Bittensor, Numerai, and the Web3 AI Supercycle

How Incentives, Intelligence, and Capital Actually Compound

🎥 Watch / 🎧 Listen

Also listen on Spotify

Why this conversation matters

This episode brings together builders and investors who have actually shipped at the intersection of AI, crypto, and incentives.
Wei Xie was an early LP in Numerai and comes from a traditional quantitative finance background. He has seen firsthand how crowdsourced intelligence can be turned into durable, institutional-grade alpha and why most attempts fail.
Max Sebti is a subnet owner on Bittensor and the founder of Score, building vision AI infrastructure inside one of the most technically demanding AI ecosystems in crypto. He operates directly at the incentive layer where theory meets production.
Together, they break down how decentralized systems coordinate intelligence, why incentive design is the real bottleneck, and what it takes for AI-native crypto networks to matter beyond narratives.
This is a conversation about how edge is created, not how it’s marketed.

Hosts

Jack Leung
Founder of Supercycle. Focused on crypto markets, narratives, and emerging tech.

Daniel Andrade
Engineer and investor-operator focused on AI, robotics, and crypto market structure.

Guests

Wei Xie
NRN Agents Founder, Early LP
Deep background in quantitative finance and incentive design

Max Sebti
Founder of Score, Subnet Owner on Bittensor
Building vision AI infrastructure on decentralized rails

What we cover

Why Bittensor is best understood as Bitcoin for AI
How subnets turn intelligence into an incentive-aligned marketplace
Why Numerai worked where most crowdsourced alpha fails
Vision AI as the missing layer for robotics and real-world agents
Why intelligence is deflating but data is compounding
How decentralized incentive systems discover edge faster than institutions
Why the real AI TAM is human labor (~$50T annually)

Key moments

0:00 – Why human work will soon cost cents
1:30 – What Bittensor actually is
5:00 – Incentives, subnets, and why they are hard to design
7:00 – “Bitcoin for AI” and the halving analogy
11:00 – How subnets make money and attract capital
18:00 – Vision AI in sports, markets, and real-world systems
21:30 – What Numerai really solved
26:30 – Why institutions trusted Numerai with massive capital
30:00 – Vision, robotics, and VLAs (Vision-Language-Action models)
32:00 – The $50T human labor opportunity
35:00 – Data, intelligence, and why decentralization matters

Watch all Supercycle episodes on YouTube
Subscribe here to get future episodes in your inbox