Slow AI Buildout Could Be Inflationary: Natixis' Herrero
THE SO WHAT
If AI demand outpaces data-center buildout, compute becomes a constrained input and pushes prices up instead of down. CFOs betting on AI-driven cost deflation should model a scenario where GPU and DC scarcity keeps unit costs elevated for several years.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIAsia’s Busiest Earnings Week Puts AI Boom in Focus After Selloff
The question on Asia’s earnings calls is shifting from AI capex volume to AI profit conversion and spillover into non-tech sectors. If you’re selling into this ecosystem, be ready to quantify ROI and downstream revenue, not just “AI strategy” and GPU counts.
Applied AIQuote of the day by Microsoft CEO Satya Nadella: 'We have to do something useful with AI, or we'll quickly lose social permission'
Social license is becoming a hard constraint on AI roadmaps, not just a PR talking point. If your AI workstream doesn’t have a clearly legible, non-speculative benefit to customers or citizens, expect more friction from regulators, employees, and buyers over the next 12–24 months.
Applied AIGrok releases the first seven minutes of its AI-generated ‘Odyssey’
An AI-generated ‘Odyssey’ that looks “okay” but thin on human interest is a clear line: AI can fill runtime, but not yet reliably carry narrative. Studios and brands should treat this as cheap experimentation and pre-viz — not a replacement for human-led stories where emotion and nuance drive value.
Applied AIOpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?
An 80% price drop in frontier LLM APIs in months is compressing margins for anyone reselling raw tokens and pushing value up into orchestration, data, and workflow depth. If your AI product’s moat is “we call the model for you,” assume your pricing power is decaying and move up-stack fast.