Technology and Business Services

AI Weekly Report: World Models, Open-Source LLMs, and AI Infrastructure Battles

14 Jun 2025

Breakthroughs in AI

Meta unveils V-JEPA 2, an open-source AI world model that enables machines to understand 3D environments and physical object behavior, advancing robotics and autonomous systems.
AlphaSense launches Deep Research agent combining LLMs (Anthropic via AWS Bedrock, Google Gemini, Cerebras on WSE-3) with proprietary and internal data via RAG, ensuring traceable insights.
Mistral releases Magistral reasoning LLMs, open-source small and enterprise-grade medium that offers transparent, chain-of-thought inference and efficient multilingual capabilities.
AMD launches the Instinct MI350 series, advancing its competitive footing in AI infrastructure with new chips optimized for large model inferencing and a push toward open, interoperable AI ecosystems.
Nvidia unveils "Climate in a Bottle", a global-scale digital twin using generative AI and Earth-2 to simulate climate patterns and help governments pre-emptively prepare for natural disasters.
Google DeepMind launches AI hurricane model predicting storm track and intensity with unprecedented accuracy and speed, and is now partnering with U.S. agencies.

Enterprise adoption and ecosystem shift

AstraZeneca deepens collaboration with China's Monster AI, leveraging advanced multimodal models to accelerate drug discovery and precision medicine breakthroughs.
Foxconn taps Google Cloud’s AI/ML for predictive maintenance, quality checks, and supply chain resilience, accelerating Electronics 4.0 and real-time intelligence transformation.
Firebird and Government of Armenia partner with NVIDIA to build a $500M AI supercomputing hub by 2026, catalyzing regional innovation and digital transformation.
L’Oréal integrates NVIDIA AI Enterprise for generative content and hyper-personalized product recommendations, elevating beauty R&D, marketing, and consumer engagement at scale.
Indonesia launches landmark AI ethics partnerships with OECD, ASEAN, and industry bodies to co-develop regulations, upskill talent, and lead trustworthy AI standards regionally.