We’ve spent the past 3 ½ years talking about AI models. I believe the next conversation will be about AI infrastructure.
After decades operating at the intersection of Silicon Valley and Hollywood, I’ve learned that the biggest technology shifts aren’t always driven by the most visible applications—they’re often enabled by the foundational layers operating quietly beneath them.
Today, I believe video has become AI’s last unstructured frontier.
As AI agents become more capable, hyperscalers invest hundreds of billions in data centers, and enterprises demand personalized, trustworthy media experiences at global scale, a new question emerges:
How do we make video itself programmable?
In my latest article, I explore why programmable video may become one of the most important infrastructure opportunities of the AI era, why data-center economics will increasingly matter, and why the next competitive advantage may not come from generating more video—but from making the world’s existing video infinitely more intelligent.
Using ION Video as a case study, I examine why a technology that may have been years ahead of its time could now be arriving at precisely the right moment.
I’d love to hear your perspective: could programmable video become the next foundational operating layer for AI?