Short reactions to what's moving in AI, infrastructure, and software.
Always linked to the source.
Even the best frontier AI models corrupted roughly a quarter of a document during a 20-step delegated workflow. Across all models tested, average degradation was 50%.
AI has given everyone the ability to build custom software in plain English. No computer science degree required.
Gartner just surveyed 350 companies with over $1 billion in annual revenue on AI and workforce reductions. The result should give every operator pause.
Anthropic's Mythos model was supposed to be so dangerous at finding security vulnerabilities that they couldn't release it to the public. That was the headline in April.
Last week I posted about the historic revenue growth at OpenAI and Anthropic. Numbers that have no precedent in the history of enterprise software.
Vibe coding is a genuine enterprise opportunity. Non-technical employees building functional internal tools in hours instead of weeks is a real force multiplier.
Yesterday I posted about who is actually winning financially from AI right now. The infrastructure layer: AWS, Microsoft, Google, Nvidia.
The cloud market just hit a $500 billion annual run rate. Growing 35%.
Ouster just announced color lidar that captures both visual and 3D depth data simultaneously from a single sensor. The CEO calls it "the holy grail of what a roboticist has always wanted.
> Only the paranoid survive. > > -Andy Grove Three attacks.
This morning I posted about indirect prompt injection. Attackers hiding invisible instructions in web pages to hijack AI agents.
McKinsey built an internal AI platform called Lilli. Three quarters of their 40,000 employees use it for strategy work, client research and document analysis.
The web is being weaponized against your AI agents. And the attack is hiding in plain sight.
AI data centers are gobbling up the world's memory supply. And your business is going to feel it.
An AI coding agent just deleted an entire company's production database and all its backups. In 9 seconds.
GPU spot prices for NVIDIA's latest chips (B200) just surged 114% in the last six weeks. Most people assume AI infrastructure costs are falling.
Software supply chain attacks have been a known threat for years. Security professionals have been worried.
The most expensive part of your AI rollout isn't the software. It's the rework no one is tracking.
Everyone is counting on AI tokens getting cheaper. That assumption depends on infrastructure that isn't being built on schedule.
McKinsey projects global data center capacity will nearly triple by 2030. AI workloads alone grow 3.
Whatever happened to under-promise and over-deliver? 241 gigawatts of data center capacity announced.
Insights covers the same topics in depth: research-backed analysis on AI, value creation, and building companies.
Read Zaruko Insights