Is China's AI video boom eating the compute it needs for LLM training?
enilea · reddit · 2026-11-05
A Reddit post argues that China's enjoyment of AI-generated video has created a strategic drag: reports claim roughly half of its AI compute goes to video inference, compute that could otherwise fund LLM training and ML research. The author notes the industry now depends on media inference revenue, making it hard to stop, even as China executes well on supply chain, grid expansion, and long-term infrastructure. Sora nearly created the same trap for OpenAI, and the author argues that on the road to RSI you can't afford to waste compute on slop videos, closing by asking whether AI video revenue justifies the opportunity cost.
More from AGI Musings
- LessWrong's flaw: treating 'being less wrong' as an identity, says blogger — StewartalsopIII · 2026-11-05
- Commentator: US should build independent AI bodies like the UK's, not data center restrictions — TaniaBabina · 2026-11-05
- Redditor floats data poisoning as a secret AI arms-race tactic, citing Anthropic's 250-doc backdoor study — cern0 · 2026-11-05
- New preprint analyzes 120k papers to show how LLMs reshaped scientific division of labor — JMateosGarcia · 2026-11-05
- SF 'Sexbot' Demo Sparks LinkedIn Debate: Normalizing Rape or Sex-Positive Critique Gone Wrong? — Graham_dePenros · 2026-11-05
- Vinod Khosla: Open Source Models Too Costly and Dangerous, Urges US-China Treaty to Constrain Them — rohanpaul_ai · 2026-11-05