XPeng's Fuyao supercomputing hub hits 600 PFLOPS, cuts training from days to under an hour
bookwormengr · x · 2026-09-08
XPeng's full-stack AI development runs on the Fuyao supercomputing center in Ulanqab, Inner Mongolia, built with Alibaba Cloud: over 600 PFLOPS sustained (240 B200s at FP16), architected for large-scale multimodal training. Distributed infrastructure compressed core foundation-model training from multiple days to under an hour — a 170x+ iteration speedup. Having merged its automotive and robotics software divisions, XPeng feeds vehicle-collected visual data, spatial maps, and corner cases into Fuyao to train the universal VLA 2.0 model, a unified fleet data flywheel.
Related event: XPeng unveils in-house Turing chip and Fuyao supercomputing center(2 posts)→
More from Infra
- Japan plans $60 billion investment in data centers — JosephJacks_ · 2026-09-08
- Vaire Computing CTO named MIT TR35 for chips that recycle waste heat into energy — MikePFrank · 2026-09-08
- GMKtec's EVO-X5 Pro mini PC claims to run a 300B LLM fully offline — Dave_Maynor · 2026-09-08
- Actian's new vector DB claims 22x Qdrant speedup, but its docs look suspiciously like Qdrant's — qdrant_engine · 2026-09-08
- 8GB VRAM runs Flux and Wan 2.2 locally fine: three wrong settings, not the GPU, were the bottleneck — leonbuilds · 2026-09-08
- China effectively leads the humanoid robot supply chain, and Optimus relies on it — JOBhakdi · 2026-09-08