Nvidia faces $50 billion risk in China as AI chip longevity remains high

Nvidia CEO Jensen Huang has warned that the combination of AI models and Huawei chips could significantly impact Nvidia’s market position in China. He noted that if large language models like DeepSeek V4 continue to utilize Huawei’s hardware instead of American technology, Nvidia could face losses of approximately $50 billion. Such a shift would threaten the network effect of Nvidia’s CUDA platform as a global AI standard.

In a separate development regarding hardware longevity, AI cloud provider CoreWeave confirmed that its Nvidia A100 GPUs remain in high demand. The company has signed contracts to lease A100 chips through 2029, suggesting that the economic utility of AI hardware may last much longer than the two to three years some critics predicted. This longevity allows older chips to be repurposed for less intensive computational tasks after initial model training is complete.