And why our 2023 regrant to Timaeus was goated
We think the for-profit funding ecosystem has some cool properties. Different funders naturally come in at different stages, from angel investors who are good at spotting early opportunities to large funds who might invest in late-stage startups to public market traders. Funders race to back promising projects. And the more they get right, the more they’re rewarded with more money to invest in the future.
It would be great if more of those properties existed in nonprofit funding as well. We think that the flow from “person makes great early donation” → “person has more influence” is too weak. And as opposed to the VC world, where investors compete against each other to invest in something quickly, nonprofit funding is beset by “funder chicken,” in which a funder waits for other funders to fill an opportunity, each trying to preserve their capital. This makes fundraising a slog.
What if donating to nonprofits worked more like investing in companies? Impact markets have been talked about in EA for a while. At Manifund, we dream of bringing impact markets to the AI safety funding ecosystem. And at the very least, we want to import more for-profit norms.
What does this look like concretely?
Timaeus: a case study in charitable ROI
Timaeus started as a planned 6-month research project on developmental interpretability, and got its first funding in August 2023, getting $143k from Manifund regrantors. They announced their launch in October 2023. Later in the fall, they got a $455,000 grant from SFF.
In 2024, they continued to get funding from Manifund and SFF, and in 2025 they got two larger grants of $2m and $1.5m from Coefficient Giving. Then in June of this year, they announced that they were merging into Resolution, which got $160m from Coefficient Giving — their largest AI safety grant to date.
What if Timaeus had been a startup? Their early investors would have earned extraordinary returns. It’s unclear how much of Resolution’s funding to credit to Timaeus; looking at the founding team, it looks to be roughly half from Timaeus and half from UK AISI. But okay, maybe Geoffrey Irving specifically gets a lot of the credit here. Even conservatively allocating 20% to Timaeus, if you compare a total of $599,700 raised in 2023 to $160m * 20% = $32m in 2026, you get a 53x return.
(Two key assumptions: all donations in a single year are treated as one funding round, and the amount of impact equity sold is constant year over year.)
Multiplied out, our $143k stake via our 2023 regrants would have grown 53x to now be worth $7.6m. This first regrant to Timaeus, alone, would more than pay back the entire Manifund regranting budget to date ($5.2m). Since the Resolution grant is so much larger than most other AIS grants, Timaeus could be the single best 2023 AI safety grant in these terms.
In an impact-market world, our regrantors might have cashed out during the Resolution merger, and reinvested proceeds to fund more early projects.
Donate early, donate fast
What lessons can you take from the VC world for your own donations?
In the for-profit world, investing in early-stage startups is high-risk, but that’s where the highest-return investments come from. Similarly, we think the best donations come from someone with good taste spotting early opportunities.
You can also have a high impact just by getting grants out the door quickly; good seed investors will say yes after one call rather than get beaten to the punch. AI 2040 predicts that the foremost AI company will have 3 months of lead time when they fully automate AI R&D. So helping an AI safety project start 3 months earlier is really valuable!
(This is why we love regranting. Jesse Hoogland, the founder of Timaeus, affirmed that “getting early support from Manifund made a real difference for us. This was the first funding we received for research and meant that we could start months earlier than we otherwise would have.”)
Finally, consider looking at who has made the best returns on their grants when considering where to direct your donations. It’s easy to look at a fund’s track record and think “oh hey they donated to MATS, I like MATS, I’ll support that fund.” We like MATS too! But donation timing can matter a lot more than the donation target. In retrospect, the first donation to MATS was worth orders of magnitude more than a marginal donation to them today.
I'd love to see something that's somewhat similar to SFF's speculation grants, though a bit different:
- Small, fast funders can fund as they do
- if the receiving org then gets a grant from a larger funder, the smaller funder gets their money back.
E.g. get 50k from a small funder with a 1M yearly budget. Then get 500k from CG, 50k of which is used to pay back the small funder.
This seems to move the incentive from "don't fund a project if you think CG will fund them anyway" to the opposite: "fund a project especially if you think CG will fund them anyway"
This seems useful if (I guess) these two conditions are met:
- dollar for dollar, the impact of smaller funders is higher
- smaller funders are faster
(I think Matt Brooks recently mentioned something like this to me, though possibly in a different form)
Yes, this is similar to the design of speculator grantors in SFF (and likely the new Lightcone Commons, too)
I think my ideal design wouldn't even require participation in a specific round or program or system - rather, just retro payback to any particular donor/funder who backed an excellent project.
Hmm Lightcone Commons uses some proportional distribution system I think -- though I didn't quite read it in enough detail to understand it -- which would be similar in spirit but sligtly different from what I meant.
Yeah I think the "ideal design" you mention afterwards is what I had in mind here. I've even read SFF grantmakers (I think in Zvi's reflection from 1-2 years ago) stating they are less likely to fund something with SFF money if they think CG will fund it. Which seems like a particularly bad outcome -- though understandable with the current mechanisms.
It's a bit confusing and not the main emphasis of Lightcone Commons, but I believe that they intend to also incorporate speculation budgets like SFF does. The proportional distribution is the main mechanism, which is what informs how the speculators get repaid.
See the section "What if a project needs funding sooner than 3 months from now"
It seems they're less prescriptive than SFF - the default is to repay but not grow the speculator ("direct grantor") budget?