Biological Computing Co. Develops Faster, Cheaper AI Video Model with AWS

The Biological Computing Co. (TBC) has developed a text-to-video model that operates five times faster and at 80% lower inference costs compared to its base model, Oasis, which has 600 million parameters. This advancement follows TBC's $25 million funding round in February 2026 and a partnership announcement with Amazon Web Services (AWS) on September 22, 2026. TBC's proprietary software layer, known as the Neural Dynamics Adapter, increases the model's size by just 0.1% while improving performance by 15% over conventional fine-tuning and 5% over LoRA adaptation. The adapter enhances image quality, achieving a 19% improvement in differential entropy, allowing sharper object edges and stable spatial relationships. The underlying technology utilizes a chip with 4,096 electrodes to grow cortical nerve cells, with the neurons remaining in the lab. TBC claims that the speed gain results from reducing compute steps per frame without sacrificing quality. This innovation could significantly impact video generation applications, making them more efficient and cost-effective.