Summary
There’s a near consensus that EA needs funding diversification but with Open Phil accounting for ~90% of EA funding, that’s just not possible due to some pretty basic math. Organizations and the community would need to make large tradeoffs and this simply isn’t possible/worth it at this time.
Lots of people want funding diversification
It has been two years since the FTX collapse and one thing everyone seems to agree on is that we need more funding diversification. These takes range from off-hand wishes “it sure would be great if funding in EA were more diversified”, to organizations trying to get a certain percentage of their budgets from non-OP sources/saying they want to diversify their funding base[1][2][3][4][5][6][7][8] to Open Philanthropy/Good Ventures themselves wanting to see more funding diversification[9]. Everyone seems to agree; other people should be giving more money to the EA projects.
The Math
Of course, I couldn’t make this post without breaking down the numbers. The math is really quite simple here. This is the best data I could find for EA funding in 2023 (numbers in $USD).
The best numbers I can find for 2023
Open Phil 691mm (source)
SFF 33mm (source)
Ea funds 14mm (source)
ACE 8.3mm (source)
Givewell 318mm (source) and (source)
Nonlinear Network 1.5mm (source)
Polaris Ventures 15mm (estimate) (source)
Other/Individual EA donors 15mm (GWWC donors, etc. based on some rough math from (source). I’m very interested if someone has a better or more accurate figure.
*I would be extremely happy to add/edit any additional numbers/figures, but I don’t expect this to significantly change the end result.
I’m not sure if Givewell and Open Phil is double counting here (since Open Phil gives to Givewell) but I’m going to ignore Givewell as EA funding since a lot of this comes from what many would consider outside of the EA community (many people and philanthropists who wouldn’t consider themselves to be EAs though I’m not sure this is completely fair).
So, according to these numbers, 89% of EA funding comes from Open Phil. This means that the weighted average of the funding from Open Phil for all EA organizations has to be around 89% unless they are getting significant amounts of outside funding. The problem, I think, is that most causes the EA community cares about don’t have a lot of outside support. Who outside of the EA community would fund shrimp, wild animals, or insect welfare? Who outside of the EA community would fund the Centre for Effective Altruism (CEA)? Who outside of the EA community would fund the Forecasting Research Institute (FRI)? Or 80000 Hours? Or Rethink Priorities? I don’t mean to say these organizations don’t have funding outside the community but they mostly don’t. Anyone who would give significant sums to these organizations is probably in the community or adjacent to it. If you are the kind of person to take shrimp welfare seriously or read Rethink Priorities reports, I’d be surprised if you weren’t involved in the EA community in some capacity.
Weighted Average
The weighted average of each organization’s funding that all EA organizations receive has to be equal to the percentage of EA money coming from Open Phil. While this makes some assumptions such as all money going to EA organizations comes from the EA community, I think it’s a decent simplification for the broad ecosystem.
There’s no getting around this fact in general. Specific organizations will be able to achieve funding diversity but the ecosystem as a whole, otherwise read as the majority of organizations won’t be able to do this.
Seeing it as a point about weighted averages also elucidates another fact. Small EA organizations (let’s say budgets on the order of <$500k) can have funding diversification because they don’t (individually) have very much effect on the weighted average but large mid-scale ($500k-5mm) organizations will have a harder time having diverse donors and it is basically impossible to have large EA organizations ($5mm+) funded by a diverse set of donors. For the EA case in particular, I would say that the average small organization will have a hard time because EA seems to be composed of many many small organizations.
