Biological Computing Company plans neuron-derived AI rollout on AWS

The Biological Computing Company and Amazon Web Services announced a collaboration to offer early access to a neuron-derived optimization layer for an open-source text-to-video model on conventional AWS infrastructure.

The company says its software can generate video five times faster while reducing inference costs by 80% compared with the underlying model. Those figures remain company claims: the available sources do not identify the base model, publish the benchmark methodology or provide independent validation.

Customers would not need to operate biological hardware. The company says it records activity from living neurons in its lab, then turns those measurements into a proprietary software layer that runs on standard AI hardware. According to the company, that layer adds less than 0.1% to the size of the model it modifies.

The planned deployment path would put the optimized model on AWS Trainium chips and make it deployable through Amazon SageMaker AI. The companies also plan commercial distribution through AWS Marketplace, but the announcement did not establish that a Marketplace listing or general availability had begun.

The company’s early-access page asks prospective users to request a free invitation and says applicants will be contacted in fall 2026. The public first-party pages do not say how many customers have access or give a date for a broader release.