Research-stage seeds are underwritten on team, a falsifiable technical thesis and cost-curve timing. Radical Numerics' record shows a $50M seed. Define the experiment that would prove you wrong, and fund that.
Key takeaways
- Pre-product seeds are underwritten on team credentials and a falsifiable technical thesis.
- Define the experiment that would disprove your thesis and fund that, not a general research programme.
- Large seeds create a Series A expectation gap — write down what the next round will need before you spend.
Standard seed advice assumes something to show: a prototype, early users, a waiting list. A category of seed rounds funds none of that. They fund a research team and a hypothesis.
The record here belongs to Eric Nguyen, founder of Radical Numerics (Stanford, California).
| | | |---|---| | Founder | Eric Nguyen | | Company | Radical Numerics (Stanford, California) | | Total raised | $50M | | Latest round | Seed — $50M | | Round date | June 2026 | | Named participants on record | Emergence Capital, Obvious Ventures, Triatomic Capital, Factory |
An entire company''s lifetime capital, raised at seed, before there is a product to describe.
1. Prior technical work. Not credentials in the abstract, but specific published or shipped results the investor can verify. 2. A falsifiable thesis. Stated so precisely that it could be shown wrong within eighteen months. 3. Timing. Something changed recently — a cost curve, an available capability — that makes the attempt reasonable now and not three years ago.
Vagueness fails here faster than anywhere else in venture. A thesis that cannot be disproven cannot be funded seriously.
The most useful framing for a research-stage plan is inversion: what experiment, if it failed, would tell you the thesis is wrong? Cost that experiment. Fund it, plus the team required to run it, plus a buffer.
This is more persuasive than a roadmap, and it produces a defensible answer to the only question that matters in month eighteen: did the thing work?
A fifty-million-dollar seed creates a Series A expectation set by the size of the seed, not by the stage. Before spending, write down the evidence a Series A investor will want and the date by which it must exist. Revisit it quarterly.
1. State the thesis so precisely that it could be falsified. 2. Design the experiment that would disprove it. 3. Cost that experiment plus the team plus a buffer. 4. Write the Series A evidence requirements before spending. 5. Review the thesis quarterly and record what changed.
Amounts, stages, dates and named participants are documented. Valuation, terms and board composition are not.
Frequently asked questions
- What do investors evaluate in a pre-product seed?
- The team's prior technical work, a specific and falsifiable thesis, and why the timing works now — usually a cost curve or capability threshold that recently moved.
- What is the risk of a $50M seed?
- The Series A expectation gap. A seed that size implies Series A-level evidence next, so decide the target evidence before you start spending.
- Where do these figures come from?
- Structured founder funding records: total raised, round stage, round amount, round date and named participants.
