Biological Compute Emerges: Neurons on Chips Challenge Silicon AI

In February 2026, Australian startup Cortical Labs demonstrated a bio-hybrid computing system where around 200,000 human neurons grown on a silicon chip learned to play the video game Doom within a week. The system, called CL1, integrates living neurons derived from human skin cells with a microelectrode array that converts digital signals into electrical stimuli the neurons can interpret. The neurons receive inputs representing the game environment and generate electrical firing patterns that translate into actions such as movement and targeting enemies. This builds on Cortical Labs’ DishBrain experiment published in the journal Neuron in 2022, where lab-grown neurons learned to play Pong through closed-loop feedback.

The CL1 system reportedly operates using roughly 850–1,000 watts of power per rack, far below the megawatt-scale energy consumption of large GPU clusters used to train modern AI models. The company is also launching “Wetware-as-a-Service,” allowing researchers to remotely access living neuron arrays through cloud infrastructure. Investors include In-Q-Tel, the venture arm of the US intelligence community, alongside private venture funds.

This development points to a potential third computing paradigm alongside traditional CPUs and modern AI accelerators. Biological neurons naturally perform adaptive learning through synaptic plasticity and operate with extreme energy efficiency. Unlike artificial neural networks, which simulate brain processes mathematically, these systems use actual biological intelligence as the compute substrate. If scalable, bio-hybrid systems could excel in environments where AI struggles, such as rapid adaptation, sparse data learning, and uncertain real-world conditions.

The future compute stack may not be purely silicon. Hybrid biological-digital systems could emerge as a new layer of intelligence infrastructure.

AI infrastructure spending already exceeds $200 billion annually in data centres, driven largely by GPU demand from companies like NVIDIA and hyperscale cloud providers. Biological computing represents a radically different cost curve. If wetware systems prove scalable, they could disrupt segments of AI hardware focused on adaptive learning tasks. Investors should monitor neuromorphic computing, bio-hybrid processors, and synthetic biological intelligence startups. Defence and intelligence agencies are already exploring the technology due to its potential advantages in low-power autonomous systems and edge decision-making.

2026 → Early research adoption of Cortical Labs’ CL1 through cloud access and developer APIs.

2027–2028 → Scaling experiments beyond hundreds of thousands toward millions of neurons on chip.

Computing systems that use living neurons as the processing substrate rather than purely silicon circuits.

Why it matters

Sources: cell.com

The signal is the high ground. Hold it.
Subscribe for monthly tactical briefings on AI, defence, DePIN, and geostrategy.
thesixthfield.com