U.S. standards push vs. China’s volume offensive: the new AI-native race

60th Anniversary National Report Conference: AI Native Korea If the focus so far has been on performance, the new agenda is the spread of AI into everyday life

사진 확대 A researcher at the National Institute of Standards and Technology (NIST) conducts experiments to establish AI and robotics standards by testing collaboration safety criteria with a robot that can recognize human attention states. [NIST]

The battlefield in the race for artificial intelligence (AI) supremacy is shifting. The competition is moving beyond a simple “spec” race to boost model performance, and into an “AI-native” phase where national competitiveness depends on how deeply AI permeates daily life and industry.

Recently, the United States has been moving fast after the National Institute of Standards and Technology (NIST) under the Department of Commerce released a draft for international AI standards. NIST is rapidly revising the draft by incorporating feedback from companies and research institutes. Breaking with the old practice where such work could take years, Washington is accelerating to secure first-mover advantage in “AI standards” that can be applied on the ground right away.

At a manufacturing plant in China visited by this reporter, developers were analyzing process data using AI agents that the government has effectively distributed free of charge. Thanks to costs that are hundreds of times lower than those of global competitors, AI is no longer the exclusive domain of large corporations.

Until last year, the core of AI competition was the performance of Large Language Models (LLM). OpenAI released GPT-4.5 in February last year and, just 46 days later, followed up with GPT-4.1. China’s DeepSeek also rolled out new models in rapid succession.

But since the beginning of this year, the axis of competition has shifted as model performance has largely converged at a high level. The decisive factor is no longer who has the smartest model, but who can secure more real-world data and user loyalty to lock in the ecosystem.

At this point, the world’s two leading AI powers, the United States and China, are pursuing sharply contrasting strategies.

The United States is staking everything on dominating global standards. The AI agent standardization work led by NIST is an effort to design certification, security, and system-connection requirements around U.S. specifications. The aim is to control the “rules” of industry and thereby steer the entire ecosystem. Mark Lemley, a professor at Stanford University, said, “Before long, thousands or even tens of thousands of AIs will be connected, negotiating with each other, making decisions, and getting work done,” adding, “We absolutely need a common set of rules led by the United States.”

China has chosen a data-driven mass deployment strategy. At China’s Two Sessions earlier this month, Beijing announced a plan to elevate AI to the level of a “universal infrastructure” like tap water or electricity. The state intends to distribute AI computing power and models as a kind of public good.

With the United States pushing standards and China pushing rapid diffusion, South Korea now stands at an inflection point. Experts warn that the country’s fate will hinge on what kind of AI strategy it adopts between the two giants. Kangchan Lee, head of the Strategy and Standards Research Office at the Electronics and Telecommunications Research Institute (ETRI), noted, “Whoever secures the standards and platforms will determine industrial competitiveness.”

■ Term explanation ▶▶ AI-native: A state in which every citizen uses AI as naturally in work and daily life as they use their mother tongue.

[Las Vegas – Park Sora / Shenzhen – Kim Hee-soo]

This article has been translated by GripLabs Mingo AI.

Originally Appeared Here