Biological Data Centers Offer Alternative to Silicon GPUs
New neuron-based platforms aim to reduce data center power draw as the AI industry faces significant energy constraints.
Updated on Oct. 31, 2026 in Data Centers
Live Poll
Do you believe biological brain tissue can eventually replace traditional chips for artificial intelligence?
Biological computing systems are now operational, with Cortical Labs launching a 20-unit facility in Singapore in August 2026. Simultaneously, FinalSpark offers remote access to its neuro-based platform, which utilizes living brain organoids to perform computational tasks.
Why it matters
As the AI industry encounters power constraints from traditional silicon-based infrastructure, biological computing presents a potential efficiency alternative. These systems leverage living neurons that function on a fraction of the electricity required by standard hardware.
Cortical Labs' CL1 units contain 800,000 neurons, while FinalSpark utilizes 16 organoids of 10,000 neurons each. The systems operate on approximately 20 watts, significantly lower than the 1,000 to 1,500 watts consumed by a single Nvidia Blackwell GPU node.
The players
Cortical Labs
A Melbourne-based developer of biological computing hardware that integrates living neurons into digital systems.
FinalSpark
A Vevey-based biotech firm specializing in brain organoid integration for remote-access computing platforms.
The details
Cortical Labs employs a biOS system—a synthetic environment that delivers stimuli to neurons and modulates the simulation based on their electrical responses. FinalSpark utilizes brain organoids—three-dimensional clusters of cells grown in a lab to mimic brain structure—networked to electrodes. These electrodes facilitate the bidirectional communication required to provide remote computational access for users.
Timeline
2014: FinalSpark was founded in Vevey.
2019: Cortical Labs was founded in Melbourne.
March 2025: CL1 hardware debuted at Mobile World Congress.
August 2026: The Singapore biological data center went live.
The Tech Race
The emergence of these platforms marks a departure from the silicon-heavy trajectory currently defined by Nvidia Blackwell GPU nodes. These firms are now competing to prove that biological substrates can provide a scalable alternative to traditional data center power consumption.
FinalSpark currently provides remote platform access for a $500 monthly fee, while the Cortical Labs CL1 hardware is sold for $35,000. Researchers and developers can utilize these systems to experiment with biological processing as a potential alternative to cloud-based GPU workloads.
The takeaway
Biological computing offers a potential path to extreme energy efficiency, though it remains in the early, experimental stages of deployment. Watch for emerging performance benchmarks and the scalability of these Wetware as a Service models to determine if they can challenge silicon-based AI training.
Further reading
Explore the evolving physical infrastructure of modern computing in our Data Centers section.
Source note: This article includes information reported by Startup Fortune.
Live Poll
Do you believe biological brain tissue can eventually replace traditional chips for artificial intelligence?
