Intel CEO: Chinese AI models are winning the race on cost
Intel CEO Lip-Bu Tan says Chinese AI models like DeepSeek and Kimi are seeing rapid adoption because they are cheap and good enough.
Business · Source: Seeking Alpha
What happened
Intel chief executive Lip-Bu Tan spoke at a panel organized by DBS in Singapore on October 12. He highlighted a massive shift in the artificial intelligence landscape. Chinese frontier models are gaining serious traction. The primary driver is not raw intelligence. It is economics. The token costs for these Chinese models are significantly lower than Western alternatives. This price advantage is driving very fast adoption across the market.
Tan specifically pointed to models like DeepSeek, Kimi, and GLM. He stated their performance is good enough for users and businesses to deploy them across a wide range of applications. Token pricing dictates the financial viability of running artificial intelligence at scale. When businesses process massive volumes of requests, every fraction of a cent matters. Lower costs put these Chinese models directly in the spotlight for enterprise adoption.
Beyond Chinese models, Tan expressed strong support for open-source artificial intelligence. He argued that open-source development is the absolute best way to drive efficiency in the industry. He believes the free market will ultimately determine which models suit different use cases best. Tan also shared his outlook on quantum computing. He predicts the technology will become mainstream within the next three to five years. Software will play a critical role in making that quantum infrastructure work efficiently.
Key facts
- DeepSeek, Kimi and GLM — Chinese frontier AI models cited by Intel CEO for rapid adoption
- Three to five years — Timeline for quantum computing to become mainstream
- October 12 — Date of the DBS panel in Singapore where Tan spoke
Why it matters
The era of defaulting to the most expensive model is over. Builders are realizing that they do not need maximum intelligence for every single feature. If a system is good enough and costs a fraction of the price, it wins the deployment race. This shifts the entire industry focus from raw technical benchmarks to practical unit economics. Startups can now build high-volume applications that were previously impossible to fund. Lower operating costs make advanced artificial intelligence accessible to companies that previously sat on the sidelines.
There is a massive second-order effect happening here. Competition in artificial intelligence is no longer just about technical capabilities. Cost, accessibility, and real-world performance are the new battlegrounds. When token costs drop, companies integrate artificial intelligence into far more of their daily operations. This explosion in usage requires massive underlying infrastructure. More adoption means more servers, more hardware, and more chips. The companies selling the compute stand to benefit the most from this race to the bottom in token pricing.
For builders
Optimize for token costs over raw power
Stop paying premium prices for basic routing and data extraction tasks. Use cheaper models like DeepSeek or GLM where performance is good enough. Startups win by keeping margins high, while expensive model providers lose market share.
Prepare for open-source AI dominance
Intel is backing open-source models to drive industry efficiency. Build your software stack to be model-agnostic so you can swap in cheaper open-source options as they improve. Founders who lock into a single expensive provider risk massive financial losses.
Keep an eye on quantum timelines
Quantum computing could reach mainstream adoption within three to five years. Startups building infrastructure software should start planning for this massive hardware shift today. Companies that fail to adapt their software for quantum hardware will be left behind.
My take
I do not think Lip-Bu Tan is just admiring the competition. He knows cheaper models mean more companies use artificial intelligence at scale. More usage means Intel sells more chips to power that massive infrastructure.
Original reporting: Seeking Alpha. This is my rewrite and opinion.