Jen Zhu

Jen Zhu

writing, researching, thinking, learning, and building

@jenzhuscott LinkedIn
About

Cofounder/CEO of Power Dynamics — designing, integrating, building ESS with advanced high-capacity cooling tech for data centers to make our digital infrastructure more sustainable with higher ROI. Building an investment thesis in robotics and AI-fueled life sciences. Chair of the SUI Foundation Board. Cofounder and Socratic Seminar Moderator of the Tech Fellowship at The Aspen Institute.

Applied mathematics and computer science. Perpetual student of quantum physics, philosophy, and history. Work spans AI infrastructure, energy systems, China's tech ecosystem. Researching the philosophy of intelligence. Mother. Motherly. Mothering. Based on airplane.

Machine Hallucination
01 Δ Energy & Infrastructure Building
Maroon Bells
Trippy Fractal
02 λ AI Economics Research
AI Hallucination
03 China Tech Analysis
West Lake
Synapses
Visual
04 ψ Philosophy of Intelligence Study
Wind Farm
05 Web3 & Protocols Practice
Data Human
Ex Machina
06 Science Communication Writing
07 Applied Mathematics Foundation
Nature
08 Quantum Physics Inquiry
Trippy Boho
Gut
Leaves
09 Socratic Method Method
Raised by Wolves
14 Grid Edge Research
AI Abstract
10 Power Systems Building
Bio Synapse
Datacenter Power
11 Data Sovereignty Advocacy
Sci-Fi Motion
15 Lifespan Science Investing
12 Cooling Systems Engineering
Google Datacenter
13 Learning Practice
Latest Thoughts @jenzhuscott
2026.06.04
The three most important infrastructure problems right now aren't AGI, aren't alignment, aren't agent swarms. They're cooling, power density, and grid interconnection. The rest is science fiction until we solve heat.
2026.06.02
Tencent was "behind" in AI. Today shares jumped 10% on news of an agent coming to WeChat — 1.4 billion users, mini-programs, payments, commerce. How is that for instant scaling.
2026.06.02
The AI industry talks endlessly about scaling laws. Zero mention of the second law of thermodynamics. Every GPU is a space heater with a marketing budget. This is why I'm working on cooling infrastructure. Too critical not to pour the best brains and resources on it.
2026.06.01
DeepSeek V4 shipped open weights at performance parity with the best closed models. Qwen matched it weeks later. When capability gaps close this fast, the moat isn't in model architecture — it's in infrastructure and distribution.