Canva Gets 70% More Video Generations per GPU Hour on NVIDIA Blackwell
pbaylies · x · 2026-09-22
- NVIDIA announced that Canva is using Blackwell GPUs to achieve 70% more image-to-video generations per GPU hour at the same compute, supporting its 265 million users while preserving speed and quality.
- NVIDIA also released an "AI Tokenomics" white paper framing inference economics around four pillars: token utility, demand forecasting (a three-level model), supply optimization, and monetization/pricing, with case studies of Cohere, Perplexity, and Canva plus a token-demand forecasting worksheet.
More from Infra
- Local H3 video gen: 15s clip in ~5 min at zero cost vs ~$0.60 per generation on platforms — AIandDesign · 2026-09-22
- MotherDuck makes text classification 50x faster at ~1% of LLM cost — josh_wills · 2026-09-22
- Google serves 3.2 quadrillion AI tokens a month — ~12GW, and energy could run out in 3 years — victor_explore · 2026-09-22
- M5 Ultra 256GB early test: Mimo2.6-Flash hits 49 tok/s at 32K context — bakawolf123 · 2026-09-22
- TRL async GRPO adds LoRA sync via storage bucket and proxy, cutting 500-step training to 53 min — SergioPaniego · 2026-09-22
- GGUF models can now run directly in Hugging Face transformers with ggml Metal kernels — pcuenq · 2026-09-22