Australian biotech company Cortical Labs has released the CL1, which it calls the world's first commercially available "biological computer." The platform combines living human neurons with a silicon chip, providing information processing via electrical feedback with a latency of less than a millisecond, IEEE Spectrum reports.
The CL1 device is designed primarily for neuroscience and biotechnology research, including disease modeling and drug development. The platform contains 800,000 lab-grown human neurons reprogrammed from skin or blood cells from adult donors. These cells remain viable for up to six months and are constantly maintained by a built-in life support system that controls temperature, provides nutrients, filters waste, and maintains environmental balance. In real time, neurons generate and receive electrical impulses, adapt to stimuli, and “learn” with each interaction.
"On one view, [the CL1] could be regarded as the first commercially available biomimetic computer, the ultimate in neuromorphic computing that uses real neurons," says theoretical neuroscientist Carl Friston of University College London. "However, the real gift of this technology is not to computer science. Rather, it’s an enabling technology that allows scientists to perform experiments on a little synthetic brain."
The first 115 CL1s will begin shipping this summer at a price of $35,000 per unit, and if you order a rack-mount of 30 devices, the price drops to $20,000 per unit. Cortical Labs also offers a cloud-based wetware-as-a-service for $300 per week per unit, which provides remote access to its in-house cell cultures.
Each rack of CL1 devices consumes between 850 and 1,000 watts, which is significantly less than the tens of kilowatts typically required for data centers running AI tasks. This makes the platform suitable for long-term laboratory studies. “We inject small electrical pulses that represent bits of information, read out neural responses, and then write new data into the culture, all in cycles lasting less than a millisecond,” explained Dr. Brett Kagan, chief scientific officer at Cortical Labs.
Scientific institutions and biotech companies have already expressed great interest. Potential applications include screening drugs for neurological disorders such as epilepsy and Alzheimer's disease by observing how neurons taken from patients process information in real time. In a recent peer-reviewed study, CL1 restored functional learning in laboratory neuronal cultures that model epilepsy: application of antiepileptic compounds allowed previously damaged cells to "learn" more effectively.
In addition to biomedical research, Cortical Labs is exploring partnerships in neurocomputing, artificial intelligence acceleration, and even unconventional ventures like Bitcoin mining. The startup has received inquiries from music and entertainment companies looking to combine biological computing with experimental art. One early proposal involved connecting CL1 to the late composer’s neurons, though that particular collaboration never materialized.
CL1 is based on an earlier Cortical Labs prototype called DishBrain, in which cultured neurons learned to play the arcade game Pong in a simulated environment. A paper published in 2022 in the journal Neuron reported that the neurons self-organized in minutes to track and respond to on-screen stimuli, in some cases outperforming deep reinforcement learning algorithms in sampling efficiency and adaptability.
The production increase takes advantage of the exponential growth of biology: “While it takes a significant amount of effort to produce 100,000 neurons, scaling up to a million cells costs only a little more,” Kagan said. The startup, which has raised more than $11 million from investors including Horizons Ventures and In-Q-Tel, says hundreds of millions of neurons are realistic before the challenges of culturing cells on a large scale arise.
Cortical Labs emphasizes that CL1 buyers must obtain institutional ethical approval and maintain appropriate cell culture facilities. The devices are not intended for unregulated use. The company’s long-term goal is to create "bioengineered intelligence" that could one day surpass both natural neural networks and silicon machines in adaptability and efficiency. "Any sufficiently advanced machine becomes indistinguishable from biology," says Kagan. "We aim to harness the adaptive, self-repairing properties of living cells to create more resilient computing platforms."