The New Money Is Coming. Generosity Is Not the Test.

Philanthropy is once again captivated by a new generation of wealth. As fortunes accumulate around artificial intelligence, a growing number of observers are forecasting a transformation in charitable giving. Some estimate that AI founders, investors, and early employees could eventually direct tens of billions of dollars annually toward philanthropic causes. Others see the emergence of a new donor class eager to apply technological ambition, analytical rigor, and unprecedented resources to some of humanity’s most pressing challenges.

I understand the excitement.

Philanthropy has always followed wealth. Industrial fortunes built libraries and universities. Oil and manufacturing wealth helped shape modern foundations. The internet era produced a generation of donors who imported the language of innovation, disruption, measurement, and scale into the social sector.

Now, many expect AI wealth to do the same. Yet much of the conversation begins with the wrong question. The debate so far has centered on how much money AI will bring into philanthropy. Supporters envision new resources for scientific progress, poverty reduction, and global public goods. Critics worry about elite influence or the concentration of private power. Both sides, however, have tended to focus on the capital itself.

Money matters, especially given the current disruptions to development finance. But every fortune carries a worldview.

In my opinion, the more important debate is not how much wealth AI will generate for philanthropy. It is what assumptions about progress, expertise, governance, and social change will accompany that wealth when it arrives.

This question feels especially urgent because AI is not simply another industry creating another group of wealthy donors. It is a technology that may reshape labor markets, alter how knowledge is produced and consumed, concentrate economic power, and challenge institutions across society. Even some of the leaders building these systems have warned that AI could produce levels of wealth concentration capable of straining the social contract itself.

That creates an unusual situation. Philanthropy may soon be funded by fortunes generated from the same technological transformation that many nonprofits, governments, and communities will be struggling to understand, govern, and adapt to.

The question, therefore, is not whether AI wealth will be generous. It is whether philanthropy is prepared for what that generosity represents.

Every Fortune Comes with a Theory of Change

Philanthropy often presents itself as the redistribution of wealth. But it is also the redistribution of ideas.

Every significant philanthropic fortune arrives with an implicit theory of how progress happens. Carnegie believed knowledge could uplift society, and built libraries to prove it. The theory came bundled with the money.

AI wealth will be no different.

The emerging discourse around AI philanthropy already reveals certain assumptions. Some advocates envision a new era of highly optimized giving, where unprecedented resources are deployed with analytical rigor toward society’s greatest challenges. Others see opportunities to build entirely new institutions, support transformative technologies, and accelerate solutions at a scale traditional philanthropy could never achieve.

I genuinely admire a lot of this ambition. Yet every theory of change contains blind spots.

I’m not worried that AI philanthropists will bring too much analytical thinking to philanthropy. I worry that the worldview that emerged from building successful technologies may be mistakenly assumed to apply equally well to solving social and political problems.

Technology often advances through optimization. Governance rarely does. Many of society’s most persistent challenges endure not because solutions are unavailable, but because people disagree about priorities, values, trade-offs, and power.

Consider democratic accountability, social cohesion, gender equality, state effectiveness, climate governance, or public trust. These are not primarily engineering challenges. They are institutional and political challenges.

I have spent years trying to get evidence into policy, and the lesson that stuck is that better evidence alone rarely changes outcomes. Evidence matters, but so do incentives, institutions, relationships, and political context. African evidence practitioners put the risk plainly at the recent Global Data Festival that I attended in Nairobi: AI’s promise can tempt policymakers into believing that complex problems have quick, attractive solutions, when the underlying issues remain deeply human and institutional.

The question therefore is not whether AI philanthropists will be generous. It is whether philanthropy can absorb their resources without quietly adopting the assumption that every social problem is ultimately waiting for a technical solution.

When the Problem Funds the Solution

There is another reason the rise of AI philanthropy deserves closer scrutiny.

