50% higher costs: Why Chinese AI giants struggle to ditch Nvidia
pstAsiatech · x · 2026-08-10
Despite pressure for domestic substitution, China's most advanced AI models still rely heavily on Nvidia chips for training. SCMP reports that prohibitively high transition costs are hampering the widespread adoption of local semiconductors.
The article notes that the core hurdle lies in the software ecosystem. While Nvidia's CUDA platform remains the industry standard, Huawei's alternative for Ascend chips, CANN, requires developers to rewrite and optimize massive amounts of code. One researcher estimates that migrating existing workflows to Ascend chips could increase team time and costs by at least 50%.
More from Infra
- Intel Announces $15B Stock Offering to Capitalize on AI Compute Demand — ryanshrout · 2026-08-10
- Hybrid Cloud + Local Model Architecture Shows Promise; Muse Spark 1.2 Open-Source Release Imminent — jack_w_rae · 2026-08-10
- 30B Model Hits 114 tps with Tuned Quants, Targeting <16G VRAM — tokenbender · 2026-08-10
- Beyond Compute: AI Data Centers Face the Power Scarcity Bottleneck — ingliguori · 2026-08-10
- Discovered Materials Raises $9M to Hunt for Novel Chip Cooling Materials — TechCrunch AI · 2026-08-10
- 1M Token Context on Single RTX 3090 Achieved via KVarN Quantization — Anbeeld · 2026-08-10