Quantum Meets Intelligence: China’s Wukong Supercomputer Boosts AI Performance to New Heights
First Encounter: Chinese AI Meets Quantum Power and Gets Smarter, Faster
In a landmark breakthrough that could reshape the future of artificial intelligence (AI) and computing, Chinese scientists have successfully used a real quantum computer to fine-tune a massive AI model — marking the first time such a feat has been achieved globally.
At the heart of this accomplishment is Origin Wukong, China’s third-generation superconducting quantum computer. Developed by Hefei-based startup Origin Quantum, the machine features 72 qubits and has now demonstrated that quantum computing is not just a futuristic concept, but a powerful tool ready to accelerate AI training in real-world applications.
The AI model used in the experiment boasted an impressive one billion parameters. Traditionally, training such large-scale models demands huge computational resources, which often leads to high energy usage, long processing times, and limited scalability. But with Wukong, researchers achieved an 8.4% improvement in training performance while reducing the model size by 76%. That means better performance with fewer resources.
Why This Matters
AI training, especially for large language models (LLMs), typically relies on classical computers that process tasks sequentially. Quantum computing, on the other hand, leverages the principles of superposition and entanglement, enabling it to process massive combinations of variables all at once. This parallelism is what gives quantum computers the edge in speeding up complex calculations.
For this experiment, the researchers introduced a novel method called quantum-weighted tensor hybrid parameter fine-tuning. This technique integrates quantum and classical computing. Quantum circuits are used to find deeper patterns in the data, while classical systems compress and optimize the model, making it more efficient.
The results were impressive. On a mental health dataset, the AI made 15% fewer errors in response generation. On math problem-solving tasks, accuracy jumped from 68% to 82% — a clear indicator that quantum tuning enhances the AI’s ability to understand and reason.
The Rise of Wukong
Launched in January 2024, Origin Wukong is already considered one of the most advanced quantum systems available commercially. What’s remarkable is that over 80% of its hardware and software components are made in China, giving it a technological independence rarely seen in the global quantum race.
The platform has attracted over 20 million visits from users in 139 countries in just a few months, completing tens of thousands of tasks across industries like biomedicine, finance, and fluid dynamics. Despite growing technological rivalry, users from countries such as the United States, Japan, Canada, and Russia remain among the most active international participants.
Interestingly, Chinese scientists emphasized their openness. "While US quantum systems are closed to China, we continue to believe in scientific exploration beyond borders," said Guo Guoping, a top physicist and co-founder of Origin Quantum.
A Technological Shift on the Horizon
This breakthrough comes at a time when the global tech landscape is witnessing significant shifts. The rapid rise of Chinese AI platforms like DeepSeek, which is being hailed as a potential alternative to Western giants like OpenAI, is already raising eyebrows in the US. The addition of powerful quantum hardware into China’s AI arsenal may further tilt the balance.
Quantum computing, once seen as decades away from practical use, is now beginning to deliver real impact. With Origin Wukong 2—an even more powerful fourth-generation quantum machine—nearing completion, China is signaling its intent to lead the convergence of quantum and AI technologies.
For now, this “first encounter” between quantum power and artificial intelligence isn’t just a scientific milestone—it’s a glimpse into the future, where machines get smarter, faster, and possibly more human-like than ever before.
