Chinas KI-Offensive: Wie das Reich der Mitte mit End-to-End-Anwendungen den wahren Wert der KI-Revolution abschöpfen will

Während der Westen über KI-Modelle philosophiert, setzt China auf die harte Währung der Technologie: Umsatz.

Vom Labor auf die Straße – und in die Kasse

Die Strategie ist klar. Statt sich im akademischen Wettlauf um die größten Parameter zu verlieren, konzentriert sich Peking auf vollständige Anwendungsketten. KI soll nicht nur denken, sondern in Fabriken, Krankenhäusern und Städten arbeiten – und dabei messbaren wirtschaftlichen Mehrwert generieren. Ein pragmatischer Ansatz, der Forschung direkt mit kommerziellem Output verknüpft.

Die Infrastruktur als Wettbewerbsvorteil

Der staatlich geförderte Ausbau von 5G-Netzen, Rechenzentren und IoT-Sensoren liefert den perfekten Nährboden. Diese integrierte Infrastruktur ermöglicht es, KI-Lösungen nahtlos vom Entwickler zum Endnutzer zu skalieren. Ein Vorteil, den fragmentierte Märkte nur schwer kopieren können.

Der globale Kontext: Ein Rennen um die Vorherrschaft

Diese Fokussierung auf End-to-End-Lösungen stellt eine direkte Herausforderung für westliche Tech-Giganten dar. Es geht nicht mehr darum, wer die intelligenteste KI baut, sondern wer sie am effizientesten monetarisiert. Chinas Ansatz könnte den Wertfluss der gesamten KI-Branche neu definieren – weg von Lizenzgebühren für Modelle, hin zu proprietären Ökosystemen.

Ein kluger Schachzug in einem milliardenschweren Spiel. Oder, wie es ein anonymer Finanzanalyst sarkastisch anmerkte: 'Endlich eine KI-Strategie, die sogar ein Value-Investor verstehen kann – sie konzentriert sich auf das Einzige, was zählt: die Gewinn- und Verlustrechnung.' Die Börsenkurse der beteiligten chinesischen Tech-Firmen scheinen ihm vorerst recht zu geben.

Cheaper energy offsets chip disadvantage for AI startups

Several Hangzhou startups are preparing to sell shares to the public. Manycore, which specializes in spatial intelligence, along with robot manufacturers Unitree and DEEP Robotics – part of a group locals call the “six little dragons” – plan to list in Hong Kong or on mainland Chinese stock exchanges, joining other AI companies going public.

Victor Huang, who helped start Manycore after working as a software engineer at Nvidia, said his company relies on chips from the California-based manufacturer because they provide better computing power for the energy they use. However, he pointed to an advantage China has: cheaper electricity costs.

Huang explained that a three-nanometer chip uses roughly 30% less power than chips that are five or seven nanometers. But companies can still compete if their electricity costs are 40%-50% lower, he said.

“Computing power cannot be viewed in isolation,” Huang told CNBC. “It depends on data quality, energy supply and operating conditions.”

Manycore made its spatial AI model available for free, a strategy China favors, unlike many U.S. companies such as OpenAI and Anthropic that charge for access. Huang said this lets the company collect user feedback, though it also limits income since people don’t have to pay.

“So you’ll get pressure from investors,” he said.

China’s AI development has concentrated on practical uses rather than pursuing the kind of super-intelligent AI that fascinates many in Silicon Valley. Examples include personalized suggestions from Baidu Map and ByteDance’s chatbot called Doubao.

In December, Doubao led China’s AI apps with 155 million people using it each week, almost twice as many as its nearest rival, DeepSeek‘s chatbot, according to QuestMobile, a company that tracks business data. Doubao’s success shows that ease of use and practical value might matter more than technical complexity.

Liangzhu emerges as China’s experimental AI hub

A more relaxed, experimental approach is growing alongside these commercial efforts. While major players like Alibaba and DeepSeek focus on advanced AI, Liangzhu has become the center of more unusual AI projects. 

After relocating to Liangzhu in 2025, Alex Wei is developing an AI software that is based on conventional Chinese fortune-telling techniques. He’s investigating how AI can meet people’s emotional needs.

Developers are drawn to Liangzhu because of its low business pressure. “You can come to Liangzhu with 1,000 renminbi ($143) and leave with your product demo,” Wei stated. “It’s a really welcoming location. You can find assistance for even a little app that serves a thousand users; you don’t need to have a unicorn product.”

This attention is changing how startups plan for growth. Many are targeting users outside China, with some founders planning to use China’s manufacturing networks to offer lower prices globally. Tough competition at home and Chinese consumers’ unwillingness to pay for apps have also pushed startups to seek international markets, observers say.

Afra Wang, who publishes the Concurrent newsletter about China and Silicon Valley, said some developers are using AI to break away from traditional jobs in uncertain employment markets. They’re working to become “superindividuals” who run profitable businesses alone or with very small teams.

Wang cautioned that some businesses are simply adding AI features for marketing purposes, from air conditioners to mirrors that check if sunscreen is properly applied. She describes some of these as “physical AI slop”, borrowing the term used for poor-quality content produced by AI.

For now, Hangzhou’s business owners are trying almost every concept, from practical to playful, in a fast-changing market.

The smartest crypto minds already read our newsletter. Want in? Join them.