Open Weight Models Reach Half of Tokens, 13% of Spend in Unexpected Shift
Open weight models now account for roughly half of all AI inference tokens served in 2026, data shows, rewriting lab economics and China's model lead.
Murtuza specializes in blockchain and digital asset intelligence. He builds AI-assisted analytical systems that turn on-chain, derivatives, ETF-flow, macro, and regulatory signals into structured, machine-readable intelligence. Trading desks, data-licensing clients, institutions and policymakers use these outputs to make decisions.
Open weight models now account for roughly half of all AI inference tokens served in 2026, data shows, rewriting lab economics and China's model lead.
Frontier labs now train models on outcomes, not text. Here's how agentic RL became the core method behind 2026's biggest model releases, and its real limit…
Hyperscalers will spend $775-800B on AI infrastructure in 2026. Fathom examines why grid capacity, not chips, is now the real constraint.
Nvidia's reported $12.9B Hugging Face acquisition arrives weeks after a breach, raising real questions about who controls open source AI infrastructure.
Anthropic Custom Chip plans include early talks with Samsung, while Microsoft chip talks could reduce costly Nvidia dependence at scale.