According to God of Prompt on X, AMI Labs raised $1.03B to pursue Yann LeCun’s world model architecture, positioning it as a thesis bet against scaling transformer LLMs that focus on next‑token prediction (as reported by AMI Labs and God of Prompt). According to AMI Labs, the company aims to build systems with persistent memory, reasoning, planning, and controllability, operating from Paris, New York, Montreal, and Singapore. As reported by AMI Labs, the round is co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions, signaling institutional support for Path B: interactive world-model learning over Path A: larger LLMs. According to God of Prompt, if world models scale, prompt engineering practices and tooling could shift toward agents that learn via interaction, offering business opportunities in robotics, autonomous systems, simulation platforms, and memory-centric AI infrastructure.

According to Yann LeCun on X, citing Johns Hopkins provost Denis Wirtz, federal funding for US biomedical research has sharply contracted, with NIH allegedly down 80% in new grants and 70% in total awarded dollars since October 1, 2025, prompting lab closures and talent exits (source: X posts by @ylecun and @deniswirtz). As reported by these X posts, this funding shock jeopardizes AI-driven drug discovery, clinical ML pipelines, and translational bioinformatics that rely on NIH-backed datasets, compute, and multi-institution consortia. According to the same X sources, immediate business risks include stalled longitudinal datasets, shrinking grant-matched cloud credits, and reduced clinical trial AI validation. However, there are near-term opportunities: industry consortia can underwrite shared biobanks and real-world evidence pipelines; payers and providers can sponsor outcome-linked AI validation; foundation grants can bridge method development for multimodal models; and enterprises can accelerate private-public data partnerships to secure compliant training corpora. According to the X posts, if the trend persists, vendors building foundation models for omics, pathology, and radiology will need to pivot toward commercial co-development and revenue-backed pilots with health systems.