loops compound each other: you need more money to survive longer, and you need to survive longer because the feedback loops won't close any faster no matter how good your team is. Rock formations don't care about your Series A timeline.
What a studio actually does here
This is roughly where the studio model earns its keep, and I say that as someone who's watched it work from the inside rather than someone selling you a deck about it. A studio (the good ones, anyway) provides shared infrastructure: legal, finance, recruiting, some of the operational scaffolding that would otherwise eat the first year of a hard tech founder's life. It de-risks the earliest capital, because the studio's already absorbed some of the cost of testing whether an idea is worth funding at all, before a founder has to go begging external investors for a bet nobody understands yet. And it embeds expertise. Not "we'll mentor you" expertise. Actual people who've built the thing before, sitting in the building, saving you the eighteen months of expensive mistakes they already made on someone else's dime.
Earth AI is a decent example of this in practice, a mineral discovery company using machine learning to find deposits that traditional exploration methods miss (we've covered their trajectory in more detail elsewhere, worth a read if you want the specifics). It's a non-obvious bet. Mining is not a sexy sector for venture capital, and "AI finds copper" doesn't have the same ring as "AI finds your next customer." But it's exactly the kind of long-horizon, capital-intensive, deep-domain problem that punishes the standard seed-to-Series-A sprint and rewards patient, structured backing instead.
The honest downsides
Now, I'd be lying if I said this was a free lunch. Studio equity isn't cheap. Founders give up more control, earlier, than they would in a traditional raise, and for some people that's a dealbreaker, full stop. You're also trading founder autonomy for infrastructure, which is fine until you're three years in and you resent someone else's fingerprints on decisions you used to make alone in your own head. Slower control. Real dilution. Not everyone should take that trade.
But here's the twist nobody likes admitting: total founder control, in hard tech, is often just a slower path to running out of money alone. The romantic version of the solo founder grinding it out for a decade works great in the Impinj story, mostly because Madrona was there the whole time absorbing the risk that would've sunk an individual. Somebody was patient. It just wasn't the founder by himself.