DeepSeek is reshaping AI industry

The rapid rise of DeepSeek, a Chinese artificial intelligence company, is reshaping the AI industry and prompting market players to explore new possibilities, experts said at the 2025 Global Developer Conference, which concluded Sunday.

DeepSeek specializes in large language models, particularly in coding-related AI, and aims to provide high-performance models with efficient training. The company, based in Hangzhou, Zhejiang province, gained global attention last month after launching its AI reasoning models. The models are fully open source, cheaper to train and perform on par with leading global counterparts.

"I hope not just Chinese AI companies, but other global AI companies learn from what DeepSeek is doing. Starting by open-sourcing your software …they've made it really financially approachable to build your own language models," said Andrew Aitken, a technical oversight committee member of the Linux Foundation's FINOS Foundation. "They've done that really well, and the rest of the world can learn from that."

Industry experts at the conference expressed admiration for DeepSeek-R1, one of the company's latest AI developments.

"DeepSeek's open-source adoption has set a role model for how AI can benefit everybody," said Gu Ruiquan, product director at Lanyun Technology. "It helps create an ecosystem to build better AI products because everybody can stand on the shoulders of giants."

Gu added that as costs decline, demand for computing power is rising across the industry, creating new opportunities for AI development.

Experts said DeepSeek's opensource approach is driving the growth of similar models, accelerating AI applications across various industries.

"I am extremely excited about the outlook of large models, which can now be developed in a cheaper, faster, better and more efficient way," said Jia Anya, product director at SenseTime.

"The recent breakthroughs in AI indicate that integration can happen more quickly, reasoning can be accelerated with large models, AI capabilities can be further optimized and better AI applications will be developed," Jia said.

Shen Haozhan, an algorithm expert at OM AI Lab, said DeepSeek's success highlights how AI models can be developed at significantly lower costs and provides a roadmap for other companies to train their own models.

"I am personally inspired by DeepSeek's open-source strategy. It gives us new research directions beyond conventional methodologies, which is a huge contribution to the open-source community," Shen said.

The 2025 Global Developer Conference, held from Friday through Sunday in Shanghai, was themed "AI Shaping the World, Unlocking Infinite Opportunities."

The event brought together global developers to collaborate, innovate and explore AI applications while promoting commercialization efforts.

"Artificial intelligence will continue to be one of Shanghai's leading industries," said Chen Jie, vice-mayor of Shanghai, during the conference's opening ceremony on Saturday.

Chen noted that under the guidance of the Ministry of Industry and Information Technology, Shanghai has made significant strides in technological innovation, industry-finance cooperation and international partnerships. The city's AI market reached a scale of more than 450 billion yuan ($62 billion) last year, and it has hosted the World Artificial Intelligence Conference for seven consecutive years and filed 60 large models.

"I'm really impressed with what Shanghai has done around AI. Large corporations are investing in AI, and the government is providing incentives, support and investment to grow the developer community and focus on open source. Shanghai is really positioning itself as a leader in this space," Aitken said.

As open-source AI models continue to evolve, Shanghai will leverage its strengths as a megacity and accelerate efforts to establish itself as an internationally influential AI hub, Chen added.

Li Junfeng contributed to this story.

wang_ying@chinadaily.com.cn