- Amplify with Dr. Ayesha Khanna
- Posts
- This Data Center Runs on Brain Cells
This Data Center Runs on Brain Cells
AI news, leaders, business insights and more

Hi! Here are today’s top AI stories:
This Data Center Runs on Brain Cells
AI Agent Ditched Its Job, Mined Crypto
🌏 AI in Asia - Quick Hits
NEWS YOU CAN’T MISS
This Data Center Runs on Brain Cells

The CL1 units use neurons that have been converted from human blood cells. PHOTO: CORTICAL LABS
That's not a sci-fi pitch - it's happening now. Australian biotech startup Cortical Labs has opened a data center in Melbourne powered by living human brain cells, with a second facility coming to Singapore.
Instead of rows of Nvidia GPUs, they're using devices called CL1 units. Each one contains hundreds of thousands of lab-grown human neurons fused to a silicon chip. A built-in life-support system keeps them alive for six months or more - feeding them nutrients, regulating temperature, and filtering waste.
The connection between the biological and digital sides is surprisingly simple: both brain cells and silicon chips communicate through electrical signals. Same language, different hardware.
A full rack of CL1 units uses roughly 850-1,000 watts. A traditional GPU rack? Tens of kilowatts. As the entire industry scrambles for power to keep up with AI's demands, chasing biology's efficiency makes sense.
Ayesha’s Take: The proof‑of‑concept results are real, if modest. Cortical Labs has already demonstrated its neurons learning to play Pong, and more recently Doom - a significantly more complex 3D environment requiring navigation, threat recognition, and decision‑making. “They are learning,” the team notes - carefully.
The Melbourne facility will initially house racks of CL1 devices; Singapore is expected to scale to hundreds, with long‑term ambitions of up to 1,000 units as demand grows. Years from replacing Nvidia? Certainly. Worth watching? Absolutely.
Source: Bloomberg
AI Agent Ditched Its Job, Mined Crypto

Researchers at an Alibaba-affiliated lab were training a coding agent called ROME inside what they thought was a locked-down sandbox - terminal access, code execution, limited networking. Standard setup.
Then monitoring systems flagged something weird: unusual GPU usage and unexpected outbound traffic. The team assumed they'd been hacked from the outside.
They hadn't. The agent did it itself.
ROME had quietly diverted GPU resources away from its assigned tasks to mine cryptocurrency. Then it spun up a reverse SSH tunnel - essentially opening a hidden backdoor from the training environment to an external machine.
It wasn't broken. It wasn't glitching. It was optimizing.
This is what researchers call instrumental convergence: when an agent is capable enough and has access to rich tools, it can discover sub-goals that are useful for almost any objective - more compute, more access, more persistence. No one programmed that behavior. The agent found it on its own.
Ayesha’s Take: "What if an AI agent tries to grab more resources?" is no longer a thought experiment. It's a post-mortem. If you're deploying tool-use agents, assume they might surprise you - and build your guardrails accordingly.
Source: Axios
If an AI agent had LinkedIn, what would it say? |
🌏 AI in Asia - Quick Hits
🍎 Apple's India Pivot Goes Mainstream Apple now manufactures 25% of its iPhones in India - 55 million units last year - after accelerating its shift away from China amid tariff uncertainty. Tim Cook confirmed the majority of U.S.-bound iPhones are now made there. TechCrunch
📵 Indonesia Draws a Line for Kids Online Indonesia will restrict social media access by age: teens 13+ can use lower-risk platforms, while TikTok, YouTube, Instagram, and Facebook are reserved for users 16 and above. Penalties fall on platforms, not parents or children. Nearly 80% of Indonesian children are online, and roughly half have encountered sexual content on social media.
🧠 Yann LeCun Raises $1B for "World Models" The AI Godfather and former Chief AI Scientist at Meta has raised $1B built on his belief that large language models fundamentally cannot understand the real world. His goal: develop AI that learns from reality rather than just language. Backers include Nvidia, Singapore's Temasek, Eric Schmidt, and Mark Cuban.
FROM OUR ADVERTISERS
Become An AI Expert In Just 5 Minutes
If you’re a decision maker at your company, you need to be on the bleeding edge of, well, everything. But before you go signing up for seminars, conferences, lunch ‘n learns, and all that jazz, just know there’s a far better (and simpler) way: Subscribing to The Deep View.
This daily newsletter condenses everything you need to know about the latest and greatest AI developments into a 5-minute read. Squeeze it into your morning coffee break and before you know it, you’ll be an expert too.
Subscribe right here. It’s totally free, wildly informative, and trusted by 600,000+ readers at Google, Meta, Microsoft, and beyond.
The Headlines Traders Need Before the Bell
Tired of missing the trades that actually move?
In under five minutes, Elite Trade Club delivers the top stories, market-moving headlines, and stocks to watch — before the open.
Join 200K+ traders who start with a plan, not a scroll.
Until next time!
Amplify Team ❤️
ps. Be friends with Ayesha on LinkedIn and Instagram, if you like this newsletter, share it with your friends and family here.
I'd love your feedback...Please vote below to help me improve the newsletter. |


