Chinese AI Firms Expand Global Reach Through Local-Language and Home-Grown Models

Dragon Media News Desk

Inside the smart exhibition hall of Safaricom, Kenya’s largest telecommunications operator, a digital human supported by a Chinese technology company demonstrates the operator’s core services and interprets financial-report data for visitors through intelligent interactive functions.

Across emerging markets with significant digital-economy potential, Chinese artificial intelligence companies are increasingly working with local partners to develop home-grown large language models tailored to national languages, industries and development needs.

Leading global large language models have traditionally focused on widely spoken languages, leaving communities that use less-common languages comparatively underserved. Experts have warned that countries with limited linguistic representation in major AI systems risk being marginalized as artificial intelligence becomes increasingly integrated into economic and social activity. Chinese AI companies are seeking to address this technological gap as part of their international expansion.

Chinese technology company iFLYTEK, which specializes in speech intelligence and natural-language processing, has intensified research and development in multilingual technologies. Its Spark speech-based large language model supports speech recognition and simultaneous interpretation in more than 130 languages.

With the ASEAN market in focus, iFLYTEK has developed the Spark ASEAN Multilingual Large Model Base in cooperation with south China’s Guangxi Zhuang Autonomous Region. The initiative draws on Guangxi’s pool of language professionals specializing in ASEAN countries, extensive linguistic data resources and digital infrastructure.

The platform currently covers 10 languages, including Malay, Indonesian, Vietnamese and Thai. According to the company, it can deliver performance comparable to leading international models while using smaller parameter sizes, enabling more efficient and secure deployment of industry-specific AI applications across ASEAN member states.

iFLYTEK has signed cooperation agreements with partners in countries including Laos, Malaysia and Thailand. In Malaysia, the company has jointly established an AI multilingual intelligent dubbing and translation center with Enjoy TV & Film Broadcasting Corporation. Powered by its multilingual model, the center supports translation in more than 130 languages and is intended to help develop an Asia-oriented intelligent dubbing ecosystem.

Dong Bin, vice president of iFLYTEK’s brand marketing center, said mainstream English-language large models still face clear limitations in many overseas markets, where demand for models adapted to local languages is substantial.

The company has also made its large model available through open-source channels. According to iFLYTEK, nearly 600,000 overseas developer teams are using its algorithms and speech tools to build customized products. The company is also expanding tested hardware-and-software AI solutions for education, healthcare and office applications into overseas markets.

Other Chinese technology companies are pursuing different approaches to providing affordable AI options for developing economies.

DeepSeek, for example, has attracted attention for offering an open-source model that can reduce entry costs for African developers. Its pricing has been reported at approximately 0.27 U.S. dollars per million input tokens and 1.10 U.S. dollars per million output tokens, substantially lowering the financial threshold for developers compared with many commercial alternatives. Its open-source structure further enables local teams to modify and deploy AI systems according to their own requirements.

In Cairo, Huawei has launched a 100-billion-parameter Arabic large language model trained on locally sourced Arabic datasets from sectors including finance, power, and oil and gas. The model has achieved a reported speech-recognition accuracy of 96 percent and is designed to serve more than 20 Arabic-speaking countries and regions.

By drawing on locally sourced data reflecting regional linguistic and cultural contexts, the system is intended to help participating economies strengthen domestic AI capabilities rather than relying entirely on externally developed models.

Beijing-based 01.AI has also partnered with Kazakhstan to establish Q.AI, which is developing and deploying large language models, AI agents and enterprise AI platforms tailored to local requirements.

Kai-Fu Lee, chief executive officer of 01.AI, said the company places emphasis on respecting local data sovereignty, on-site deployment and local operation under an open, collaborative and sustainable framework.

He argued that emerging economies such as Kazakhstan should not depend solely on purchasing foreign models or outsourcing their AI capabilities. Instead, they need self-sustaining national AI capacity built around their own data, languages, industries and governance frameworks.

A common feature of the overseas strategies pursued by several Chinese AI companies is an emphasis on strengthening local capabilities rather than simply replacing domestic systems with imported technology. This approach is increasingly linked to broader efforts to narrow the digital divide and promote more inclusive participation in the global AI economy.

At the 2026 World Artificial Intelligence Conference held in Shanghai in July, China advocated an open and mutually beneficial approach to AI development and called for stronger support for AI capacity building in Global South countries.

China has also pledged to provide 5,000 training and seminar opportunities in artificial intelligence for developing countries over the next five years.

Xue Lan, dean of Schwarzman College at Tsinghua University in Beijing, said China encourages open-source AI development and that Chinese open-source large models are among the most widely downloaded globally, supporting a growing range of applications.

He said the broader objective goes beyond making technology available. Building local expertise, he argued, can enable countries to develop AI applications suited to their own linguistic, economic and social conditions.

The international expansion of Chinese AI companies therefore increasingly represents more than the export of technology. It reflects a broader effort to support AI ecosystems built around local languages, locally governed data, national industries and domestic technological capacity.