Intel chief executive Lip-Bu Tan said cheaper Chinese artificial intelligence models are now "good enough" for most tasks and are being adopted "very fast" because of their lower token costs, speaking at a DBS panel in Singapore reported by TheStreet.
Tan named DeepSeek, Kimi and GLM as examples of Chinese open-weight systems undercutting many closed U.S. models on inference pricing, the cost users pay per token to run a model rather than the cost of training it.
He said he is "a strong believer in open source" as the best route to improving efficiency, a view that aligns with Intel's broader push into open models and lower-cost infrastructure.
The remarks mark a shift in framing from "largest model wins" to lowest cost per useful task, a dynamic that cuts both ways for Intel.
Cheaper, more efficient models could widen overall AI usage and lift demand for servers, networking and inference hardware; but if tasks need fewer or cheaper accelerators, spending could shift away from the premium chips that Nvidia dominates.
Intel is simultaneously trying to rebuild its position in AI chips and contract manufacturing against Nvidia and Taiwan Semiconductor.
Tan also told the panel that quantum computing could go mainstream within three to five years, though he stressed that software tooling will be needed to use that infrastructure efficiently.