Several prominent AI founders have publicly framed their philanthropic commitments as a response to the very disruptions AI may create: economic concentration, labor market displacement, and institutional upheaval.

We all should welcome those commitments. However, they raise a difficult question. What happens when philanthropy is funded by the same forces that are generating the social disruptions philanthropy is later asked to address?

History offers precedents. Industrial fortunes funded efforts to soften the consequences of industrialization. Financial wealth has supported initiatives aimed at reducing inequality. AI may simply be the latest chapter in this story. But the scale of potential disruption makes the tension harder to ignore.

Imagine a future in which AI contributes to widespread labor displacement while simultaneously creating unprecedented private wealth. Philanthropic capital might help workers retrain, support new forms of education, or cushion communities affected by change. All of that might be valuable.

But it doesn’t resolve a deeper question: who gets to decide what adaptation looks like?

This is not a new question. Philanthropy has always had to answer for who decides - it is why the sector now argues about shifting power, participation, and whose voice sets the agenda. What AI changes is where the money comes from. Ordinarily a fortune and the problem it addresses have separate origins - Carnegie’s steel did not cause the illiteracy his libraries answered. AI wealth is different. It is generated by the same force now reshaping labor, concentrating power, and straining institutions. The funder is no longer an outsider bringing resources to a problem. The funder is implicated in it.

Funding Its Own Oversight

If legitimacy is the problem, the instinct will be to reach for solutions. Fund retraining. Fund education. Fund the institutions that hold power to account.

Now think about that last one. If AI wealth is a product of the disruption, then funding the universities, journalists, and civic groups meant to scrutinize that disruption is the disruption paying for its own oversight.

This is not a reason to refuse the money. It is a reason to be honest about what the money cannot buy.

Scrutiny funded by the thing being scrutinized is not independent by default. It becomes independent only if the funder builds in the freedom to be turned against itself - and honors that freedom when it is used. Most philanthropy is not built to do this. I have watched foundations fund independent institutions and still expect them to arrive at the funder’s conclusions. The money advances the funder’s theory of change; it rarely underwrites dissent from it.

So the legitimacy problem does not dissolve when philanthropy starts funding good institutions. It moves inside the funding relationship itself.

The harder discipline is not choosing what to fund. It is funding it in a way that leaves the recipient free to turn against you - and staying at the table when they do.

The point is not to solve the problems AI creates. Societies solve their own problems, or they do not solve them at all. The point is to fund the capacity to do so: the ability of a society to make collective choices, and to change its mind when it gets them wrong.

That capacity is not a solution. A solution is a fixed answer. Capacity is what a society uses to keep answering, long after any single funder has lost interest.

The Humility Test

Why, then, is the instinct always to fund solutions?

Because the people who will hold this wealth built things. They optimized systems, shipped products, watched effort convert into outcomes. That experience teaches a powerful lesson: hard problems yield to enough intelligence applied with enough rigor.

It is the wrong lesson here.

Building a powerful system teaches you nothing about how a society should govern it. Optimizing a process teaches you nothing about what a good society owes its members. These are not harder versions of engineering problems. They are different in kind.

This is what humility means in philanthropy - not modesty of ambition, but accuracy about the limits of one’s own expertise. The ability to build is not the ability to govern. Resources are not wisdom.

The sector rarely admits this. Philanthropy is built around certainty: strategies, theories of change, metrics, plans. Yet the advances that last have tended to come not from the certainty of powerful actors, but from institutions that leave room for learning, disagreement, and correction.

The arrival of AI wealth will test whether the sector can tell the difference between having resources and having wisdom.

The coming wave – if it arrives – will reshape philanthropy. The question is whether philanthropy sees itself as the architect of that future or its steward.

One path concentrates influence, however good the intentions behind it. The other accepts a smaller and more democratic aspiration: helping societies build the capacity to govern themselves in an age of extraordinary change.

The defining test of AI philanthropy will not be how much it gives away. It will be how much power it is willing to leave in the hands of others.