Making a lot of money to donate is difficult
I sometimes wonder if the EA movement is exceeding or underperforming expectations and by how much. I think it’s fair to say that the EA movement started sometime around November 2009 but certainly 2009-2012ish feels fair. This means that the EA movement has been around for about 15 years. I think if you were to ask a well-calibrated Toby Ord/Will Macaskill back in 2009 the odds of the movement having billions of dollars committed to it in a few years, they would have put the odds incredibly low. There would be some arguments for it happening; the arguments for effective altruism are rather simple and extremely compelling but the base rate on getting billions of dollars committed to anything is insanely low. To that end, I essentially think the movement got extremely lucky at getting Good Ventures/Open Philanthropy in the first place, especially so early on, and that this is a fairly incredible and fortunate outlier event. Without this, I think it would be extremely unlikely for effective altruism to have anywhere near its current funding or impact levels. My rough calculation here is that Good Ventures/Open Philanthropy multiplied annual EA donations by a factor of ~100. Nothing similar has occurred since and I think there are some obvious reasons that this couldn’t happen again. A decent primer here is that with the exception of Jaan Tallinn (increasing EA wealth by maybe 10%), a few adjacent funders (who fund things the EA community likes but are more adjacent to the community, like Vitalik Buterin) and the FTX catastrophe (where we had some temporary billionaires) there just has never been any kind of big increase to EA wealth and these “exceptions” didn’t multiply the pie, at least not anywhere near the level that GV/OP did. Also, there are only ~100 people/families that can give on the order of Dustin/Cari Moskovitz at this point.
The obvious reason is that making a lot of money is hard. In the United States today, the richest country there has ever been (except for a few tiny countries), the top 10% of earners earn about $170,000 per year in adjusted gross income (AGI, the other kind of AGI). This is pre-tax. If you factor in taxes of about 25% and just make a few simple assumptions of living expenses, etc. you can see how it gets difficult to donate above $50,000, enough to fund maybe half of a current person doing direct work. Of course, the EA community is exceptionally talented. It has high-value market skills to do the highest-paying jobs (quant trading, software engineering, entrepreneurship, etc.). Still, you aren’t going to get much above 100-200k per year on average for highly committed EtGers who are aggressively trying to earn and donate as much as possible. It seems that there are ~10,000 EAs (~2500 if you count only highly committed/engaged) and about 10% of the community does earning-to-give. This means that broadly speaking, $25mm-50mm per year could be donated whereas with Open Phil, the movement is moving nearly $1000mm/year. I don’t want to say $25-50mm is nothing. But I do want to point out that it would be 2.5-5% of the total figure.
And in practice, it seems that effective altruists donate significantly less than this. In 2018, 80% of EAs donated less than 10% of their income and the median EA donated less than $1000/year with a mean of around $10000 so we are nowhere near getting enough money from the broad community.
It is difficult for a movement composed of ~10000 people to move over a billion dollars a year and I would submit the only way this has been possible is due to the funding concentration known as Open Phil.
It doesn’t always feel like it, but EA has been winning. A lot. Given our relatively small scale (in number of people) it’s surreal how much EA has accomplished. This is in no small part due to our support from billionaires, mainly Dustin/Cari, and giving this up would be a mistake.
Solutions
1. Get more funders
This is very very very easy to say but extremely hard to do. “Get people to give you millions of dollars” is incredibly, incredibly difficult. First of all, I want to tip my hat to AIM (formerly Charity Entrepreneurship) here alongside GWWC. A few months ago they launched 5 organizations aimed at addressing this problem and organizations like Giving What We Can and other meta organizations have been working at this. But it is really, really hard. There just aren’t that many people who have a lot of money to fund the EA movement at the scale it is operating and every entity that has money has lots of people/organizations asking for it, let alone the fact that they have their own priorities/things they want to spend the money on.
However, this is the exception. Most people asking for a more diverse funding landscape are mainly just asking/hoping for it. It’s extremely easy to want more diverse EA funding. It is far far far harder to make millions of dollars yourself and then donate it. Outside of the above, I haven’t seen many well-thought-out mechanisms for the EA community to get more diverse funding at a large scale.
2. Spend Less
This doesn’t get talked about often because, well, this would mostly involve people taking pay cuts, having fewer perks, and spending less. Salaries are very sticky. People don’t want to take pay cuts or give up perks/comforts they have gotten used to.
