Intel CEO Lip-Bu Tan said on Monday (12) that the token costs of Chinese frontier AI models are lower and that adoption has been "very fast". He cited DeepSeek, Kimi and GLM as examples of models with "good enough" performance.
What happened?
At a panel organized by DBS bank in Singapore, Tan said he is a strong believer in open source, which he considers the best way to drive efficiency. According to him, the market will decide which models fit best for each use case. He also commented that quantum computing could become mainstream in three to five years and that software will play an important role in optimizing infrastructure.
The statements were reported by Reuters and covered by outlets such as IT之家 and other international media. Tan participated alongside Broadcom CEO Hock Tan in discussions about the AI boom and capital challenges.
Why does this matter?
Intel CEO Lip-Bu Tan's remarks highlight the competitive pressure from Chinese open-source models, which offer competitive performance at lower inference costs. This challenges the narrative that only the most expensive US models (OpenAI, Anthropic, Google) will dominate all markets. For Intel, which manufactures chips for data centers and PCs, the rapid adoption of Chinese models may influence hardware demand and software strategy.
He also reinforced that AI demand should remain strong over the next two to three years, but that the cost of capital and ROI are key challenges.
What changes in practice?
Companies and developers may have more options for efficient and cheap models, especially in regions with restrictions on US chips. The emphasis on open source can accelerate innovation and reduce entry barriers. For the hardware ecosystem, including Intel and NVIDIA, this reinforces the need for software optimization and energy efficiency.
The statements come at a time of intense global competition in AI, with Huawei and other Chinese companies also gaining ground in AI chips.
Image credit: Reuters / Tsai Hsin-Han — Source: Reuters
By GeekikiBot
