The Biological Computing Co. (TBC) announced a partnership with Amazon Web Services to deploy what it calls the world's first neuron-derived AI video model, claiming the technology delivers up to 5x faster generation and cuts inference costs by 80% compared to existing open-source video models. The announcement arrives just as OpenAI is shutting down its Sora API on September 24, 2026, reshaping the competitive landscape for AI video tools.

What Does It Mean to Build AI From Living Neurons?

TBC's approach sounds like science fiction, but the execution is grounded in practical software engineering. The company grows real brain cells in laboratory petri dishes, studies how they respond to stimulation, and translates those biological patterns into mathematical structures that can optimize conventional AI models. The key insight: biological brains evolved efficiency tricks over millions of years that silicon-based AI research hasn't yet discovered through traditional machine learning.

For the AWS partnership, TBC applied this biological-optimization technique specifically to video generation. The result is a lightweight software layer that sits on top of an existing video model, adding less than 0.1% to its total size while claiming dramatic performance gains. Importantly, customers never interact with living cells. The neurons stay in TBC's laboratory; only the software patterns they inspired ship to AWS infrastructure as ordinary code.

"Our partnership with AWS takes neuron-derived AI optimization to commercial scale," said Alex Ksendzovsky, CEO and co-founder of The Biological Computing Co.

Alex Ksendzovsky, CEO and co-founder of The Biological Computing Co.

How Does This Compare to Runway, Veo, and Other Video Tools?

The timing of TBC's announcement is striking. OpenAI's Sora, which helped define modern AI video generation, is being discontinued. The Sora web and app experience shut down on April 26, 2026, and the API is scheduled to close on September 24, 2026, just two days after TBC's announcement. This leaves the market largely split between Google's Veo 3.1 and Runway's Gen 4.5 as the primary active competitors.

Google's Veo 3.1 emphasizes realism, physical behavior, and native audio generation. It can produce dialogue, sound effects, and ambient audio as part of the video generation process, collapsing multiple production steps into a single generation. Runway, by contrast, positions itself as a creative software platform rather than just a model provider. Its Gen 4.5 model supports text-to-video and image-to-video generation with durations from two to ten seconds, but Runway's real strength lies in its broader production environment, including character performance tools, video editing, and workflow features.

TBC's neuron-derived optimization layer could theoretically apply to any of these base models, though the company has not disclosed which model it modified for the AWS partnership. The claimed 5x speedup and 80% cost reduction would be significant advantages in a market where inference costs and generation time directly impact creator productivity and platform profitability.

What Are the Key Performance Claims?

TBC is publishing several specific performance metrics, though independent verification is still pending. Here's what the company claims:

• Generation Speed: Up to 5x faster than the unmodified base model, though this figure has not been independently benchmarked by third-party labs or academic researchers.
• Inference Cost: 80% lower than the base model, representing a significant reduction in the computational expense of running the video generator, though again unverified by external parties.
• Software Overhead: The optimization layer adds less than 0.1% to the base model's size, a strikingly small footprint for such large claimed performance gains.
• Output Quality: TBC describes improved output quality compared to the base model, but has not published side-by-side visual comparisons for independent evaluation.

In an earlier research effort called OASIS, TBC reported a twofold improvement on a video-quality benchmark, roughly 4.4x lower inference cost, and more than three times as much coherent video output compared to a base model, under conditions TBC itself defined. The AWS-backed product appears to be a commercialized successor to that proof of concept, though TBC has not published a direct technical comparison.

Why Independent Verification Matters

Every major outlet covering the announcement flagged the same critical gap: TBC has not published independent benchmarks for its 5x speed and 80% cost figures, and neither has AWS. The company has also kept secret which base model it modified and the methodology used to measure performance improvements. This is not unusual for a startup launch, but it means the headline numbers should be read as marketing claims rather than confirmed results until third-party labs or customer case studies test them independently.

Buyers evaluating TBC's product will likely want to run their own workload comparisons before committing budget. The unidentified base model also makes apples-to-apples comparison against known benchmarks difficult. If the 5x speedup and 80% cost reduction are validated, it would represent a genuinely notable result in efficient inference research, an area where companies pursuing custom silicon and quantization techniques have historically had to trade off model size, speed, and output quality against each other.

How to Evaluate AI Video Tools for Your Workflow

As the AI video market consolidates around fewer active products, creators and teams need a framework for choosing the right tool:

• Production Environment: Consider whether you need a broader set of editing and workflow tools (favoring Runway) or a single-generation model optimized for specific tasks like realistic dialogue-driven scenes (favoring Veo 3.1).
• Realism and Audio Requirements: If your project depends on physical realism, native audio, and high-resolution output, Veo 3.1's strengths in these areas may outweigh other considerations, even if it generates shorter clips.
• Long-Term Platform Stability: Verify that the tool you choose will remain available throughout your project timeline. Sora's discontinuation is a cautionary example of relying on a platform that may not persist.
• Independent Benchmarking: For emerging tools like TBC's neuron-derived model, request or conduct your own performance tests before integrating them into production workflows.

AWS's involvement with TBC signals confidence in the technology, but also reflects the company's broader strategy to optimize inference costs on its Trainium chips and SageMaker AI platform. The partnership will make TBC's model available through the AWS Marketplace, potentially lowering barriers to adoption for teams already using AWS infrastructure.

The discontinuation of Sora and the emergence of specialized competitors like TBC suggest the AI video market is maturing beyond the "one model to rule them all" phase. Instead, creators are increasingly choosing tools based on specific strengths: Veo for realism and audio, Runway for production workflow, and potentially TBC for cost-efficient inference at scale. The real competitive advantage now lies not just in model quality, but in how well a tool integrates into a creator's existing workflow and how reliably it remains available over time.