From @ElliotJDavies Consider Earning Less
When the Future Fund was founded in 2022, there was a simultaneous upwards pressure on both ambitiousness and net-earnings in the wider EA community. The pressure to be ambitious resulted in EAs really considering the opportunity cost of key decisions. Meanwhile, the discussions around why EAs should consider ordering food or investing in a new laptop pointed towards a common solution: EAs in direct work earning more.
The funding situation has significantly shifted from then, as has the supply-demand curve for EA jobs. This should put a deflationary pressure on EAs' salaries, but I'd argue we largely haven't seen this effect, likely because people's salaries are "sticky".
Organizations tightening their purses would result in less reliance on OP. If your organization is currently 50% funded by OP and 50% by non-OP. If you spend half as much, you don’t need any money from OP.
But it’s not just salaries that would get scaled back to make this happen. It would be things a lot of EA organizations have gotten used to like catered lunches, discretionary budgets for productivity, offsites, etc. It would generally mean that working in EA organizations would cause one to have a slightly lower quality of life (by working harder and having to do more work yourself that you might be used to outsourcing) and that might also mean that there is less talent in EA on the margin.
Outside of animal welfare organizations, broadly speaking, people working in EA non-profits earn significantly more than those who work in “non-EA” non-profits and expenditures are much higher generally. Before people get too upset with me, and I'll write a bit more below, I think it is broadly worth it and I’m happy to listen to arguments that it is worth it but you can’t deny that it’s true.
3. Splitting up Open Philanthropy into Several Organizations
A couple of years ago, there were calls to democratize EA funding (11). If you split up Open Philanthropy into a few independent entities, you will duplicate a lot of work but you will also have less reliance on Open Phil. That said it’s important to remember that Open Phil is nearly entirely (I think entirely?) funded by Dustin Moskovitz/Cari Tuna’s Good Ventures and they rightfully, desire control over how their funds are spent. This would likely result in a lot of duplicated work.
Broadly speaking, I think this is a bad idea for reasons outlined in this post, primarily because Dustin/Cari wouldn’t do it but this is a legitimate option. It is after all their money.
Another approach that has more merit would be for Open Phil to do a lot more regranting but given that they have moved away from giving funding to the LTFF, EAIF, and WAI for regranting and the fact that funding downstream of Open Phil does get back to them from critics.
4. More For-Profit EA Work/EA Organizations Charging for Their Work[10]
I expect this post to already get some pushback but I expect it on this point in particular. I think there are costs and benefits to EA organizations selling their services and the specifics/details always matter. Sometimes it will be a good idea and sometimes it won’t be. But here are some of the Pros/Cons of this.
Pros
-More money for the organization from a “diverse” source. This is what this post is about so it is worth spelling it out clearly. If an organization can sell its services/products to others, this will give the organization money.
-A way for organizations to know that there is demand for what they are doing. Very often (and I don’t want to call out specific projects) people in EA will start a project and spend a lot of time on it. This can be a research project, a service, a tool, a website, etc., and spend a lot of time on building it without having asked a sufficient number of people/organizations who are their prospective users if they are even going to want/need the thing being built . The project/service/tool then goes on to be minimally used. If you have intended customers, you ought to find out if they would even want it. I’m not opposed to having a lot more “markets” in EA work where different organizations sell/provide services to other organizations. This will lead to more intentional work where research done is meant to inform certain specific questions that will change outcomes and a forecasting tool is only developed if it will be used.
Cons:
-A lot of the people/sentient beings/stakeholders that EA organizations “serve” aren’t well represented in markets. Non-human animals don’t buy things. The poorest people in society don’t have the capital to buy the services, and that’s why they need help in the first place. Future people aren’t going to pay for allowing them to exist or at least not yet.
- It’s possible that this will cause organizations to not optimize on doing good because they have to make sure they will be paid for it. You’ll build the version of the product/service that can sell the best but not have the most impact.
-When there is a profit motive, this will skew incentives towards profit as opposed to doing what is good. We have seen this repeatedly with work in AI safety for example. When people have the opportunity for great amounts of profit and what is good for society, they often/usually will succumb to the incentive of their profit motive.
