LWiAI Podcast #237 - Nemotron 3 Super, xAI reborn, Anthropic Lawsuit, Research!

2026-03-13 · Source: Last Week in AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Robotics & Autonomous Systems · Depth: Intermediate, short

Summary

The 237th episode of a weekly AI news brief, recorded on March 13, 2026, covers significant developments across AI tools, applications, business, policy, and research. Key announcements include Perplexity's "Personal Computer" local Mac agent, Anthropic's GitHub PR code review pricing and in-chat interactive visuals, and Cursor's trigger-based "Automations" for coding agents. Nvidia released its 120B-parameter Natron Free Super hybrid Transformer–Mamba latent-MoE model, trained natively at 4-bit for Blackwell GPUs, but halted H200 production for China. Business news featured xAI co-founder departures, Anthropic's Claude Marketplace preview, Yann LeCun's aMI raising $1.3B, and humanoid robot maker Sanctuary reaching a $1.15B valuation. Policy discussions centered on Anthropic suing the Pentagon over a "supply chain risk" designation. Research highlighted topics like activation steering resistance, chain-of-thought control limits, inference-scaling for cyber tasks, and multimodal pretraining.

Key takeaway

For CTOs and VPs of Engineering evaluating AI integration strategies, the rapid pace of agentic tool development and specialized model releases like Nvidia's Natron Free Super indicates a need for continuous assessment of deployment options. Your teams should prioritize solutions offering both enhanced security, such as local agents, and specialized capabilities, like code review tools, while closely monitoring policy shifts and supply chain risks that could impact hardware availability and regulatory compliance.

Key insights

The AI landscape is rapidly evolving with new agentic tools, model releases, market shifts, and critical policy and research challenges.

Principles

• Local AI agents enhance security.
• Hybrid model architectures improve performance.
• Inference scaling boosts cyber task success.

Method

Nvidia introduced Natron Free Super, a 120B-parameter hybrid Transformer–Mamba latent-MoE model, trained natively at 4-bit for Blackwell GPUs, designed for agentic reasoning.

In practice

• Utilize Perplexity's "Personal Computer" for local Mac AI.
• Explore Anthropic's code review for AI-generated code.
• Consider Cursor's "Automations" for always-on coding agents.

Topics

• AI Agents
• Large Language Models
• AI Hardware
• AI Policy
• Multimodal AI

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, AI Product Manager, Tech Journalist

Related on AIssential

• NVIDIA Releases Nemotron 3 Super: A 120B Parameter Open-Source Hybrid Mamba-Attention MoE Model Delivering 5x Higher Throughput for Agentic AI — Nemotron 3 Super is an open-source 120B parameter hybrid MoE model for efficient multi-agent reasoning.
• LWiAI Podcast #235 - Sonnet 4.6, Deep-thinking tokens, Anthropic vs Pentagon — The AI landscape is rapidly evolving across model capabilities, compute infrastructure, and critical safety/policy considerations.
• Last week in AI | 23 March — NVIDIA is integrating hardware and software into "AI factory" platforms, while new research enhances LLM multi-step task capabilities.
• Hermes Agent Foundation Update: Real-Time Agents, DeepSeek V4 FREE, Native Windows Support, & More! — Hermes Agent's 0.14.0 update unifies AI tooling, boosts performance, and expands autonomous capabilities.
• Build enterprise-ready Agentic AI with DataRobot using NVIDIA Nemotron 3 Super — Transforming powerful LLMs into trusted, production-grade enterprise AI systems requires robust infrastructure beyond model capabilities.
• NVIDIA AI Releases Nemotron-Terminal: A Systematic Data Engineering Pipeline for Scaling LLM Terminal Agents — High-quality data engineering, not just scale, drives proficiency in LLM terminal agents.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Last Week in AI.