Z Potentials | Exclusive Interview with the Millennial Investor Behind DeepSeek in the US: Where Does He See the Next Trillion-Dollar Company?
🚀 AI Investment Insights from Brian Zhan
Early investments in Reflection AI, Skild AI, Dyna Robotics, Periodic Labs, and several next-generation AI infrastructure companies have shaped the frontier before consensus emerged.
Image source: Official website
---
🔍 Key Takeaways
- Foundation models will expand beyond language — RL, robotics, and AI for Science will become breakthrough levels of intelligence.
- Next-gen Agents aren’t just “smarter ChatGPTs” — they will have systemic cognitive structures that remember, collaborate, and learn from deployment.
- True opportunities emerge before they’re obvious — the next $100B company could be solving a problem no one is tracking yet.
---
🎙 Exclusive Interview: Brian Zhan, Partner at Striker Venture Partners
ZP: You’ve joined forces with Max Gazor, a four-time Midas List honoree, after building a portfolio featuring Reflection AI, Skild AI, Dyna Robotics, Periodic Labs, Lepton, Voyage, and LanceDB — most before they were mainstream. What did you see first?
Brian:
I look for top-tier technical talent tackling problems most consider “nearly impossible.”
Example: Reflection AI assembled elite RL researchers when everyone else focused on Transformer scaling laws. RL is foundational to reliable Agent models — allowing them to make mistakes, self-correct, and improve through interaction for true reasoning ability.
For infrastructure plays like Lepton, Voyage, and LanceDB, our edge was spotting critical AI base layers before the market recognized their necessity.
---
🤖 Betting Early on Robotics
ZP: You backed Skild AI and Dyna Robotics when VC considered robotics a “capital graveyard.” Why?
Brian:
I believed robotics’ GPT moment had arrived.
Key changes:
- Foundation models matured
- Massive robotics datasets (e.g., Open X-Embodiment) became available
- Sim-to-real gap bridged by capable teams
Skild is building a model that generalizes to any task, environment, and robot hardware, promising paradigm-shifting cost reductions.
Dyna Robotics takes a more vertical approach, enabling rapid expansion in select fields — both project types represent architectural breakthroughs, not just performance tweaks.
---
🔬 AI for Science: A New Era
ZP: Why is AI for Science real now?
Brian:
True AI-for-science will arrive 2025–2030.
Previously, models only matched patterns in “text about science.” Now they can reason within scientific concept space: rediscover theorems, interpret unpublished data, propose valuable experiments, and form novel mechanistic hypotheses.
This is concept space traversal, not “faster database lookups.”
---
🌿 Digital Biological Systems
Two core meanings:
- Multi-scale unified perspective — model understands biology from molecular to tissue level in a dynamic system.
- Efficient combinatorial navigation — rapid discovery in protein design, materials science.
Compound scientific intelligence accelerates the entire loop: literature → hypothesis → experiment → analysis → follow-up — shrinking months-long cycles to hours.
---
🧠 Why Current Agents Still Fall Short
Brian:
Agents fail due to:
- Insufficient intelligence
- Immature true multimodality
- Weak computer-use skills
- Lack of continuous learning/self-improvement
Most crucially, Agents cannot share written records or transfer knowledge — each “relearns the world from scratch.”
---
🛠 Teams to Watch
The best are attacking system-level challenges:
- New cognitive architectures
- Long-term context retention
- Continuous world-model updates
- Knowledge sharing across Agents
Once solved, automation will leap forward.
---
🎯 Striker’s High-Conviction Model
- Only 10 companies per fund
- Up to $30M per project at an early stage
- Extreme selectivity → deep collaboration and shared journey from day zero
---
🕵️♂️ Spotting “Build Before It’s Obvious”
Ask:
- What constraint is being broken?
- Why is this viable now, but impossible before?
If founders clearly explain the specific breakthrough, it’s worth attention.
---
💡 Most Non-Consensus Bet
Next $100B company:
- Stealth technical team
- Problem not tracked by VCs
- No label, no market narrative — pure technical depth and purpose
---
📣 Advice to Founders
- Act before market validation
- Hire curiosity-driven exceptional talent
- Define your own problem space
- Avoid distraction by competitors
---
🌐 Platforms Empowering AI Creators
Tools like AiToEarn官网 are advancing open-source global AI content monetization, integrating AI generation, multi-platform publishing, analytics, and model rankings — connecting creators across Douyin, Kwai, WeChat, Bilibili, Xiaohongshu, Facebook, Instagram, LinkedIn, Threads, YouTube, Pinterest, and X.
---
🚀 Recruiting next group of interns
🚀 Seeking creative Gen Z entrepreneurs
---
📜 About Z Potentials
Opportunities for entrepreneurs, innovators, and creators to leverage tools like AiToEarn官网 for AI-powered content creation, cross-platform publishing, and monetization.
Read Original | Open in WeChat
---
Would you like me to also convert this into a concise executive summary version so decision-makers can read the insights in under 3 minutes? That way, you’ll have both the detailed interview and a quick-read briefing format.