5. Acceptance
The last option is to realize that funding diversification is a tradeoff that simply isn’t worth making given the current state of affairs. While every organization would ideally not be reliant on a single donor, should they cut their budget by 50%+ to make this a reality? Should they not allow Open Phil to account for more than 50% of their budget and scale down their work/impact? All else equal, funding diversity would be great, but all else isn’t equal and funding diversification is super costly and will result in less impact.
My Personal Solution
I’d propose we go for a mix of 1,2,4, and 5. We should be working on getting more funders in the long term (and investing in it). In the meantime, we should spend a little bit less and do a bit more for-profit work in certain circumstances but on the whole, we should more or less accept that having funding diversification is a tradeoff that is broadly not worth it in the current state of affairs.
Conclusion
In conclusion, the EA ecosystem’s funding is primarily concentrated in one donor and thus, due to the law of weighted averages, most organizations can’t have a diverse donor base that doesn’t rely on Open Phil. Making a lot of money is difficult and we shouldn’t expect in the short term to have a dramatic change in our top-line funding distribution while the movement is so small. We therefore need to grapple with the tradeoffs imposed by trying to have the majority of EA organizations have a diverse funding base.
Further Readings
https://forum.effectivealtruism.org/posts/ZbaDmowkXbTBsxvHn/historical-ea-funding-data#comments
TLDR: This forum post is well-written but overlooks two key points: the high-status perception of OP funding within EA, which skews community inclusion and metrics, and the reputational risks for non-EA-branded donors and organisations in aligning with EA, leading to a disconnect in recognising their contributions.
This forum post effectively outlines the options and includes a (partial) call to action to stop complaining, acknowledge issues, and take steps forward. However, it overlooks two important aspects that I believe warrant discussion:
First, the post does not account for how funding from OP is perceived as a marker of being an EA organisation. This perception creates a feedback loop where organisations funded by Open Philanthropy are seen as high-status and placed centrally within the EA community. In contrast, organisations or sub-teams not viewed as fully aligned are often excluded from community metrics. This dynamic significantly influences which organisations are recognised as part of the EA community.
Additionally, in these kinds of statistics, there is little recognition of sub-teams within organisations or EA-adjacent groups that contribute to EA goals without formally associating with the movement. For example, many civil service roles are funded through diverse portfolios and remain underrepresented in these discussions. Similarly, some organisations prefer not to publicly align with EA for strategic reasons, despite their close alignment in practice.
Second, for figures like Dustin and Cari, donating to EA-branded organisations may make sense given their personal brands are closely tied to the community. However, for other donors, explicitly associating with EA poses reputational risks and diminishes the credit they receive for their philanthropic efforts. Schmidt Futures exemplifies this—it is a family-funded organisation doing work aligned with EA interests but avoids formally associating with the movement, as I think there’s little incentive to do so from their perspective.
Similarly, I work for an organisation largely funded by a donor not widely considered an "EA billionaire," yet the organisation is almost entirely staffed by EAs doing aligned work. Despite this, the donor is unlikely to appear in lists of EA funding sources, highlighting a gap in how contributions are recognised.
This is a great point that I strong upvoted.
1 is just a good point. happy you made it. I wish people owned the EA reputation more and we should work towards it having a better brand. I also acknowledge that "what is EA" is fuzzy and also that in an ideal analysis, I would have summed up EA organizations (by my definition) receiving funding as opposed to tallying EA sources of funding.
On 2, I agree but I don't think this will change the calculation very much. Even if I were to grant Schmidt Ventures as an EA funder (I don't fwiw, they mostly fund science), this doesn't change much in the realm of being able to significantly diversify funding (there are 2 funding sources and not one).
Withr respect to the org you work for being funded by a donor who isn't widely considered an "EA billionaire", that's awesome but again, I don't think will change the math very much. It would be similar (at best) to adding another SFF. It would take OPs funding percentage from 89% to say 85%. It helps but doesn't change the big picture.
I should also note that this isn't surprising. In the hedge fund or VC space, firms usually have a few investors representing the bulk of their funds. Every firm is different but generally speaking, fewer than 5 investors make up 80% of a firm's AUM and it very much follows a power law.
