๐Ÿ“ŠStalecollected in 63m

Brain Cells Power Data Centers

Brain Cells Power Data Centers
PostLinkedIn
๐Ÿ“ŠRead original on Bloomberg Technology

๐Ÿ’กBrain cells on silicon challenge Nvidia in data centersโ€”bio-AI hardware breakthrough.

โšก 30-Second TL;DR

What Changed

Cortical Labs launches brain-cell data centers

Why It Matters

Pioneers neuromorphic alternatives to GPUs, potentially slashing AI compute costs. Signals shift in data center tech toward bio-hybrid systems.

What To Do Next

Check Cortical Labs' site for biocompute platform demos and research papers.

Who should care:Researchers & Academics

๐Ÿง  Deep Insight

Web-grounded analysis with 6 cited sources.

๐Ÿ”‘ Enhanced Key Takeaways

  • โ€ขCortical Labs' CL1 is the world's first code-deployable biological computer, allowing users to deploy code directly to lab-grown neurons on custom silicon chips.[3]
  • โ€ขThe company's DishBrain system uses around 200,000 neurons on a microelectrode array to interface with software, demonstrated by learning to play Pong and Doom via feedback loops.[4][5][6]
  • โ€ขBacked by In-Q-Tel (US intelligence community VC) and Horizons Ventures, Cortical Labs plans revenue from biological computer sales by end-2024 and cloud services by early 2025.[1][5]
  • โ€ขApplications include drug testing for neurological disorders like dementia and epilepsy using human neurons to assess efficacy and side effects, partnering with firms like Bit Bio.[1]

๐Ÿ› ๏ธ Technical Deep Dive

  • โ€ขCL1 device integrates lab-grown human neurons (from adult stem cells via skin or blood) onto custom silicon chips with microelectrode arrays, housed in a shoebox-sized unit providing nutrients, air, and waste removal.[1][3]
  • โ€ขNeurons act as the CPU, receiving electrical signals from simulations (e.g., Pong or Doom) and responding with firing patterns interpreted as actions, enabling closed-loop learning in minutes without extensive training data.[4][5][6]
  • โ€ขDishBrain setup features ~200,000-1,000,000 neurons; a rack of 30 CL1 units offers low energy consumption compared to traditional AI hardware.[2][6]
  • โ€ขCortical Cloud provides remote 'Wetware as a Service' access to clusters (e.g., 120 units) for programming neurons on tasks.[1][3]

๐Ÿ”ฎ Future ImplicationsAI analysis grounded in cited sources

Biocomputing racks consume far less power than equivalent AI hardware for training models
A rack of 30 CL1 units maintains a very low electrical envelope, addressing the industrial energy costs of current AI infrastructures.[2]
Human neuron chips will replace animal models in neuroscience drug testing by 2025
Cortical Labs targets pharma partnerships like Biogen for in-vitro neuron assays on tasks like Pong to evaluate drugs for dementia and epilepsy.[1]
Biological cloud services launch commercially in early 2025
Cortical Labs plans clusters of 120 biological computers for remote access, with negotiations underway with US cloud providers.[1]

โณ Timeline

2022-10
Published high-profile study on DishBrain neurons learning to play Pong, drawing media attention for 'embodied sentience' claims.[4]
2022-11
Announced neurons (800k-1M cells) mastering Pong in five minutes via real-time learning.[6]
2024-12
Secured fresh capital from In-Q-Tel and Horizons Ventures to scale biological computers and begin sales.[1][5]
2025-01
Introduced CL1 desktop biocomputer and announced plans for Cortical Cloud biological computing platform.[1][3]
2025-12
Demonstrated neurons learning Doom via cloud API in under a week, showing enemy-seeking and shooting behaviors.[6]
2026-03
Launched two small brain-cell powered data centers in Singapore and Melbourne.[article]
๐Ÿ“ฐ

Weekly AI Recap

Read this week's curated digest of top AI events โ†’

๐Ÿ‘‰Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: Bloomberg Technology โ†—