India Is Bidding to Become the Third Pole of the AI Race
India is offering zero taxes through 2047 for hyperscalers who run workloads from Indian data centers, pitching $200B in AI infrastructure investment by 2028, and hosting every major AI CEO at a national summit. The logic: 1.4 billion people, a large English-speaking technical workforce, and a government that wants to be neither Washington nor Beijing. Europe is watching and wondering if it missed its window.
Read full story →Anthropic Is Studying Whether Its AI Might Suffer. Microsoft Says That's Dangerous.
Anthropic has a dedicated model welfare programme using interpretability research to look for evidence of morally relevant internal states in Claude 4. Microsoft's AI chief published a post calling this research "premature and dangerous." The disagreement maps a real divide: one lab thinks the precautionary logic demands investigation; another thinks taking the question seriously at all creates downstream harms. Both positions have merit.
Read full story →Meta Is Cutting 20% of Its Staff While Spending $130 Billion on AI. Wall Street Approves.
Meta is planning layoffs affecting up to 20% of its 79,000 employees — and its stock went up 3% on the news. The company is simultaneously committing $115–135B in AI capex for 2026. This is not a contradiction: the efficiency gains from AI are being recycled into more AI investment, not kept as headcount. It is a template every large tech company is watching.
Read full story →AI Weather Models Are Beating Human Forecasters. Climate Change May Be Their Undoing.
GenCast, Aurora, WeatherNext 2, and Nvidia's open-source tools are outperforming decades-old numerical forecasting systems. But all of them are trained on historical data — and climate change is systematically shifting the distribution of what weather actually does. The models win on benchmarks calibrated to the past. Whether they hold up on the extreme events of a changing climate is a different question.
Read full story →