OK, I give up. @GoogleDeepMind Neon feels different. A real step change.
Either Google finally figured out how to put its enormous trove of training data to work, or we’re seeing the beginnings of recursive self-improvement. Maybe both.
The pace is getting ridiculous. Each stronger model helps build the next one, and the window to catch up keeps shrinking.
If this keeps up, Anthropic’s IPO timing might be perfect. Because how long can “we have the best model” remain a viable business model when your lead lasts weeks and Google can fund the race indefinitely?
Race @GoogleMaps Pegman through Tokyo's Shibuya Scramble Crossing to reach the Google building 🏃
Built with Gemini 4 Neon, this 3D runner turns nearly 1M tokens of real OpenStreetMap data (654 roads, 70 rail lines, and 280 building footprints) into a playable arcade map with custom Web Audio.
The more I learn about cybersecurity breaches, the more convinced I am that “AI transformation” is a myth.
Companies won’t survive by bolting AI onto broken legacy systems. They need to rebuild from scratch, become AI-native, and attract people obsessed with AI.
Very few companies or countries grasp the magnitude of what’s coming, let alone the urgency to act.