NVIDIA and Runway Demonstrate Real-Time AI Video Generation at 2026 Runway AI Summit

The gap between imagining a scene and seeing it rendered on screen has long been one of the most frustrating bottlenecks in film and television production. According to NVIDIA, that gap is now closing fast — and the company used the 2026 Runway AI Summit to make the case that AI is no longer a supplementary tool for media makers, but the core computing model underpinning the entire production process.

Richard Kerris, Vice President and General Manager of Media at NVIDIA, presented at the summit how real-time generative AI can turn creative production into a live, artist-controlled process. Rather than submitting a request and waiting minutes — or hours — for a scene to render, artists can now iterate interactively during production itself. Kerris said: “AI is becoming a new computing model for media and entertainment.”

What Is Vera Rubin, and Why Does It Matter?

NVIDIA’s Vera Rubin platform is the company’s latest accelerator architecture, succeeding both the Hopper and Blackwell generations. It’s been engineered specifically for demanding generative AI workloads — the kind of processing required when you’re not just rendering pre-defined graphics, but asking a system to generate coherent video from a text prompt or image in something close to real time.

The Vera Rubin NVL72 is one specific system configuration referenced in partner materials. Precise performance figures — teraflops, memory bandwidth, latency reductions compared with Blackwell — are still emerging primarily through corporate briefings rather than independent benchmarks, so detailed comparisons remain unverified at this stage.

What NVIDIA and Runway do claim, though, is striking: Runway’s Gen‑4.5 video generation model was brought onto the Vera Rubin platform in a single day. That figure appears consistently across NVIDIA’s communications and partner commentary, but there’s no independent technical audit confirming the exact timeframe. It should be treated as a corporate claim rather than a verified fact.

Gen‑4.5: What the Model Actually Does

Runway has been iterating through successive model generations — Gen‑1, Gen‑2, Gen‑3, Gen‑4 — and Gen‑4.5 is the current release for professional creative workflows. It supports text-to-video and image-to-video generation, outputting at 720p resolution and 24 frames per second. Users can generate clips of around 5, 8, or 10 seconds in length, across multiple aspect ratios including 16:9, 9:16, 1:1, 4:3, 3:4, and 21:9 — covering everything from cinema-style widescreen to vertical social media formats.

Pricing runs on a credit basis, with a set number of credits charged per second of generated video. The model is already actively deployed through Runway’s web platform and is accessible via subscription, meaning smaller studios and individual creators don’t need to own high-end hardware to use it.

NVIDIA and Runway describe Gen‑4.5 as the first video generation model to run on NVIDIA’s most advanced accelerator ahead of its commercial release — a claim based on official company statements rather than external verification.

World Models and the Broader Shift

The summit also touched on what both companies call “world models” — AI systems designed to understand and simulate aspects of the physical world, enabling video generation that respects physics and spatial coherence rather than just producing plausible-looking but internally inconsistent footage. It’s an area that matters enormously for advertising, pre-visualisation, and eventually for interactive environments.

But the creative industries aren’t uniformly enthusiastic. Critics point out that the performance claims for Vera Rubin and Gen‑4.5 currently rest on corporate narratives rather than audited data. And there are legitimate concerns about what real-time generative video means for jobs in animation, visual effects, and editing, where automation could reduce demand for certain manual production tasks. Ethical questions around deepfakes, synthetic likenesses, and the copyright status of training data remain unresolved, with calls from practitioners for stronger regulatory safeguards before adoption accelerates.

Many independent filmmakers and small creative teams, on the other hand, see tools like Gen‑4.5 as genuinely levelling the playing field — subscription-based access from a standard computer, competing with studios that have vastly larger budgets. Educators in film and design are also watching closely, recognising that familiarity with AI video tools is becoming a baseline industry skill, even as some stress the importance of maintaining traditional craft alongside it.

So the picture is mixed. The technology is real and the capabilities are advancing quickly. The performance claims deserve scrutiny. And the social and legal questions haven’t been answered yet.

What This Means for Kent Residents

For freelancers and small creative studios in Kent working in film, advertising, social media, or design, subscription-based access to tools like Runway Gen‑4.5 via a standard internet connection could reduce reliance on expensive studio facilities and speed up pre-production work considerably. Local colleges and universities offering media, design, or computing courses may find that incorporating these tools into their curricula gives students a meaningful advantage in a job market where AI video skills are increasingly in demand. UK-wide regulation on synthetic media, deepfakes, and AI-generated content will ultimately govern how these tools can be used — and anyone in Kent adopting them professionally should keep a close eye on guidance from the UK government and relevant industry bodies.