Sora and the Myth of AI's Inevitability

Sora is officially dead. Less than two years after teasing it and six months after opening it up to consumers. In spite of Disney committing $1 billion to a partnership built around it. Some reports say that Disney found out less than an hour before the public did.

So, I understand if you’re feeling a little whiplash right now. After all, the tool that was going to end Hollywood couldn’t retain a single user for more than 60 days.

For me, it represents a teaching moment because nothing, not even AI, is inevitable. Every time I deliver AI literacy training, I have participants who thank me for calming them down, explicitly referencing being buried in social media hype. “Doomscrolling” is a term entering the public lexicon.

Let’s look at how we got here, from a hype perspective.

“Hollywood Is Done”

When OpenAI first previewed Sora in February 2024, the reaction was immediate. Tyler Perry, a filmmaker and studio owner, put an $800 million studio expansion on hold after seeing a demo.

“Being told that it can do all of these things is one thing, but actually seeing the capabilities, it was mind-blowing,”

“So I am very, very concerned that in the near future, a lot of jobs are going to be lost. I really, really feel that very strongly.”

- Tyler Perry

Other Hollywood insiders, like Ashton Kutcher, jumped in with similar messages. Fear, uncertainty, doubt. Embrace AI or fall behind. Sound familiar?

Some filmmakers and producers took a more pragmatic “wait and see” approach, but those kinds of messages don’t get clicks, and we all know the media needs attention to survive, so doom predictions dominated the headlines.

What the Data Shows

I’ve been saying for a while now that AI tooling has both a capability and an economic gap that need to be addressed. The emerging details about Sora confirm my skepticism. Here’s what happened when Sora went from controlled and cherry-picked demos to real-world users:

• Downloads peaked, then fell 32% month-over-month in December 2025.
• They fell another 45% in January 2026
• 7-day user retention rate: 2%
• 30-day user retention rate: 1%
• 60-day user retention rate: a big fat 0%

And the financials? $2.1 million in revenue versus $15 million per day in costs.

Olivia Moore flagged this trend in December 2025, months before the shutdown. Even the head of Sora, Bill Peebles, noted that “the economics are completely unsustainable” in an X post.

What this confirms is two gaps:

1. Consumers, already battered by inflation, a poor job market, and a desire for tools that… you know… work as advertised, weren’t willing to pay for this tool in any meaningful sense.
2. OpenAI, like other purveyors of this technology, spends a lot of time coming up with impressive-looking demos that are completely detached from real-world results.

I think OpenAI made the correct call in killing this product. It frees them up to focus on enterprise customers. Or as I like to call them: the organizations that have money.

The Apex Short Film

One of the biggest splashes to hit social media was a short film by the Dor Brothers. “What if a 200 million dollar film no longer required 200 million dollars?” is their claim on the site.

Now, the Dor Brothers aren’t random guys who typed in a few prompts. One of them, Yonatan Dor, has reportedly been using these tools daily since 2022, completing 300+ projects. So, right away, we expose the deception that anyone, without training or skills, can produce such things. But second, and more importantly, I thought this video was hot garbage, nearly unwatchable, because I pay attention to details.

Here are just a few of the things I noticed, and I’m not trained in filmography:

• Kinetic Impacts: When debris hits the ground, there’s no secondary interaction. Dust kick-up, cratering, shards.
• Clipping: In many places, falling objects in the background pass through buildings rather than colliding. An airplane’s wing clips a concrete building and shears right through it, leaving the wing intact.
• Gravity: Objects frequently fall at the same “floaty” speeds.
• The Cybertruck: The wheels have no continuity between scenes, the windows are closed, but the woman’s hair is blowing in the “wind”, the engine sounds like a big diesel engine, when it’s an electric vehicle, etc.
• The woman: Her jacket changes between cuts, and her face, while pretty consistent, doesn’t portray the range of emotions and expressions that one would have in that chaos.

So, yeah, neat demo, but there is a 0% chance I could sit through 2 hours of that.

The AI Hype Pattern

Big tech seems to be drifting further and further from any empathy for its customers or understanding of how they work. Sora is a near-perfect case study. How many times in the last few years have you seen this play out?

1. Ship a demo video with cherry-picked outputs.
2. Watch the media repeat whatever a CEO says and amplify the existential fear, because that gets clicks.
3. Sit back while hype bros on social media evangelize, because that attention helps them raise capital or prey on less-savvy users.
4. Collect investment at valuations completely disconnected from reality.
5. Release the product to real users.
6. Burn through capital trying to grab market share for products that effectively have no moat.
7. Shut it down and reallocate the compute.

Most AI companies are sitting somewhere in this pattern. There currently is no off-ramp to profit because no one, except Nvidia, which sells the shovels, is turning a profit.

Marketing, or Deception?

What makes this insidious is that there’s a somewhat blurry line between marketing tactics and outright lying. In Sora’s case, the demo videos were real, and the technology is real too, but the gap between “an impressive demo from a company that had effectively unlimited compute” and “an average person can create Hollywood quality video” is enormous.

AI Isn’t Inevitable. Demand Proof.

You may have noticed that I haven’t claimed that the Sora story means AI video is dead and gone forever. Other tools still exist in the space, and I expect that experiments and incremental improvements will continue. But, when you see a scary headline claiming that a product like Sora could be “The end of Hollywood.” just stop, take a deep breath, and ask questions:

• Who is making this claim? Do they have a financial interest in it?
• Is there any independent data on real-world usage with actual results?
• And my favorite: If AI can’t effectively manage drive-through orders at a fast food restaurant, is it likely that it can replace all of Hollywood?

The media needs clicks. Hype bros want attention that converts to capital and revenue. Neither of those incentives rewards the pragmatic approach of “oh, that’s a neat demo, but no one would actually pay for that quality level today.”

See also: the Metaverse, NFTs, Blockchain, 3D TVs, Full-Self-Driving Cars, Windows Phone, etc.