Chinese Companies Are Dominating the Patent Race for Generative AI

What We’re Showing

The top companies by patent ownership in GenAI models, as of April 2024. The data is from the World Intellectual Property Organization (WIPO), the United Nations agency for innovation and creation.

Key Takeaways 

China is dominating the patent race for generative AI, with Tencent and Baidu topping the list.

Among the GenAI programs, or models, with most patents are:

• Generative Adversarial Networks (GANs): GANs use a generator to create data and a discriminator to evaluate it, refining the generator's output. They are key in image generation, style transfer, and data augmentation
• Variational Autoencoders (VAEs): VAEs encode data into a latent space and decode it back, allowing new data generation by sampling the latent space. They are used in image generation, anomaly detection, and semi-supervised learning
• Decoder-based Large Language Models (LLMs): LLMs, like GPT, generate text by leveraging the transformer architecture and vast pre-training data. They excel in text completion, translation, summarization, and conversational AI