China's AI Firms Challenge US Giants Amid Chip Curbs

Chinese AI startups and tech giants are rapidly advancing their large language models, challenging US dominance despite export restrictions on advanced microchips.

NGN Market

Written by NGN Market

·3 min read
China's AI Firms Challenge US Giants Amid Chip Curbs

China's Artificial Intelligence boom is accelerating, with top startup DeepSeek releasing a new large language model (LLM) on Friday. This development underscores the nation's swift progress despite US export restrictions on advanced microchips, positioning Chinese firms as significant challengers to US and European counterparts.

Internet giants Baidu, Alibaba, and Tencent are aggressively investing in AI, capitalizing on their extensive user bases and cloud infrastructure. Baidu, often referred to as China's Google, has been a long-time advocate for AI's potential, developing its "Ernie" AI chatbot. E-commerce leader Alibaba, known for its open-source approach, saw its Qwen chatbot mobile app reach over 200 million monthly active users in January, according to AICPB.

Tencent, a major player in gaming and social media, launched an AI model in 2023 and a chatbot the following year. Founder Pony Ma recently emphasized AI as "the only field worth investing in," signaling increased investment.

Advertisement

ByteDance, the company behind TikTok, is increasingly focusing on AI as pressure mounts on its overseas social media operations. Its AI chatbot, Doubao, is the most popular in China with over 100 million daily active users. The firm's AI video generator, SeeDance 2.0, has raised concerns about copyright and potential job displacement with its high-quality, prompt-generated clips.

Startup DeepSeek, which began as a side project in 2023, made waves in the global AI scene with its "R1" model in January 2025. Its low-cost, high-performance R1 chatbot was seen as a "Sputnik moment" for AI, challenging the perception of US dominance. DeepSeek's open-source models are noted for their competitive performance.

The company's newest model, DeepSeek-V4, released on Friday, aims to match the performance of leading closed-source models. DeepSeek-V4 features an ultra-long context of one million tokens and, for its Pro version, 1.6 trillion parameters, which dictate the model's input capacity and decision-making capabilities. DeepSeek-V4-Pro reportedly leads in world knowledge benchmarks among open-source models.

Emerging startups Zhipu AI, MiniMax, and Moonshot AI are dubbed China's "AI tigers" for their work in AI foundation model research, directly competing with established tech giants. Zhipu AI, originating from Tsinghua University, is a key provider of chatbot tools to Chinese businesses, with its "GLM-5" model receiving praise from developers.

MiniMax focuses on consumer-facing multimedia tools, including AI companions and video generators. Both Zhipu AI and MiniMax experienced significant stock price increases upon their Hong Kong IPOs in January. However, Zhipu AI was placed on a US export control blacklist due to national security concerns, while MiniMax faces copyright infringement lawsuits from US entertainment companies.

Moonshot AI, whose Chinese name Yue Zhi Anmian references Pink Floyd, has seen its "Kimi K2.5" model become highly popular on the developer platform OpenRouter. The company reportedly achieved its 2025 full-year revenue targets within weeks of Kimi K2.5's launch.

Advertisement

Advertisement