Hello, dear readers! This is our weekly brief on remarkable AI topics, so you can keep up without drowning in the hype.

Today’s focus — Apple may be rethinking its AI strategy, shifting from building everything in-house to potentially hosting an AI marketplace. If true, that’s a meaningful pivot: instead of trying to win the model race, Apple leans into what it already does best — distribution.

Also in this week’s edition:

Apple May Stop Chasing AI — and Start Hosting It

For the past year, every major tech company has been chasing the same idea: build a do-it-all AI stack.

Apple might be quietly stepping away from that. According to Bloomberg, the company is exploring an AI marketplace model in future iOS versions — opening the door for multiple AI providers inside its ecosystem.

Instead of betting on a single, fully Apple-built intelligence layer, users could interact with different models depending on the task.

That’s a notable shift. Apple has historically preferred tight control — hardware, software, services, all vertically integrated. An AI marketplace would be the opposite: a controlled environment, but with external intelligence plugged in.

There’s a practical logic here. Competing head-on in models is expensive, fast-moving, and uncertain. Owning distribution, on the other hand, is familiar territory. Apple doesn’t need to win the AI race if it can host it.

Sora Stumbles, But the Video Race Isn’t Slowing Down

OpenAI Sora may have hit a wall — but that hasn’t stopped others from rushing in.

Following reports that OpenAI scaled back or limited Sora’s rollout, competitors are already positioning themselves to fill the gap. Most notably, Elon Musk has teased an “epic” video generator for xAI.

For now, that promise is just that — a promise. Musk has a track record of ambitious timelines, not all of which materialize as planned.

Still, the broader point stands. Even if Sora isn’t ready for prime time, the demand for AI video hasn’t gone anywhere. If anything, the pause creates more space for competitors to experiment, iterate, and claim momentum.

In other words, Sora may have stumbled — but the category is very much alive.

Meta Turns AI Into a Job Requirement

Meta is taking a more direct route: making sure people actually use AI.

According to Business Insider, the company has been running internal “AI weeks,” training employees to integrate AI tools into their daily workflows — from coding assistants to automation tools. Leadership, including Mark Zuckerberg, is framing AI usage as a core expectation rather than an optional skill.

This is less about launching new technology and more about changing behavior.

If AI tools are powerful but underused, the bottleneck isn’t the model — it’s adoption. Meta’s approach suggests the next phase of AI may depend less on breakthroughs, and more on whether people actually incorporate these systems into their work.

There’s also a subtle shift in tone: not “AI can help you,” but “you are expected to use AI” — or else.

Thanks for reading AIport. Until next Monday — by then, AI will almost certainly have launched another breakthrough… and quietly walked back another one.