The YouTube Tip of the Google Spear (stratechery.com)

Nvidia’s deepening financial and technical entanglement with OpenAI—culminating in up to $100 billion in investment—underscores how the AI stack is consolidating around a few dominant ecosystems. Nvidia’s multi-year bets on chips, networking and software have made it the de facto infrastructure provider, and an equity stake in OpenAI both secures its largest customer and converts hardware revenue into long-term upside. That deal is emblematic of many cross-company arrangements (OpenAI-Oracle, Nvidia-Intel, etc.) that reflect both the capital intensity of generative AI and the strategic desire to own more of the value chain. At the product level Google is pushing AI where it can win: YouTube. DeepMind’s Veo 3 model, in a mobile-optimized “Veo 3 Fast” variant, will power new YouTube Shorts features that generate 480p video with sound from phones—animate photos or one video with another, stylize clips, add objects, assemble footage with voiceover, and even convert speech to song. Technically this is a latency- and scale-focused deployment of generative video+audio on-device/cloud hybrid pipelines. The implication: massive growth in churned AI-generated content paired with YouTube’s recommendation engine could raise the absolute quantity of engaging videos while lowering marginal costs (and changing creator economics). Risks—quality, copyright and deepfakes—remain, but the move highlights a bigger strategic bet: AI may shift long-term platform value from text/search to video-driven entertainment.