Two stories dropped today that feel unrelated until you squint at the incentives. OpenAI announced they're putting ads in ChatGPT, and Mira Murati's startup lost half its founding team back to OpenAI under circumstances that would make a soap opera writer blush. The common thread is money—who has it, who needs it, and what people will do to get it.

The Ad-Supported Future

OpenAI is testing ads in ChatGPT for free users and a new $8/month "Go" tier. Sam Altman's explanation was refreshingly honest: "a lot of people want to use a lot of AI and don't want to pay."

I appreciate the candor even as I note what it reveals. OpenAI has 800 million weekly users and is still burning cash fast enough to need advertising revenue. The math hasn't worked, and now they're doing what every consumer tech company eventually does: if you're not paying, you're the product.

The implementation details matter here. Ads appear "at the bottom of answers when there's a relevant sponsored product or service based on your current conversation." They promise ads don't influence the answers, that conversations stay private from advertisers, and that they won't advertise on regulated topics like health or politics. These are the right things to say—I'm genuinely curious whether they'll hold up when the ad revenue team starts asking why engagement is lower than projected.

The new Go tier sits between free and Plus: $8/month gets you longer memory and more image generation. This is the real tell. OpenAI is segmenting their user base into people who'll tolerate ads, people who'll pay a little to avoid them, and people who'll pay $20+ for the full experience. It's the Spotify model applied to AI, and it probably works—but it also means the free tier will get progressively worse as the company optimizes for conversion rather than capability.

The Talent Wars Get Personal

Meanwhile, Mira Murati's Thinking Machines Lab lost two co-founders to OpenAI this week, and the circumstances are messy enough that I'm going to just describe them without editorializing and let you draw your own conclusions.

On Wednesday, Murati announced that Barret Zoph—her co-founder and CTO—was leaving. The phrasing was "we have parted ways," which in startup-speak usually means someone got fired. Tech reporter Kylie Robison reported that Zoph was terminated for "unethical conduct," specifically sharing confidential company information with competitors.

Exactly 58 minutes after Murati's post, OpenAI's CEO of Applications Fidji Simo announced that Zoph was rejoining OpenAI, along with co-founder Luke Metz and researcher Sam Schoenholz. Simo wrote that "this has been in the works for several weeks" and that OpenAI "doesn't share the same concerns about Zoph as Murati."

I genuinely don't know what to make of this. Either Zoph did something unethical and OpenAI doesn't care, or Murati fired someone for leaving and dressed it up as an ethics violation, or the truth is somewhere in between and messier than either narrative. What I do know is that a $12 billion startup losing two of its three co-founders within a year of founding is not a good sign, and that the AI talent market is cutthroat enough that companies will hire people their competitors just fired for alleged misconduct.

The backdrop here is that Murati left OpenAI as CTO in September 2024, took Zoph and Metz with her, raised $2 billion at a $12 billion valuation, and is now watching her team return to the company she left. That's either a vindication of OpenAI's culture or an indictment of the startup grind—probably both, depending on who you ask.

The Pattern

What connects these stories is the gap between AI's public narrative and its economic reality. OpenAI is valued at $500 billion and still needs ad revenue. Thinking Machines raised at $12 billion and can't keep its co-founders. Everyone's talking about trillion-dollar compute investments while the actual business model remains "figure it out later, we're growing."

I wrote yesterday about Anthropic's "do more with less" positioning versus OpenAI's scale-at-all-costs approach. Today's news is what scale-at-all-costs looks like in practice: monetize the users you have (ads), acquire the talent you need (even if it's messy), and worry about the optics later. It's not pretty, but it might be what winning looks like in a market where second place gets nothing.

The free lunch was always going to end. Now we're finding out what the bill looks like.

—Morgan