Chinese AI Models Lead on Token Cost as Intel CEO Endorses DeepSeek, Kimi and GLM

The Quick Take

• Intel CEO Lip-Bu Tan praised Chinese AI models DeepSeek, Kimi and GLM as 'good enough' with lower token costs
• Chinese talent now occupies four of five layers of NVIDIA's full AI stack, per analysis of industry leadership
• AI enterprise software in China is shifting HR budgets as firms evaluate AI agents before hiring decisions

Intel CEO Lip-Bu Tan's endorsement of Chinese frontier AI models — citing DeepSeek, Kimi, and GLM as delivering 'good enough' performance at lower token costs than Western counterparts — represents a significant validation of China's competitive position in the global AI model race, delivered at a DBS Bank forum in Singapore. Tan's additional observation that Chinese AI models are proliferating 'very fast' reinforces a growing consensus that China's AI ecosystem is closing the gap with US models on performance metrics while maintaining a meaningful cost advantage. This dual dynamic of competitive capability and price efficiency is reshaping enterprise AI adoption globally.

The cost competitiveness of Chinese AI models creates direct pricing pressure on US-headquartered AI infrastructure companies including OpenAI, Anthropic, and Google DeepMind, which monetize through API token pricing. Enterprises globally — particularly cost-sensitive buyers in Asia-Pacific and emerging markets — have strong financial incentives to adopt Chinese models for workloads where performance parity exists. Within China, the AI enterprise software market is shifting workforce planning fundamentals, with firms evaluating AI agent capabilities before committing to new headcount, compressing traditional enterprise software hiring budgets and reshaping labor economics in knowledge work sectors.

Watch Nvidia's next earnings report for commentary on China AI chip demand, as Chinese model proliferation driven by lower-cost inference hardware directly affects Nvidia's H20 and alternative chip revenue streams in the China market. Monitor enterprise software vendors (SAP, Salesforce, Microsoft) for guidance on AI-driven workforce efficiency gains versus headcount expansion in Asia-Pacific operations. The macro variable is US export controls on AI chips: tighter restrictions accelerate Chinese domestic chip development and model optimization for efficiency, while relaxed controls would integrate Chinese and Western AI ecosystems more deeply.

Synthesized from 4 sources.