Global AI Race Heats Up as China Narrows the Quality Gap

The United States continues to dominate the global landscape of artificial intelligence (AI) model development, with 40 notable models expected in 2024 alone. According to the most recent Artificial Intelligence Index Report, China is rapidly closing the performance gap, indicating a transformative shift in the global AI race.

China adds quality, while the United States builds numbers

While the United States has been the dominant player in developing top-tier AI models, China has made significant progress in quality. In 2023, Chinese and American models had a double-digit performance gap on industry-standard benchmarks like Massive Multitask Language Understanding (MMLU) and HumanEval (coding performance). By 2024, the gap had nearly closed.

China has advanced its AI capabilities with the introduction of Alibaba's Qwen Series, DeepSeek's R1, ManusAI, and Tencent's Hunyuan Turbo S, among others. This progress is due to China's aggressive investments in AI infrastructure, advanced computing capabilities, and state-sponsored research initiatives.

The low-cost model, developed in two months with a budget of less than $6 million, stands in stark contrast to the $100 million that OpenAI reportedly spent on training its GPT-4 model.

Top AI organizations include OpenAI, Google, and Alibaba

The global AI race to build agentic capabilities and infrastructure has piqued the interest of the world's biggest tech companies and academic institutions.

In 2024, OpenAI emerged as the leading organisational contributor, releasing seven notable AI models and establishing itself as a key player in general purpose AI systems.

Google quickly followed with six significant model launches, reinforcing its long-standing leadership in machine learning (ML) innovation. Over the last decade, Google has maintained a dominant position, contributing a staggering 186 notable models since 2014—more than doubling the next player.

Meta and Microsoft have also been prolific, developing 82 and 39 models during the same time period, respectively.

On Saturday, Meta unveiled its latest suite of open-weight AI models under the Llama 4 family, including two new variants — Llama 4 Maverick and Llama 4 Scout — aimed at delivering personalized, multimodal systems.

Notably, Alibaba, which represents China's growing presence in foundational AI development, finished third in 2024 with four notable models. This represents a shift in the global innovation landscape, with Chinese firms not only scaling deployment but also contributing to cutting-edge research and model design.

Since 2014, the most prolific academic institutions have been Carnegie Mellon University, Stanford University, and Tsinghua University, with 25, 25, and 22 notable models, respectively.

Amplifying AI research and patents

In addition to model quality, China is the world leader in AI research volume. In 2023, Chinese researchers accounted for 23.2% of all AI-related publications, compared to 15.2% in Europe and only 9.2% in India. Since 2016, China's share has steadily increased, while European contributions have declined and US publication output has plateaued.

Despite America's ban on AI chip supply, China has emerged as the world's second largest producer of AI models for text, images, video, and audio. China produced 36% of the world's 1,328 AI large language models (LLMs), second only to the United States.

While China dominates in volume, the United States retains an advantage in influence. Over the last three years, American institutions have contributed the vast majority of the top 100 most cited AI papers.