What the World Is Missing About China’s AI Sprint

Zhu Dan

Beijing-based journalist and geopolitical analyst with two decades of experience in international journalism. She specializes in analyzing shifts in global politics and Chinese diplomacy. She regularly covers major multilateral issues for CGTN’s French-language service. The opinions expressed in this article are those of the author alone.

On February 26, German Chancellor Friedrich Merz stood inside the headquarters of Unitree Robotics in Hangzhou and watched humanoid robots perform « Wu BOT », a kung fu routine. Ten days earlier, machines built by the same company had performed the same routine on live television before over a billion Chinese viewers during the Spring Festival Gala. The fact that an European leader chose a Chinese robotics startup on his first official visit to the country says something about where the global AI competition now stands.

Unitree’s moment in the spotlight was part of a broader wave. In January, reports emerged that DeepSeek was preparing to release its V4 model. A year earlier, the startup’s R1 had rattled the Nasdaq, wiping nearly $600 billion off Nvidia’s market value in a single day. In February, ByteDance rolled out Seedance 2.0, an AI video tool that unsettled Hollywood. And at the Spring Festival Gala on February 16, four robotics startups showcased humanoid robots performing martial arts with a fluidity that sparked worldwide debate.

Lian Yingying, a team member at Unitree Robotics, described what that fluidity cost: « To synchronize over twenty robots with the music, the martial artists, and the shifting stage terrain, we fine-tuned every movement to an almost obsessive degree. Sometimes we polished a single 0.1-second transition over and over again. »

The speed of progress is striking. In 2025, Unitree was the only company to stage a group robot performance at the Gala. Twelve months later, four companies competed on the same stage, each with greater range, smoother movement, and more sophisticated coordination. In China’s robotics industry, a year is a long time.

What happened during the broadcast may matter even more. Within two hours of the Gala’s start, JD.com reported that searches for robot products surged over 300% month-on-month, with orders jumping more than 150%. By 10 p.m., several robot models from JD merchants sold out within minutes of launch, including two high-end units priced at nearly 630,000 RMB combined. Demand wasn’t confined to Beijing or Shanghai. It reached smaller cities and rural markets, revealing a nationwide appetite for AI products that few analysts had predicted. According to Counterpoint Research, China held over 80% of installed capacity share in the global humanoid robot market in 2025.

This quick shift from TV spectacle to consumer purchase reflects a level of digital integration that few countries have matched. Mobile payments, e-commerce platforms, and logistics networks are wired into a single pipeline. A product seen on TV at 8 p.m. can be bought, paid for, and shipped within minutes to virtually any address in the country.

Meanwhile, a quieter but arguably bigger competition was unfolding. In the week before the Spring Festival, China’s four tech giants (ByteDance, Tencent, Alibaba, and Baidu) spent a combined 4.5 billion RMB, roughly 625 million USD, promoting their AI assistant apps. The timing was deliberate: a nine-day national holiday gave hundreds of millions of smartphone users the downtime to try new things.

Yu Ren, an architect on Alibaba’s Qwen app, put it this way: « During the Spring Festival, people run into real needs around travel, spending, socializing, and entertainment. It’s a critical moment for AI to go mainstream. »

These are not just chatbots. They are evolving into all-in-one life management tools: ordering meals, booking rides, paying bills, managing schedules, all by voice. The goal is to make AI the default gateway for daily life, much like how WeChat and Alipay became super-apps for payments and social communication.

This « infrastructure-first » approach helps explain why AI adoption in China looks so different. Companies aren’t selling AI as a standalone product. They’re embedding it into systems people already use. Over time, these assistants will likely evolve into comprehensive personal managers. Sci-fi fans will recognize the vision: J.A.R.V.I.S., Tony Stark’s AI in Iron Man. Whichever company wins this race will shape China’s next-generation digital ecosystem.

Several things explain this velocity, and they feed off one another.

A market of 1.4 billion digitally connected consumers gives any new AI product the chance to amass millions of users almost overnight. That instant scale generates enormous volumes of real-world data, which developers pour back into improving their models at a speed few other markets can support. Meanwhile, the country’s power grid and telecom networks, built out over the past two decades, let services expand nationwide without the infrastructure constraints that slow deployment elsewhere.

Manufacturing adds another layer. The industrial clusters of the Pearl River Delta and the Yangtze River Delta can take a product from prototype to mass production in months. When a humanoid robot goes viral on national television, the factories to build it at scale already exist.

Chinese companies also tend to start from a specific consumer problem and work backward, rather than waiting for lab breakthroughs to find commercial uses. It is a model that prizes getting to market fast. And the government plays a coordinating role at every level, from national AI road maps down to city-level subsidies. The Spring Festival Gala itself is a case in point: part prime-time entertainment, part state-endorsed product launch, it folds industrial promotion into cultural broadcasting in a way that is particular to China’s system.

A technology this new, advancing at this pace, inevitably carries problems with it. Ahead of the 2026 Gala, market regulators summoned several major tech companies for talks, citing concerns over subsidy-driven user acquisition and price wars that risked squeezing smaller competitors. As the AI assistant race heats up, the business model remains an open question: will the massive marketing spend build sustainable ecosystems, or fuel an unsustainable race? AI’s rise has also prompted public debate about job displacement and data protection. After the Spring Festival, ByteDance disabled the Seedance feature that generated videos from real human faces, responding to growing privacy concerns.

China doesn’t wait for perfection, it ships, then refines. That means regulation will lag: some rules will arrive late, others won’t survive contact with the market. No country has figured out how to govern a technology it is still inventing. None of this means other countries should, or could, copy China’s path. Market size, regulatory traditions, and social expectations differ too widely. Countries will take different paths. Yet they face the same core questions: data governance, the gap between research and commercialization, and whether AI ultimately serves the public good.

Next February, the Spring Festival Gala will return. If the past two years are any guide, more companies will compete, the robots will move in ways we can’t yet picture, and the real battle will play out not on stage but in the AI apps that hundreds of millions of viewers open during the commercial break. When Chancellor Merz left the Unitree showroom in Hangzhou, the robots he watched were already last month’s models. That is the pace the rest of the world is trying to read.