Thanks for these helpful points! Minor note: Did you mean Schmidt Futures and not Smith Futures?
Yes!
Worth noting that although high EA salaries increase the risk to EA organisations, they reduce risk to EA individuals, because people can spend less than their full salary, thereby saving for a time when EA funding dries up.
(In general, the salaries which I will work for in EA go up with funding uncertainty, not down, because indeed it means future funding is more likely to dry up, and I have to pay the high costs of a career transition, or self-fund for many years)
Thanks for raising this!
What about applying for funding from non-EA funders? I know that most of these funders won’t be interested in EA charities because they don’t fit into their view of what good charities look like, but the non-EA funding landscape is big and there are definitely funders out there that could be convinced to fund some EA projects. My sense is that many many EA orgs don’t even try to access this funding.
Granted - it’s a lot of work, and may take years of relationship building, and all of that is a distraction from execution. But the rewards are significant because the counterfactuals of that funding is far ‘cheaper’ than the counterfactuals of the money orgs take from EA funders. Plus I think fundraising from non-EA funders is really good practice for EAs at communicating about effective charity work in ways that are appealing to non-EAs. This is something we’re generally very bad at, and becoming good at it is a necessary condition for us to really grow the base of support and resources for this movement, so I think we should lean into opportunities to engage with the non-EA world instead of staying in our silo
Yup, strong agree here. Many reasons to raise money from non-EA sources. Hard but extremely valuable.
One factor this post might be failing to account for: the wealth of Anthropic founders and early-stage employees, many of whom are EAs, EA-adjacent, or at minimum very interested in supporting existential AI safety. I don't know how much equity they have collectively, how liquid it is, how much they plan to donate, etc. But if I had to guess, there's probably at least $1B earmarked for EA projects there, at least in NPV terms?
(In general, this topic seems under-discussed.)
I also have very wide error bars on my $1B estimate; I have no idea how much equity early employees would normally retain in a startup like Anthropic. That number is also probably dominated by the particular compensation arrangements and donation plans of ~5–10 key people and so very sensitive to assumptions about them individually.
Forbes estimates that the seven co-founders are now worth $1-2 billion each.
This, along with OAI is likely worth investigating.
Good point. OpenAI as well.
Indeed, though EAs were less well-represented at the senior managerial and executive levels of OpenAI, especially after the Anthropic departures.
I agree that salaries in EA should be more in line with the rest of the non-profit sector[1], that Open Philanthropy is the main funder of many projects, that funding diversification has tradeoffs, and that members of the EA community should donate (much) more.
But I think this post exaggerates the % of effectiveness-oriented funding that comes from OpenPhil, at least for projects outside of EA community building.
I think the main reason is that most effectiveness-minded donors (including billionaires and agencies like USAID) are not part of the EA community, but still fund "causes the EA community cares about".
Here are estimated appproximate amounts donated via the GWWC donation platform in the past year, by rough cause area. (Note, these have not been double-checked and should not be considered official numbers, and they don't include donations reported by pledgers made outside the GWWC platform)
For donors not using the GWWC platform, I think this hinges a lot on how you define "EA donors" (see below)
I think most GWWC donors also don't consider themselves part of the EA community[4], I don't think this matters much in terms of our willingness to fund the most impactful projects that help improve the lives of others.
Here are some other effectiveness-oriented sources of donations:
As you mention in a footnote, the Navigation Fund is funding the Shrimp Welfare Project, and many other high-impact projects in causes that the EA community cares about, even if (as far as I know) it's not explicitly part of the EA community, and you don't include it in the funding amounts in this post.
Crustacean Compassion was started in 2016 and only got funding from OpenPhil in 2021. I don't think that only people in the EA community donate significantly to crustacean welfare.
I think it's weird to mention "fund shrimp, wild animals, or insect welfare" as causes in a post on how OpenPhil is the main funder of many EA projects, given that OpenPhil stopped funding those.
