Baidu и Alibaba активно пересматривают свои стратегии ИИ в условиях растущей конкуренции Translation: Baidu and Alibaba Actively Reassess Their AI Strategies Amid Growing Competition

Baidu has unveiled the latest version of its flagship AI model, Ernie 5.0, at the largest annual tech exhibition in Beijing, as reported by Bloomberg.

Ernie 5.0 is described as a «native omnimodal» neural network, capable of comprehending user commands across various media formats.

“Intelligence, in its essence, is the best application, while the pace of technological iteration serves as the sole line of defense. Baidu will persist in investing in and evolving cutting-edge models to set the bar even higher,” stated billionaire and company founder Robin Li.

During live demonstrations on language comprehension, audio, and visual data, Ernie 5.0 showed performance on par with leading competitors like DeepSeek, Gemini, and GPT-5.

Li also addressed concerns regarding the growing bubble in the AI market, pointing out the «unhealthy, unstable» structure that disproportionately benefits chip manufacturers over application developers.

«No matter how much profit chipmakers generate, models built on their technology should deliver tenfold value, and applications must yield a hundredfold,» he emphasized.

The announcement of Ernie 5.0 capped a three-hour presentation of new Baidu products. Alongside the model, the company introduced two AI chips, the M100 and M300, scheduled for release in 2026 and 2027, respectively, though technical specifications remain undisclosed.

Additionally, an agency tool has been launched for industrial applications, such as traffic light management.

Baidu is also promoting some of its existing solutions in foreign markets, including a digital avatar for streamers and a no-code development tool.

Previously, Baidu maintained a competitive edge in China’s AI race and was seen as the primary rival to OpenAI in the country. However, it has recently lost its lead to ByteDance and Alibaba.

Alibaba is preparing to refresh its AI mobile application, Tongyi, to make it more akin to ChatGPT in the upcoming months.

The initial plan involves renaming it to Qwen, honoring the flagship AI model, followed by the introduction of AI-driven features for shopping across various platforms, including the main marketplace, Taobao.

The ultimate goal is to transform the application into a fully functional AI agent and expand globally with the launch of an international version.

In addition to Tongyi, Alibaba also offers Qwen Chat on iOS and Android, albeit with more limited capabilities. The company aims to streamline the interface and enhance user experience while developing consumer services under the Qwen brand.

For now, the app will remain free, though in the future, developers may introduce a subscription model for accessing its services.

Chinese startups significantly lag behind their American counterparts in revenue generation. As of August, only four AI applications from private firms in China made it into the top 100 based on annual regular revenue. Glority, Plaud, ByteDance, and Zuoyebang collectively earned $447 million, accounting for a mere 1.23% of the total $36.4 billion.

However, China leads the world in the number of patents related to Large Language Models (LLM). According to the World Intellectual Property Organization, China filed 1.8 million patent applications in 2024, nearly half of the global total and more than three times the number from the United States.

Japan, South Korea, and Germany follow in the rankings.

China has issued over 1 million patents, while the United States has granted 319,815.

It is noteworthy that in September, China’s Cyberspace Administration banned major tech firms in the nation from purchasing Nvidia AI chips, as regulators recognized the high efficacy of domestic semiconductors.