[SYS_OK] MEA_0 ACTIVE -- 862 Hz mean spike rate -- donor 924A consent verified -- 37.0 C -- rack draw 914 W -- 0.85 Gop/W -- MEA_1 ACTIVE -- 743 Hz -- donor 482B consent verified -- 37.1 C -- perfusion nominal --
OrganoidGrid
Platform profile · Cortical Labs

Cortical Labs CL1

The CL1 is a desktop biocomputer that runs cultured cortical neurons on a planar high-density electrode array, packaging incubation, perfusion, and electrophysiology into one box so a software team can address living tissue without building a wet lab.

In March 2026 the CL1's commercial launch produced a search spike an order of magnitude above its baseline. Most of that interest is people meeting the idea of a "biological computer" for the first time. This profile is for the rest: what the unit is, what it can and cannot do, and which numbers to verify before you wire a budget to them.

A compact desktop neural-culture bioreactor unit with a glass chamber and faint green electrode glow inside, on a black background.
The CL1 collapses an electrophysiology rig into a desktop unit. Imaging is illustrative.

What is the Cortical Labs CL1?

The CL1 is a self-contained unit that grows human iPSC-derived cortical neurons (or primary rodent neurons) on a CMOS multielectrode array and keeps them alive while a host computer stimulates and records them. The selling point is integration: the incubation, the microfluidic perfusion, the digitizers, and the gas handling are inside one footprint, so the barrier to entry drops from "build a cleanroom" to "find bench space and aseptic technique."

~1,024active MEA channels
20-30 kHzper-channel digitization
37.0 Cchamber setpoint
~850-1000 Wrack-level draw

Figures track vendor documentation (accessed 2026-06-12) and are operational estimates; confirm against current specifications before relying on them.

How much power does it really use?

The desktop unit itself draws under about 100 watts. The number that matters for a data center, though, is the rack: with perfusion pumps, gas mixing, low-noise amplifiers, digitizers, and a processing host, a full configuration sits in the rough range of 850 to 1000 watts under continuous operation. The neurons are not the load. Their life support is. Any "ultra-efficient biocomputer" claim that quotes the cell-level microwatts and omits the rack is, at best, incomplete; the honest accounting lives on the biocomputing primer.

How the HD-MEA interfaces the neurons

The computational surface is a CMOS planar array of platinum-black or gold microelectrodes at pitches under about 20 micrometers, close enough for near-single-cell resolution. Each electrode both listens and speaks. Recording captures the microvolt-scale extracellular deflections of firing neurons through low-noise amplifiers and fast analog-to-digital conversion. Stimulation injects charge-balanced biphasic pulses, on the order of tens of microamperes, to drive firing without electrolyzing the medium or eroding the electrode.

Cross-section of a microelectrode array interfacing cultured neural tissue A planar electrode array at the base, an electrical double layer at each electrode, neural tissue above with neurons and synapses, and bidirectional arrows showing stimulation downward and recording upward. Neural tissue (organoid) CMOS electrode array soma axon synapse double layer stimulate (uA) record (uV)
The CL1's planar HD-MEA records extracellular field potentials and delivers charge-balanced microampere stimulation through its electrodes. Charge-balanced biphasic pulses protect both the electrode and the tissue.

How do you program it?

You do not program the tissue so much as wire a loop around it. Cortical Labs exposes a Python SDK that treats the culture as a non-deterministic node: you map data onto specific electrodes as stimulation patterns, capture and spike-sort the response, and route an outcome straight back into feedback stimulation. That low-latency closed loop is the same machinery the DishBrain study used to train cultures on Pong.1

Signal acquisition and feedback pipeline A left-to-right chain of processing stages from the electrode array through amplification, digitization, spike sorting and decoding, then back to the stimulator. Encode data to electrodes Stimulate biphasic pulses Record spike-sort Decode action Feedback reward/entropy
The CL1 software loop. Input data is mapped to a stimulation pattern, delivered to the culture, recorded and spike-sorted, decoded into an action, and converted into feedback stimulation, all at latencies low enough for real-time closed-loop control.

Availability and what you need to run one

The CL1 is sold on an inquiry basis, with pricing handled per buyer rather than published. Even though the unit is self-contained, it grows human cells, so a site needs at least Biosafety Level 1 practice: a laminar-flow hood for media prep, aseptic technique, and proper biological waste handling. Cortical Labs supplies training and starter cultures. Budget for the laboratory discipline, not just the hardware.

Frequently asked questions

What is the Cortical Labs CL1?

A desktop biocomputer that runs cultured cortical neurons on a high-density electrode array, with incubation, perfusion, and electrophysiology integrated into one unit.

How much power does the CL1 use?

The desktop unit draws under about 100 watts, but a full rack with life support and data acquisition is in the rough range of 850 to 1000 watts. The neurons are not the load; their life support is.

How do you program the CL1?

Through a Python SDK that maps data onto electrodes as stimulation, records and spike-sorts the response, and routes outcomes back as feedback, forming a closed loop.

Can I just buy one?

It is sold on an inquiry basis, and a site needs Biosafety Level 1 practice to handle the human cells. Pricing is not published.

Profile changelog

Dated revisions to this profile
DateChange
2026-03-15Initial profile after the CL1 commercial-launch news cycle.
2026-06-12Reframed power budget honestly (rack vs unit), added HD-MEA and SDK detail, added citations.

References

  1. Kagan BJ, et al. In vitro neurons learn and exhibit sentience when embodied in a simulated game-world. Neuron. 2022;110(23):3952-3969. doi:10.1016/j.neuron.2022.09.001. Accessed 2026-06-12.
  2. Mueller J, et al. High-resolution CMOS MEA platform to study neurons at subcellular, cellular, and network levels. Lab on a Chip. 2015;15(13):2767-2780. doi:10.1039/C5LC00133A. Accessed 2026-06-12.