China’s artificial intelligence (AI) sector is experiencing rapid growth, fueled by aggressive investment and a strong open-source ecosystem. While major Chinese companies like Alibaba and Baidu are leading the charge, AI startups are also making significant strides. One such startup, Monica, has developed an AI agent called Manus, already dubbed “the second DeepSeek.” As China continues to expand its AI dominance, voices are growing in South Korea calling for increased investment and a stronger open-source ecosystem to stay competitive.
According to the IT industry, Manus, released on Wednesday, March 5, is drawing attention as an autonomous AI agent with capabilities comparable to DeepSeek R1, another Chinese AI model. While large language models (LLMs) typically function based on user commands, Manuscan performs tasks independently. For example, suppose a user uploads a compressed PDF file containing multiple resumes and requests an Excel summary. In that case, Manus will automatically extract the files, analyze the resumes, and organize them into a spreadsheet. Beyond this, it can even provide insights on industry trends and relevant technologies, demonstrating advanced AI autonomy.
Industry experts view Manus as more advanced than OpenAI’s Operator feature, which is currently offered to Pro-tier subscribers (priced at $200 per month). Unlike conventional AI models focusing on natural language processing and image recognition, Manus extends its capabilities to complex fields like financial analysis. According to Monica, the startup behind Manus, the AI agent functions as a general-purpose AI that processes information and delivers actionable results. Benchmark tests evaluating AI problem-solving abilities have confirmed that Manus outperforms OpenAI’s Deep Research feature, showcasing its superior analytical capabilities. Additionally, Manus has been launched as an open-source project called OpenManus, allowing researchers and developers to participate in its real-time improvement and feedback process.
@@@
Image Generated by Alibaba’s Wan 2.1 / Yonhap News
Major Chinese tech companies, including Alibaba, ByteDance, and Baidu, actively embrace open-source AI development and have established a robust ecosystem. On March 6, Alibaba unveiled QwQ-32B, an open-source AI inference model that is only one-twentieth the size of DeepSeek R1 but claims to match its performance. Meanwhile, Alibaba Cloud recently launched Wan 2.1, the latest version of its video foundation model Tongyi Wanxiang, which has drawn significant attention.
China is making massive national investments in AI research and development (R&D) while easing regulatory restrictions. The Chinese government has established a national venture capital guidance fund worth approximately $138 billion to channel funding into AI, quantum science, and other advanced technology sectors.
Chinese corporations are also doubling down on AI investments. Alibaba has announced a three-year plan to invest approximately $52 billion in AI infrastructure, including data centers. Meanwhile, ByteDance is allocating $12 billion solely for AI development in 2025.
In contrast, South Korea is struggling with a weaker open-source foundation and insufficient AI investment, raising concerns about its global competitiveness. AI technology requires enormous data and computational power, yet South Korea lacks the necessary infrastructure and industrial investment to support these demands. The country’s insistence on closed-source development makes it difficult to keep pace with open-source advancements worldwide. Choi Byung Ho, a professor at Korea University’s AI Research Institute, emphasized the need to learn from China’s AI development strategy.
“We need to take cues from China’s AI strategy,” he stated. “To remain competitive, South Korea must make massive state-level investments in leading AI companies and foster a strong open-source community to drive AI advancements.”
