TBC's Bio-Algorithm: A New Frontier in AI Efficiency

SAN FRANCISCO, CA – July 30, 2026 – In an industry grappling with the voracious energy and computational demands of artificial intelligence, a San Francisco startup is presenting a radical solution drawn not from silicon, but from biology. The Biological Computing Co. (TBC) today unveiled a public proof of concept demonstrating that software algorithms derived from experiments on living neurons can dramatically improve the performance and efficiency of existing AI models. The company’s claims are striking: a 4.4-fold reduction in inference cost and a threefold increase in coherent output on a test video model. This development moves beyond AI merely being inspired by the brain to being directly derived from it, presenting a potential paradigm shift in how we optimize our most complex digital systems.

From Wetware to Software: A New Optimization Playbook

At the heart of TBC's announcement is its “Algorithm Discovery Platform,” a novel process that sidesteps the limitations of conventional digital optimization. For decades, AI efficiency has been pursued through a shared toolkit of software and hardware refinements like quantization, pruning, and architectural tweaks. TBC, founded by neurosurgeon-scientists Alex Ksendzovsky and Jon Pomeraniec, argues that this toolkit is inherently limited and that a vast, untapped library of computational strategies exists within biological neural networks.

Their method is as fascinating as it is ambitious. The process begins in a laboratory, where TBC’s interdisciplinary team of biologists and neuroscientists runs targeted experiments on living neural networks. They present these networks with specific computational problems and record their responses, analyzing the neural activity to identify novel and efficient computational principles. These biological insights are then translated by computational neuroscientists and engineers into lightweight software algorithms. Crucially, these algorithms are designed to integrate seamlessly into existing AI models and run on conventional GPU and cloud infrastructure. The living neurons are confined to the discovery phase; the end product is pure software.

This approach aims to harness the unparalleled efficiency of the brain, which operates on a fraction of the power required by today's supercomputers. “Until today, the industry optimized models with the same shared toolkit, while neuroscience offers a wealth of entirely new techniques,” said TBC co-founder and CEO Alex Ksendzovsky in the company's announcement. “This is the initial proof that principles derived from the brain, the most efficient computer in the world, can deliver real gains for applied AI.”

Putting Biology to the Test: The OASIS Demo

To validate its claims, TBC applied its optimization technology to Decart's OASIS 500M, an open-source interactive video model trained on Minecraft gameplay. The choice of a smaller, interactive model was strategic, allowing for rapid testing and a tangible, playable demonstration of the technology's impact. The results, according to the company, were validated through matched external testing conducted by AI infrastructure provider Bluesky Compute, a specialist in high-performance inference operations.

Under these controlled conditions, TBC’s neurally-optimized version of the OASIS model reportedly demonstrated a twofold improvement on a selected video-quality benchmark. More critically for businesses deploying AI at scale, it required approximately 4.4 times less inference cost to run. The model also produced more than three times as much coherent video—a key metric for generative models where outputs can often degrade into nonsensical noise. By partnering with a reputable infrastructure provider for verification, TBC adds a layer of credibility to metrics that might otherwise be dismissed as internal hype.

The public demo, available on the company’s website, serves as the first concrete evidence of this unique process yielding tangible results. It represents an early but significant proof point for a platform that, until now, has operated primarily behind closed doors.

The Billion-Dollar Question of Efficiency

TBC's entry comes at a critical juncture for the AI industry. As models grow in size and capability, their operational costs are spiraling. Inference—the process of running a trained model to generate predictions or content—is shifting from a secondary concern to a primary driver of operational expenditure. For many companies, these costs represent a formidable barrier to profitability, turning cutting-edge AI features into unsustainable financial burdens. Consequently, the market for AI efficiency solutions is experiencing explosive growth, with analysts projecting the AI biocomputing market to surpass $11 billion by 2030.

This economic pressure has created a fertile ground for innovation. While competitors like Australia’s Cortical Labs are developing hybrid biological-silicon hardware, TBC has made a strategic bet on a software-only deployment model. This allows potential customers to reap the benefits of biological discovery without overhauling their existing hardware stack, a significant advantage in a market built on NVIDIA GPUs and standard cloud infrastructure. This software-centric approach, combined with the technology's promise, has attracted significant investor interest, including a $25 million seed funding round led by Primary Venture Partners.

Navigating the Neuro-Ethical Frontier

The use of “living neural networks” inevitably raises profound ethical questions. The prospect of integrating biological matter into computational processes, even at the discovery stage, demands a high degree of transparency and ethical oversight. TBC appears to be tackling this challenge proactively. Company leadership has been clear that the neurons used in their labs are not sentient, lacking the structure, blood supply, or capacity to experience pain. According to research, the company has established “very clear guardrails” and employs bioethicists to help navigate this nascent field.

This distinction between a biological discovery process and a digital software product is central to TBC's ethical and commercial proposition. However, as the lines between biology and technology continue to blur, maintaining clear and publicly understood ethical frameworks will be crucial for the entire field. For TBC, demonstrating responsible stewardship is as important as demonstrating performance gains.

Building on the foundation of the OASIS demo, TBC is now applying its optimization process to a much larger, state-of-the-art text-to-video model, which it plans to launch as its first commercial offering later this year. This move targets one of the most computationally demanding sectors of the generative AI market. The company has provided a compelling proof point that its bio-derived algorithms can deliver. The next, and perhaps greater, challenge will be to prove that this novel discovery engine can be scaled into a reliable commercial product that consistently outpaces the relentless, silicon-based innovation of the established AI industry.