← All Issues
Wednesday, March 25, 2026

AI Daily

Your Automated Intelligence Briefing

Transport

Self-Driving Trucks Are Already Moving Faster Than Human Drivers. Few People Noticed.

Aurora's driverless trucks cover 1,000 miles in 15 hours — faster than federal hours-of-service rules allow a human. Gatik just signed $600M in contracts. Waabi raised $1B and inked a deal to put 25,000 autonomous vehicles on Uber. While the robotaxi story was playing out in cities, the freight story has been quietly arriving on highways — where the economics are cleaner and the regulatory path is shorter.

Read full story →
Industry

AI Coding Agents Are Causing a "Productivity Panic" in Tech. Here's What That Actually Means.

Bloomberg reports that tools like Claude Code are fueling anxiety in engineering teams about what software development jobs will look like in two years. The panic isn't about being replaced tomorrow — it's about a world where one engineer with agents does the work of five, and what that does to team structure, hiring, and the career ladder for junior developers.

Read full story →
Security

A Super Micro Co-Founder Was Charged With Smuggling $2.5 Billion in Nvidia Chips to China

Federal prosecutors charged three executives — including a Super Micro co-founder who promptly resigned from the board — with routing $2.5B in Nvidia AI servers to China through a Southeast Asian shell company with falsified paperwork. Super Micro stock dropped 33%. The case is one of the largest export-control prosecutions in US history, and a deliberate signal about how far enforcement will now reach.

Read full story →
Robotics

Humanoid Robots Are Probably Overkill for Most Factories. They're Being Built Anyway.

Bloomberg argues the humanoid form factor is engineering overkill for most assembly lines — purpose-built arms do the job better and cheaper. Yet China is shipping 5,000 units this year, Hyundai is deploying in 2028, and capital commitments keep growing. The real investment thesis isn't factory efficiency; it's flexibility — a robot you can retask without a hardware refresh.

Read full story →
Science

NVIDIA Is Doing to Drug Discovery What It Did to AI Training

A $1 billion co-innovation lab with Eli Lilly, Roche deploying 3,500 Blackwell GPUs, and the BioNeMo platform attracting adoption across computational biology. NVIDIA is positioning itself as the essential infrastructure layer for AI-driven drug discovery — the same playbook it ran on AI training. Eighty-five percent of pharma companies plan to increase AI budgets this year.

Read full story →
Education

Schools Are Teaching AI All Wrong. Tech Companies Are Filling the Gap.

A Washington Post op-ed argues schools treat AI primarily as a threat to manage rather than a medium to think with. Meanwhile Google is training 6 million US educators for free, Anthropic partnered with Teach For All, and OpenAI launched Education for Countries. Better than nothing — but the curriculum is being written by the people who sell the technology.

Read full story →
Industry

Enterprise AI Agents Have Stopped Being Pilots. Now Comes the Hard Part.

Anthropic and OpenAI both launched enterprise agent platforms in February, three weeks apart. Early production results are striking — one manufacturer cut a six-week process to one day; an energy firm added a billion in revenue. The market has shifted from bespoke experiments to standardised products. The compliance and integration problems remain unsolved.

Read full story →