In his 2013 TED Talk, Peter Singer claims that Bill and Melinda Gates and Warren Buffet were "the most effective altruists in history", and in 2023 MacAskill keeps defining EA not in terms of a specific movement/community but in terms of "using evidence and careful reasoning to try to do more good."
I think probably in their mind EA (as they define it) already had billions of dollars committed to it in 2009.
See also We need more nuance regarding funding gaps from 2022 with an estimate of the number of funding sources for different cause areas at different scales. I think for most cause areas the number of funding sources of more than ~$1M/year increased since 2022.
Besides the Shrimp Welfare Project mentioned above, other interesting examples of non-OpenPhil funding are that Lightcone stopped receiving funding from OpenPhil but managed to raise >$1.1M in a month and that EAIF isn’t *currently* funding constrained.
I think currently EA salaries are higher for non-leadership roles and probably lower for leadership roles
Mostly this $10M donation
Note that this area might be over-represented in this table, as the main way to donate to GWWC and to Effective Altruism Infrastructure Fund is via the GWWC platform, while donations to projects in other cause areas are usually made outside of it
For what it's worth, I also don't consider myself part of the EA community.
See https://en.wikipedia.org/wiki/User_talk:IanFoundersPledge#You_can't_write_about_Founders_Pledge
I agree that 5 (accepting OP-dominated balance sheets) seems like the best solution.
I think a different but related point is that an org that can fundraise outside of EA is that much more valuable than an org producing identical outputs but fundraising from within EA. The big example of this of course is GiveWell - using EA principles but getting money from a far wider set of people. Raising $1 from OP (and even more so other EA sources) has pretty direct opportunity costs for other high-impact projects, but raising $1 from someone else mainly trades off against that donor's consumption or their other donations which we (putatively) think are a lot less impactful.
I agree, raising money from non-EA sources is hard but extremely valuable for multiple reasons; the counterfactual probably isn't as charitable, it grows the EA pie, it might bring them into more effective giving, etc.
I agree overall but I want to add that becoming dependent on non-EA donors could put you under pressure to do more non-EA things / less EA things -- either party could pull the other towards themselves.
Another good point. It's amazing how much donors can influence priorities, even subconsciously
Thanks for the useful post Marcus!
If people reading might be a good fit for running a project helping to improve funding diversification, I encourage them to apply to the EA Infrastructure Fund. We are keen to receive applications that help with this (and aren't currently very funding constrained ourselves).
As for ideas for projects; Marcus lists some above, I list some on my post, and you might have ideas of your own.
This doesn't change your conclusions at all, but it's hard to count the Nonlinear Network donations without accidentally double counting because a lot of the largest donors are also on the platform.
Sure. While I am a fan of the NLN, a big point of mine is that individual donors is a very tiny portion of the funding ecosystem, especially for larger projects, they can't really use independent donors to diversify.
The Nonlinear Network was designed to help increase funding diversification in the movement.
It was also designed to be maximally low effort on both the funder and the applicant side. This is why we allow people to apply with any existing fundraising materials and there are very few required questions so if you've already fundraised, it should take you minutes to apply.
It's not nearly enough to solve the whole problem, but it's low cost and high upside so good EV for most AI safety orgs
"First of all, I want to tip my hat to AIM (formerly Charity Entrepreneurship)"
I think EA community could also be more proactive in helping more EA-aligned business people - potential future big donors - to appear.
Why not form a group of EA activists, who have business experience, in order to:
1) support EA-aligned people starting their business careers - to help them with advice and mentoring regarding how to be more effective and successful in their business careers
2) think about how to support EA-aligned people in their startup ventures (for example, by giving advice on strategy, marketing, fund-raising, investor-relations, etc.)
Marcus, do you want to discuss?
You go into it a bit, and your post certainly answers some very important questions, but two that should be dug into further by someone are Why do other rich donors avoid the EA space (intentional, accidental, misinformed) and How do we get them to change that/those reasons.
Thanks for the synopsis! It's a very significant topic, especially because of what it means to have a very small number of individuals either directly or indirectly steer the whole (89%) EA community.