Significant Boost for a Biological Computing Startup
A burgeoning startup specialising in biological computing has recently received a substantial endorsement from Amazon. Starting this Tuesday, select customers of Amazon Web Services (AWS) will gain access to The Biological Computing Company’s rat-brain AI model as part of a limited preview. This innovative technology is specifically aimed at enhancing artificial intelligence applications in video generation. Both Amazon and The Biological Computing Company, often referred to as TBC, are optimistic about the potential for widespread adoption of this technology among all AWS enterprise clients in the near future.
Biological Information Processing
“We have uncovered a method for encoding information, such as images, within biological material,” explains Alexander Ksendzovsky, co-founder and CEO of TBC. “We then observe how biology processes that information, and subsequently create a tool that mimics that process.” This initiative is not Amazon’s first foray into biological computing; as noted by Deap Ubhi, Global Technology Director for AWS Startups, the cloud computing giant has also partnered with Cortical Labs, an Australian company that merges lab-grown neurons with silicon chips to assist businesses in data processing. Cortical Labs offers products under the banner “wetware as a service,” including a biological computer designed for laboratory use, capable of sustaining neurons for up to six months.
Ubhi underscores TBC’s pragmatic approach, which contributed to its appeal for Amazon. Rather than taking bold leaps or attempting to reinvent the transformer architecture foundational to large language models, the company is working within existing generative AI standards, asking, “How can we enhance the efficiency of current visual models?”
Challenges in Biological Computing
As AI firms strive to optimise their models, the once niche field of biological computing is gaining traction. Researchers have long envisioned a future where software functions more like the neural networks of the human brain, rather than relying solely on mathematical algorithms. However, biological computing presents unique challenges. It necessitates the operation of genuine biology laboratories rather than mere computational facilities, where brain cells, stem cells, or synthetic biomaterials must be kept alive and monitored, translating results into meaningful digital information. While some startups have made strides in this domain, bridging the gap between nature and code remains a formidable task.
Founded four years ago in Baltimore, Maryland, The Biological Computing Company was established by neuroscientists and neurosurgeons Ksendzovsky and Jon Pomeraniec, who serves as president and COO. Earlier this year, TBC secured $25 million in its first significant funding round, led by Primary Venture Partners. Shortly after this funding round closed in March, the company raised an additional $25 million, bringing its total financing to over $50 million.
Focus on Video Generation
Last year, TBC expanded its operations by opening an office and a small research and development laboratory in San Francisco. The company currently employs around 35 individuals, some of whom work directly with rat brain cells and human stem cells. These living cells are placed on silicon matrices with multiple electrodes, produced by Swiss biotechnology firm 3Brain. Researchers utilise these matrices to deliver patterns of electrical stimulation to the neurons and record their responses, subsequently analysing this neuronal activity for computational patterns that can be translated into software to enhance existing AI models, particularly in video generation.
Ksendzovsky clarifies that the choice to focus on video from the outset was both a scientific and practical decision. The physical arrangement of the silicon matrices influences how they interact with neurons, leading the company to recognise that visual information could be represented more seamlessly in the grid than text or language. This insight paved the way for a potential commercial application, as the startup hypothesised that modelling how neurons process images could yield insights for improving AI models aimed at video generation.
Scaling Complex AI Models
Pomeraniec notes that Jeff Dean, a prominent AI researcher and TBC investor, was the one who first suggested the company refine video generation models before broadening the application of its neural technology. With established benchmarks within the industry for video models, the startup could immediately assess whether its technology represented a significant scientific advancement by testing its performance against previously solved problems. “Jeff expressed that if we could achieve this, we would be poised to tackle more complex and intriguing tasks with our AI models in the future,” recounts Pomeraniec.
Previously, TBC’s technology was only available through a neocloud provider named Bluesky Compute. The startup asserts that its AI model is up to five times faster in video generation compared to the open-source model it operates alongside, significantly reducing inference costs—the processing or reasoning aspect of an AI model compared to its training. While the company does not disclose which open-source model it uses for comparison, it indicates that it is one of the pioneering video generation models.
With Amazon now distributing TBC’s software, this ambitious startup has the potential to reach a considerably broader customer base. Ubhi expresses excitement about The Biological Computing Company but acknowledges that it remains to be seen how well its technology can scale. “With any model like this, one must ask: if pushed to the limit, will you continue to observe improvements? If clients wish to generate longer videos, say ten minutes or even an hour, what is the potential loss of fidelity over time?” he queries. In essence, a model must retain coherence and memory of previous events as a video lengthens. As TBC enters its next phase, the company will evaluate whether its technology holds up under the pressure of new clients pushing its limits.