Half of Global Enterprises Expected to Use Chinese AI in Two Years

Predictions indicate that half of all enterprises worldwide will adopt Chinese artificial intelligence (AI) models within two years.

As the AI technology dominance previously led by the United States rapidly shifts, evaluations suggest that China has built an independent ecosystem spanning semiconductors, AI models, AI agents, and robots.

Analysts note that Chinese AI, once merely regarded as an affordable alternative, has jumped into the top-tier frontier AI competition while swiftly expanding into corporate deployment, bringing global AI hegemony competition into a new phase.

According to a report titled "Key AI Trends in China for 2026" released by global market research firm Gartner on the 9th, the adoption rate of Chinese AI models among global companies is projected to surge from just 5% last year to 50% by 2027.

This implies that as the market reorganizes away from reliance on a single AI supplier toward utilizing multiple models concurrently, Chinese AI is emerging as a core choice by simultaneously boosting performance and cost competitiveness.

"Kimi K3," unveiled this year by Chinese startup Moonshot AI, clearly illustrates this change.

This open-weight model, equipped with approximately 2.8 trillion parameters, recorded performance levels in various benchmarks right behind the top-tier models of Anthropic and OpenAI.

Experts analyze that the US-China AI technology gap has narrowed from the past 6 to 9 months down to a recent 2 to 3 months.

This is why evaluations suggest that Chinese AI is no longer a budget model, but has emerged as a core axis of the frontier AI competition.

Behind the rapid rise of Chinese AI is the competitiveness of its overall ecosystem, spanning chips, AI models, services, and robots, going beyond the success of individual models.

Analysts view this as the result of consistently pursuing the construction of an independent ecosystem despite US semiconductor export controls and advanced AI technology restrictions.

The starting point is semiconductors.

US semiconductor supply restrictions have rather accelerated China's development of domestic chips.

Huawei proposed its proprietary path, the "Tau Scaling Law," bypassing import-dependent High Bandwidth Memory (HBM) bottlenecks through 3D advanced packaging technology, while DeepSeek and Zhipu established systems running natively on Huawei's Ascend AI chips starting this year.

Gartner forecasts that the proportion of domestic AI accelerators in private AI infrastructure within Chinese companies will exceed 50% by 2030.

In addition, the open-weight strategy is further accelerating the speed of diffusion.

Kimi K3, along with GLM-5, DeepSeek V4, and Alibaba's Qwen series, have made their model weights public so companies can freely download and apply them to their own services.

It is a strategy to rapidly expand corporate adoption by leveraging high utility and low barriers to entry.

The diffusion of models is leading to the deployment of AI agents.

The Chinese government has set a target to raise the penetration rate of AI agents and intelligent terminals to over 70% by 2027.

Chinese companies are also entering a stage of actually deploying AI agents across overall operations, moving beyond simple proof-of-concept stages.

The final axis of the ecosystem is robots.

China is expanding manufacturing automation by deploying AI-based humanoids and industrial robots into automobile, home appliance, and semiconductor production lines.

AI is rapidly expanding beyond digital services into the physical industry.

In Gartner's survey, 34% of Chinese companies responded that they are actively reviewing so-called "physical AI" applications.

Amidst this upheaval, South Korean companies are also deepening their concerns.

Adopting Chinese AI models driven by price competitiveness brings data security and geopolitical risks, whereas turning away from them forces companies to endure cost burdens and export control risks stemming from reliance on US AI.

Recently, as both the US and China have moved to control AI technology, the point that reliance on any specific country's technology itself constitutes a new risk is also being highlighted.

Following the US export controls on Mythos last month, China has also initiated responses to prevent the overseas outflow of its domestic AI technology.

The Chinese Ministry of Commerce is reportedly discussing measures with major AI companies such as Alibaba, ByteDance, and Zhipu to restrict the overseas transfer of core data used for AI model training and block foreign users from downloading AI model weights.

An AI industry official stated, "We must face the reality that both the US and China can restrict technology access whenever necessary," adding, "Securing an independent AI ecosystem is more important than anything else."

Just as China has built a self-reliant ecosystem from chips to AI models and services, voices calling for South Korea to move beyond foreign-dependent structures and secure AI sovereignty are gaining momentum.

The government is also beginning to fully review strategies to secure world-class frontier AI, going beyond developing independent AI models, triggered by the rapid catch-up of Chinese AI.

The National Assembly Research Service recently suggested in its "Issues and Points" report, "We must revise our strategy in the direction of strengthening sovereign control over the entire AI ecosystem, going beyond securing independent AI foundation models."

※ Please note: This article was translated by AI and may contain errors.