Andrew Feldman

Andrew Feldman appears in Open Source Wins, AGI Is Here, and Scorsese’s AI Toolkit with CEOs of Cerebras & Black Forest Labs as the Cerebras CEO interviewed by Jason Calacanis and the All-In hosts. He frames AI infrastructure as a capacity-constrained industrial buildout: data centers, power, chips, and capital are racing to meet current model-company and hyperscaler demand.

His strongest technical claim is that reasoning is inference. Long-running reasoning systems consume many internal tokens, so faster chips can make multi-hour or multi-day reasoning loops practically usable rather than only theoretically better.

Feldman also links open-source models to model sovereignty. He expects enterprises to route ordinary and sensitive workloads across frontier, cheaper, open, domestic, and custom models instead of treating one provider as the only answer.

Connections

• Cerebras - company context.
• AI Inference Cost Structure, Low-Latency Inference Chip, and AI Chip Specialization - hardware and inference-economics branch.
• Token Maxxing, Unlimited Token Workflow, and Loop Maxxing - reasoning-token and recursive workflow branch.
• Open Source AI Models, Model Sovereignty / 模型主权, and Model Routing Cost Control - open-model and enterprise-selection branch.
• Frontier Model Release Governance, AI Cyber-Defense Utility, and AI-Enabled Vulnerability Discovery - staged-release and cyber-risk branch.
• AGI Narrative and AI Abundance Narrative - AGI-threshold and abundance claims made in the episode.