'Post-Silicon' AI revolution... Using living neurons to challenge brain-based computing

The new turning point in AI technology is beginning with non-silicon “biological brains.” The startup biological computing company (The Biological Computing Co., hereinafter referred to as TBC) is advancing the commercialization of computing technology based on living neurons and has shown ambitions to replace the existing silicon-centric artificial intelligence infrastructure. The company recently secured $25 million in seed funding from Primary Ventures.

TBC was founded on three technological turning points: the rapid development of neuroscience, the limitations of current AI models, and the energy crisis triggered by climate change. Its founder and CEO, Alex Ksendovski, believes that the current generative AI models, which continuously optimize and scale computations indiscriminately to improve performance, have inherent limitations. He emphasizes, “Using a real brain for computation may seem contradictory, but it is the clearest answer.”

According to the company, TBC’s technology utilizes living neurons to directly encode image, text, and video data into neurons, decode neural activity in the brain, and convert it into high-dimensional representations. This approach can significantly reduce computational costs while enhancing the performance of generative AI models. Notably, compared to traditional silicon-based computing, this model can achieve high performance with less power and offers advantages such as continuous learning and enhanced memory.

John Pomeranets, co-founder and COO of TBC, stated, “This direction is the beginning of a major transformation in the post-silicon era,” and “We are building a new foundational layer for AI infrastructure that perceives the world and facilitates communication.”

As companies invest billions of dollars into AI infrastructure, technologies that improve computational efficiency are gaining industry attention. Analyst Holger Muller from Constellation Research commented, “AI architectures that mimic the structure of the human brain are a completely new approach, but if it can be proven that such models are truly scalable, the market will open explosively.”

Currently, TBC is using its technology to improve the video processing quality and structural capabilities of generative AI models and is testing its contribution to maintaining long-term consistency. Especially in applications such as optimizing visual autoencoders and long-term predictive video models, its potential has been demonstrated, increasing expectations for its technological commercialization.

TBC plans to launch a cloud-based hybrid neural-silicon cluster platform by 2027. This platform directly targets the AI infrastructure market, which is expected to grow to $1.7 trillion (approximately 2,448 trillion Korean won) by 2030.

Bryan Shakht from Primary Ventures stated, “Until now, AI has developed on silicon-based foundations, but this approach has reached a bottleneck,” and “Biological computing will become the core of next-generation AI infrastructure.”

TBC’s experiments are not limited to technological advances. It is considered the first case proposing a biological solution—“the human brain”—as an alternative to fundamentally address the current AI infrastructure’s issues with power consumption, bottlenecks, and model limitations.