Adobe creates 50-person CAVA research lab for video and audio AI

When OpenAI dropped Sora, everyone in the creative industry looked straight at Adobe with one question: what is the company that built Premiere and After Effects going to do about it?

It turns out Adobe was already making moves behind closed doors.

According to industry insider Bilawal Sidhu, Adobe quietly formed a brand-new 50-person AI research organization dubbed CAVA, dedicated entirely to generative media.

Meet CAVA

CAVA stands for Co-Creation for Audio, Video, & Animation.

While details leaked via posts on X and LinkedIn, Adobe had not yet made any formal announcements on its official website or the Adobe Research accounts.

The team reportedly gathers around 50 researchers and engineers tasked specifically with pushing generative AI across moving pictures, sound, and animation.

That head count makes it clear this is not just a casual side project. It is a dedicated unit aiming straight at the multimodal frontier.

A wake-up call from Sora

The timing tells the real story here. The news surfaced right on the heels of OpenAI stunning the world with Sora's hyper-realistic video generation.

Sidhu and other industry observers wondered aloud if Sora served as an urgent wake-up call for Adobe to formalize and accelerate its video and multimodal roadmap.

Adobe already made significant noise with Firefly in static image generation, especially around commercial safety. But dynamic video and synchronized audio represent an entirely different battlefield.

If Adobe fails to lead here, its core fortress of video editors and motion designers could start looking elsewhere.

Why it matters

For creators, this is not an overnight fix. CAVA is an R&D unit, which means its models will take time before they surface as one-click features inside Premiere Pro or After Effects. Do not expect magic tools in your creative suite next week.

For the wider industry, this marks the official beginning of the video AI platform war. OpenAI has the raw frontier compute, but Adobe owns the actual professional timeline, the editing workflows, and the enterprise trust.

We think highlighting the word Co-Creation in the group name is a deliberate strategic move. Instead of promoting AI as an autonomous replacement for artists, Adobe continues to pitch AI as a collaborative copilot for pros.

The open question is whether a 50-person research squad can move fast enough against sheer foundation-model brute force. But by formally organizing around video and audio, Adobe is making sure it stays in the fight.