$100 Billion in AI Philanthropy Sounds Great! Think Again.

Nan Ransohoff shook up a lot of nonprofits and foundations when she took to Substack last week to describe the so-called third wave of philanthropy now coming from AI wealth. Ransohoff, who heads climate work at the technology company Stripe, estimates that OpenAI, Anthropic, and the people who built them, have accumulated enough wealth to start distributing up to $100 billion annually in new philanthropic capital — a staggering sum given the nearly $600 billion that now flows to charities from individuals, foundations, and companies. 

In her more than 3,600-word essay, with accompanying data, Ransohoff presents a detailed account of what a growing chorus of tech professionals say is needed to leverage this new tidal wave of wealth. The nation’s first philanthropic wave, Ransohoff reminds us, built libraries, universities, and the modern foundation. The second built an apparatus to measure whether grants were making a difference, promoted cost efficacy, and created the donor-advised fund. The third, she writes, will require new institutions, new talent, and new thinking — what Ransohoff describes as a Silicon Valley of new public goods.

Before we accept that premise, let’s remember the adage that if you have an hour to solve a problem, spend 50 minutes thinking about the problem and just 10 on the solution. That’s harder than it sounds. Most failures of strategy are not failures of execution. They are failures upstream, at the level where the problem was first named.

Defining the Problem

So let us spend a moment naming this problem carefully. The story we are being told is that the constraint on philanthropy is capital. But the 50-year record suggests otherwise.

Over the past five decades, Americans have given roughly 2.5 percent of GDP to philanthropy. Real GDP nearly quadrupled in that time, from roughly $6 trillion to more than $23 trillion in inflation-adjusted dollars, according to the Bureau of Economic Analysis. The percentage of giving to GDP did not move. Wealth concentrated to a degree without modern precedent. The philanthropy profession built itself into a multi-billion-dollar industry, with graduate programs and credentials and think tanks. 

The percentage, however, did not move. We invented donor-advised funds, the Giving Pledge, and GivingTuesday. We added technology and behavioral science and AI insights to our CRMs. The percentage, however, did not move.

Some outside the social sector argue that the percentage has stayed flat because the existing nonprofit infrastructure cannot absorb more capital well. Build new institutions, the argument goes, and the share of GDP given to philanthropy will rise.

This gets the causality backwards. Over the same 50 years that giving held flat, the underlying social problems philanthropy seeks to eliminate have not meaningfully disappeared. Homelessness in major American cities rose. Income inequality reached a level not seen since the 1920s. Life expectancy in the United States saw its steepest decline in a century. If money were the lever, that lever has been failing for half a century regardless of how hard we pulled it.

The flat line is not evidence that we ran out of money to give or that the existing institutions failed to deploy it. It is evidence that we have been asking the wrong question. The question is not how much capital flows. It is who controls the wealth, through what structures it moves, and how far that wealth sits from the people closest to the work.

Giving Well Requires More Than Data

To better understand the real problem at hand, we need to examine the values of the people who hold the wealth and the inequities created through wealth generation. Giving well is harder than simply giving money away. That is, it requires the kind of knowledge that develops over time and through proximity — an understanding of context, history, and relationships that data alone cannot produce.

The venture capital model is built on the opposite premise. It assumes the investor stands outside the system, applies capital to it, measures the results, and adjusts. The investor is not transformed by the work. The system is treated as a machine to be tuned. That posture is fine for funding software. It is much less suited to the slow, relational, often unmeasurable work of providing care to people in our communities. The problem is not lack of money. It’s an imbalance of power.

The pattern I’m describing is something we all fall into. We mistake the inside of our own bubble as a reflection of other realities outside of it. We assume the thing that seems most important to our world is the thing that is most important to others’ worlds. In this third wave, tech gurus say that “effective and rapid philanthropic deployment” is the solution. But the American tech sector has not historically been good at solving problems outside its own “shareholder” field of vision. 

A Dismal Track Record

OpenAI and Anthropic, the two companies generating essentially all of the philanthropic capital that might flow to charitable causes, are headquartered in San Francisco. Their largest investors, Microsoft and Amazon, are in the Seattle metro area. According to the 2024 Annual Homelessness Assessment Report from the U.S. Department of Housing and Urban Development, San Francisco counted 8,323 people experiencing homelessness on a single night that year. Seattle’s King County counted nearly 17,000, more than half of them unsheltered. 

These figures persist despite extraordinary local investment: San Francisco expanded shelter capacity by 60 percent from 2018 to 2024, funded in part by a 2018 ballot measure that has since directed nearly half a billion dollars to homelessness services. Yet the city’s own Department of Homelessness and Supportive Housing estimates that for every one person it successfully houses each year, three more become homeless. The wealth that was supposed to lift these cities has not. 

Moreover, the sector has not improved labor conditions in the cobalt and rare-earth supply chains that build its hardware — a new form of colonialism, as Pope Leo XIV declared this week. The tech sector has also failed to repair the attention economy its platforms engineered. 

This is not an indictment of the people involved, many of whom are thoughtful and well-intentioned. It is an observation: The sector’s track record on problems close to home is not strong enough to justify its confidently proposed redesign of the philanthropic sector. The instinct to import Silicon Valley’s operating system into philanthropy is being advanced by the same culture that has not yet figured out how to care for its own neighbors.

The Fence Problem 

G.K. Chesterton, an early 20th-century English writer, offered an important principle: Before you tear down a fence in a field, he urged, find out what the fence is for. If you cannot say what its function is, you are not the person who should tear it down. The nonprofit sector has many fences. Some of them look inefficient or pointless to people who arrived this morning: The long timelines. The focus groups. The program officers who hold information sessions that appear to produce nothing. 

So what would it look like to spend the 50 minutes trying to fully understand the problem?

It would involve questioning why the percentage of philanthropic giving to GDP hasn’t changed, why increasing amounts of money are sitting in DAFs, and what conditions over the past 50 years have kept giving from rising as substantially as billionaire riches. And it would mean examining why social inequities have worsened, even as wealth has exploded. 

The solution is unlikely to be more agile philanthropic institutions. Instead, it will likely involve the slow, uncomfortable work of changing values, behavior patterns, and the distance between funders and those served. Or perhaps it will require something even slower: a willingness by grant makers themselves to be changed by encounters with social inequities in their own backyard — not as donors but as citizens invested in their community. 

A new wave of capital is coming. No argument there. But the loud voices and confident prescriptions about how to deploy it must not drown out the more useful question that almost no one is asking: Is the assumption that more money equals more good simply wrong? The issue is not whether the sector can absorb the third wave. It is whether the model of new philanthropic startups can enable what the AI billionaires believe it will.

The wealth we are preparing to deploy is part of the system we are seeking to repair. So perhaps we begin by pausing — by daring to understand the problem before we rush, once again, toward the solution.