Hey {{First name|there}}! It’s Aaron.

OpenAI just killed one of its most impressive AI products.

Not because it didn’t work. But because it cost too much to exist.

Here's what's shifting in AI this week:

OpenAI didn’t discontinue Sora because it was bad.

It looks more like the economics stopped making sense.

According to The Wall Street Journal, Sora was burning roughly $1 million a day and had fallen to fewer than 500,000 active users. That’s not a scaling product. It’s a cost centre.

When Sora launched, it felt like AI video had finally arrived. You could generate cinematic scenes — and even place yourself inside them.

I called it wrong.

I thought closing the distance between idea and execution was enough. What I missed: shorter distance means nothing if nobody’s walking it.

So the real question is this: where does AI actually make money right now?

And right now, video does not look like the answer.

Most creators don’t need “Hollywood-level video instantly.” They need to know what to make, whether it will perform, and how to produce consistently.

Sora solved a different problem. It made creation easier, but not necessarily more useful.

That’s the gap.

The Wall Street Journal reported that Sora’s retention weakened even as costs stayed high, and that there was no clear revenue path strong enough to justify the burn. That’s a tough combination to defend in any market, let alone one where inference costs can destroy margins fast.

OpenAI’s attention appears to be moving toward coding models and enterprise tools — the kind of products that are easier to price, easier to measure, and easier to sell.

Not as flashy. But much easier to justify.

Because those tools don’t just generate — they execute. And execution is where companies can actually measure return.

Sora’s discontinuation doesn’t kill AI video. It clarifies the economics.

The tools still standing tend to make the trade-off more explicit:

The pattern is simple: AI video works when costs are controlled.

Sora tried to do too much, too early.

The Wall Street Journal also reported that Disney had a $1 billion partnership tied to Sora and was informed of the shutdown less than an hour before the public announcement.

That’s not just a product decision.

OpenAI’s press cycle mattered more than a billion-dollar relationship. That’s a power move — and a preview of how AI companies treat enterprise dependency.

Because underneath all of this:

AI isn’t being built for what’s impressive. It’s being built for what sustains itself.

Use AI video when:

Avoid it when:

Because here’s the trade-off:

AI video gives you speed. It doesn’t give you stability yet.

OpenAI chose infrastructure over impressiveness.

That’s the real story.

You can build your workflow around what looks exciting. Or around what’s still running in six months.

Right now, those aren’t the same thing.

See you in the next one,

I hope you found value in today’s read. If you enjoy the content and want to support me, consider checking out today’s sponsor or buy me a coffee. It helps me keep creating great content for you.

Check out my previous posts here