What OpenAI's Non-Profit Got Wrong on Mission Driven Ventures

In 2015, Sam Altman and Elon Musk founded OpenAI with one belief. AI should benefit all of humanity. Not shareholders. Not investors. Humanity.

Ten years later, they're suing each other in federal court.

I've been following the Musk v. Altman trial closely. Not for the drama. Because what it exposes matters to every company trying to build something good.

Including us.

The Original Promise

OpenAI started as a nonprofit. Their own words: build AI that benefits humanity, "unconstrained by a need to generate financial return."

Then reality hit.

Building frontier AI costs extraordinary amounts of money. Training GPT-4 cost roughly $100 million in compute alone. During the trial, Greg Brockman revealed OpenAI expects to spend $50 billion on computing power in 2026. Just this year.

A nonprofit can't raise that. Nobody would argue otherwise.

So in 2019, they created a for-profit arm with a "capped profit" structure. Investors could earn returns up to a limit. Anything beyond that went back to the mission.

Clever idea.

Raise the capital.

Keep the soul.

When the Structure Changes, Everything Changes

By late 2025, OpenAI had completed a full conversion to a Public Benefit Corporation. Microsoft now owns 27% of it. The valuation is north of $850 billion.

On the stand, Musk said the for-profit arm became "the tail wagging the dog."

Altman's counter: Musk abandoned the company and wanted total control from the start.

Both of them are probably right, at least partly. And that's the point.

Neither of them could stop what happened. The structure made it inevitable. Once you take capital from investors who expect returns, you've introduced a force into the room that doesn't care about your founding document.

Good intentions don't survive bad structure.

Why This Hits Different for Us

Manifest isn't building AI infrastructure. We're not spending billions on compute. But we're building inside the same tension every tech company faces.

What happens when what's good for users and what's good for investors start to pull in opposite directions?

Most companies don't even have to decide. The structure decides for them.

Publicly traded companies are legally obligated to put shareholders first. When a feature would be great for users but bad for engagement metrics, the metrics win. When privacy defaults would protect members but reduce data collection, data wins.

The people running these companies often care. The structure overrules them.

Our Doctrine says it plainly: "We don't believe the people running these companies are villains. We believe they are caught in a structure that makes it impossible to choose the member over the metric."

That's what happened to OpenAI. We refuse to repeat it.

The Structure Is the Promise

We're in the process of becoming a Public Benefit Corporation.

Not as a badge. Not as marketing. As a legal commitment that ties our decisions to the people we serve, not just the people who invested in us.

A PBC doesn't eliminate investors or returns. That's not the point. The point is it changes the legal obligation. In a standard corporation, values vs. shareholder returns has one answer: shareholders win. In a PBC, the company is required to balance profit with public benefit.

The mission gets teeth.

We're also staying private. No IPO. No quarterly earnings calls where we have to explain why we chose trust over growth.

Our Doctrine puts it this way: "We would rather be a steward of trust for a generation than a platform that had to apologize to one."

We Made the Rules Public So We Can't Change Them Quietly

Most companies put values on a wall. We published our Doctrine publicly so that when pressure shows up, we can't quietly walk it back.

Every product decision passes through seven principles. They're not aspirations. They're filters.

We don't ship features that grow engagement by interrupting someone's life. We don't build for virality. We don't measure success in time spent or notification clicks.

And when values and growth conflict, trust wins. Every time.

The final check before anything ships at Manifest is one question:

"Would this make someone more or less willing to share something real with people they trust?"

If the answer isn't clearly "more willing," we stop.

That's a different bar than improving a DAU metric. And yes, it means we'll sometimes leave growth on the table.

OpenAI didn't get to make that choice once the capital was in the room. We're building a company where that choice is already made. In advance. In public.

What This Means for You

If you're using Manifest, here's what I want you to know.

The goals you share in this app are real. The relationships you're building are real. The vulnerability it takes to say out loud what you're working toward, to your people instead of a public feed, that's real.

You deserve a product that can't exploit that.

The OpenAI trial is a window into what happens when a mission meets capital without the right structure around it. I watched it unfold and felt the weight of it. Because we're in the trust business too. Just a different kind.

We're building the structure first. The promise second. The product around both.

That's what Manifest is. That's what we intend to keep it.