Founder Mode: Andy Lapsa, Founder & CEO, Stoke Space

Summary

This The Social Radars YC offsite episode has Andy Lapsa explain [[Stoke|Stoke Space]], a vertically integrated launch service provider building fully and rapidly reusable rockets. The episode extends Reusable Rocket Economics by isolating [[SecondStageReuseConstraint|second-stage reuse]] as both a cost driver and flight-rate limiter after SpaceX proved first-stage reuse. It also adds a hard-tech version of Founder Mode: early survival requires doing whatever is needed, but a scaling company should distribute ownership and urgency across the organization.

Key Claims

  • [[Stoke|Stoke Space]] is building rockets intended to go to space, move around in space, and return to the ground.
  • Andy Lapsa frames Stoke’s long-term mission as making transportation to, through, and from space a natural extension of existing mobility systems.
  • Satellite delivery is the main current application, but Andy argues that lower price and better availability could make space another node in a broader logistics network.
  • Stoke officially started at the end of 2019, spent its first year doing diligence and searching for a technical solution, joined Y Combinator Winter 2021, closed its seed round, and then ramped.
  • Andy says pre-SpaceX launch was fully expendable and compares that to an airplane being thrown away after every flight.
  • [[Falcon9|SpaceX first-stage reuse]] is presented as transformational, but Andy says current partial-reuse systems still discard the second stage on every flight.
  • The discarded second stage creates direct cost because it must be rebuilt and also creates a production-rate bottleneck through engines, certifications, tests, and manufacturing capacity.
  • Andy identifies two major launch cost drivers: vehicle cost and infrastructure cost. Reuse amortizes the vehicle, while higher flight frequency amortizes factories, test sites, launch complexes, and labor.
  • He says SpaceX’s first-stage reuse helped the company shift factory capacity toward second stages and grow to roughly 150 flights per year, while noting that this is still small compared with ordinary transportation.
  • The source says Starship is intended to be fully reusable but had not achieved that state at the time of the episode.
  • Stoke had grown to a little north of 260 people, hiring both specialized rocket people and outsiders with transferable skills.
  • Andy describes difficult fundraising conditions around Winter 2021 and again in 2023 and 2024 for hardware growth companies, especially pre-revenue companies.
  • Stoke tried to avoid looking like a marketing-only hardware company by building an early rocket test stand in a shipping container in a co-founder’s yard.
  • Andy defines early founder mode as survival urgency and doing whatever has to be done when nobody else is available.
  • At Stoke’s current scale, he says founder mode should become organizational ownership and urgency; if he personally has to do something, that can mean the organization has not yet been set up to succeed without him.
  • He says founder intervention still happens because growing companies have unfilled holes, and adjacent opportunities require judgment about whether to pursue them or protect focus.

Key Quotes

“PowerPoint rocket” - Andy’s contrast for a startup that markets a rocket before proving serious hardware execution.

“go all in” - Andy’s rule after the company decides an adjacent opportunity is worth pursuing.

“they’re not there yet” - Andy’s source-time caveat about Starship and full reusability.

Connections

Contradictions

  • No direct contradiction found. The source extends the earlier Stoke Space mention in the Paul Graham founder-mode episode by grounding the company name, founder identity, YC batch timing, rocket thesis, and operating details. Its Starship and flight-rate claims are preserved as Andy’s source-time account from 2025-09-29 rather than current independent launch-status validation.

Source Notes

  • Ingested from the TSR-YCOffsite-AndyLapsa-v1-Audio Markdown export in the podcastatlas episode corpus.