TL;DR:
- Zhipu AI, a Chinese foundational model developer, secures $340 million in funding in 2023.
- The company was founded in 2019 and is affiliated with Tsinghua University.
- Zhipu’s success comes amidst increased US-China competition in AI, with restrictions on Nvidia AI chip exports.
- China’s AI firms are stockpiling semiconductors due to anticipation of further bans from the US.
- Zhipu diversifies funding sources by attracting yuan-denominated funds, departing from the USD trend.
- President Biden’s executive order restricts U.S. investments in critical Chinese tech sectors.
- Prominent venture capitalists in China invest in Zhipu amid a shifting investment landscape.
- Zhipu receives support from Chinese tech giants, including Alibaba and Tencent.
- The company open-sources its bilingual conversational AI model and a massive 130 billion parameter foundational model.
Main AI News:
In the realm of artificial intelligence, a dynamic competition is brewing between the United States and China, each fostering its own formidable contenders. While the US boasts the likes of OpenAI and Anthropic, China is witnessing the emergence of its own AI champions. Zhipu AI, a foundational model developer founded in 2019 and spun out of China’s esteemed Tsinghua University, has just announced a significant milestone. This pioneering entity, led by Professor Tang Jie from the university’s Department of Computer Science and Technology, has successfully secured a staggering 2.5 billion yuan ($340 million) in total financing for the current year.
However, this triumph comes against a backdrop of evolving geopolitical dynamics. The Biden administration, in an attempt to curtail China’s rapid technological ascent, has recently imposed additional restrictions on the export of Nvidia AI chips to China. These constraints further complicate China’s pursuit of training large language models, intensifying the rivalry between the two AI powerhouses. In response, China’s deep-pocketed AI firms have been fervently amassing semiconductor reserves, investing hundreds of millions of dollars in these coveted chips.
To maintain its competitive edge in this high-stakes AI race, Zhipu is proactively securing its financial future by tapping into local investors. Notably, the $340 million investment infusion was sourced from yuan-denominated funds, marking a notable departure from a two-decade tradition where USD funds were the preferred financing avenue. Geopolitical tensions have given rise to this seismic shift, underlining the importance of diversifying funding sources.
In a significant policy move, President Joe Biden signed an executive order in August, prohibiting U.S. investments in critical Chinese tech sectors, including AI, semiconductors, and quantum computing. The primary objective is to curb China’s military expansion, but it has also sent shockwaves through China-focused American venture capitalists. Many are now shying away from investments in sensitive sectors, opting instead to restructure their China-focused entities. For instance, Sequoia Capital China, now HongShan, and GGV Capital have undergone such transformations.
HongShan, in collaboration with other prominent venture capital firms like Shunwei Capital and Hillhouse Capital, alongside a state fund managed by Legend Capital, has made a strategic investment in Zhipu. This collective effort underscores the recognition of Zhipu’s pivotal role in China’s AI ecosystem.
Remarkably, Zhipu has garnered support from a diverse array of Chinese internet giants, bringing together archrivals such as Alibaba and Tencent in a rare display of cooperation. The roster of investors includes Ant Group, Alibaba, Tencent, Xiaomi, Meituan, Kingsoft, TAL Education Group, and Boss Zhipin.
Not content with merely securing funding, Zhipu has also contributed to the broader AI community by open-sourcing its bilingual conversational AI model, ChatGLM-6B. This model, trained on a monumental six billion parameters, claims the capability to perform inferences using a single consumer-grade graphics card. Additionally, Zhipu has made available its foundational model, the GLM-130B, trained on a staggering 130 billion parameters. These open-source initiatives mark Zhipu’s commitment to advancing AI innovation and collaboration on a global scale.
Conclusion:
Zhipu AI’s substantial funding, support from major Chinese players, and open-source contributions underscore China’s determination to compete in the global AI market. Geopolitical factors are reshaping the funding landscape, with a focus on local currency investments, and the race between the US and China in AI continues to intensify.
