Next $1T Thesis
A thinking partner for AI startup ideas that refuses to flatter you. Names the tier, scores against four tests, and mutates weak ideas before killing them.
- Claude Code
- Open Source
- Skill
the single most important idea in it is the sign flip.
a copilot company treats every new model release as an existential threat, because the labs eat wrappers. an outcome company treats every new model release as a tailwind: cheaper cogs, same price, fatter margin. if model progress hurts your business, that is close to fatal, and the skill asks this about every idea.
it names a tier for everything. t0 is the copilot trap and its ceiling is an acqui-hire. t1 displaces an existing service at a discount and tops out in the billions. t2 serves the un-served at a fraction of legacy price, and that is where the trillion dollar company actually lives. not the ai cravath. the ai walmart of professional services.
hard rules: no flattery, ever. name the tier with a confidence number. reject "businesses" and "users" as answers, because vague inputs produce vague verdicts. separate the buyer from the user, since the person writing the check is never the person using it. mutate before killing.
it also argues with its own source. the reference material has a section on what the thesis gets wrong, because a framework you cannot criticise is a religion.
independent interpretation of sequoia's services thesis. not affiliated with them, and not investment advice.