Behind the First Hyperreal AI Film Built on AMD GPUs using TensorWave Cloud

When creativity meets compute, you get VOID RUN—the world’s first hyperreal AI-generated film. Directed by the team at Higgsfield and powered by the AMD GPU cloud at TensorWave, VOID RUN isn’t just a cinematic achievement—it’s a statement on what’s possible when open ecosystems, cutting-edge infrastructure, and AI-driven storytelling converge.

At Advancing AI 2025, Jeff Tatarchuk (Chief GPU Officer and Co-Founder of TensorWave) sat down with Alex Mashrabov (Founder of Higgsfield) to unpack how this one-of-a-kind film was brought to life.

The Story Behind the Pixels

VOID RUN follows a female developer named Em—curious, fearless, and relentless—as she uncovers something powerful and shares it with the world. It’s a story designed by humans, rendered by machines, and brought to life with a new kind of creative pipeline.

According to Mashrabov, the stages of production—pre, post, and everything in between—remained familiar. But instead of traditional rendering engines, Higgsfield used AMD GPUs at every step: from script co-creation and character modeling to full-pixel generation. “It’s a tool for creativity,” he said, “not a replacement.”

Why TensorWave?

When Higgsfield began building the film, they ran into a wall: memory. The workloads required for generative video models couldn’t run effectively on traditional GPU setups. That’s where TensorWave came in.

Jeff shared how TensorWave’s AMD MI300 and MI325X infrastructure—with up to 256GB of VRAM per GPU—enabled Higgsfield to scale creative generation without compromise. “Memory makes all the difference, especially in inference,” Jeff noted. “We were the first to render a film on the MI325. That’s a milestone for everyone building on AMD.”

A Win for Open Ecosystems

This collaboration wasn’t just about raw power—it was about alignment on values. Both TensorWave and Higgsfield are all-in on open source and developer empowerment. Jeff emphasized that open inference stacks are now outperforming proprietary ones, and VOID RUN is proof that openness fuels innovation.

“We’re here to support the next wave of creators—people like Alex, scaling from 20,000 to millions of users,” Jeff said. “That takes serious infrastructure. That’s what we’re building.”

The Future Is (Hyper)Real

VOID RUN is just the beginning. As generative video AI continues to catch up with LLMs, creators will need compute that’s fast, memory-rich, and open. TensorWave’s AMD GPU cloud is ready for it.

And if VOID RUN is any indication, the next generation of storytelling won’t just be watched—it’ll be built.

About TensorWave

TensorWave is the AMD GPU cloud purpose-built for performance. Powered exclusively by Instinct™ Series GPUs, we deliver high-bandwidth, memory-optimized infrastructure that scales with your most demanding models—training or inference.

Ready to get started? Connect with a Sales Engineer.