AI News Feed
Market watch
Study Finds Climate Feedback Loops Could Add 0.2C to 0.4C by 2100iRobot’s Roomba Duo Concept Puts a Tiny Robot Vacuum in Its HatAnt Lingbo, Amap and Daxiao Robotics Push World Models Toward Single GPUs, 3D Cities and Robot SkillsTencent's WorkBuddy Opens AI Office Ecosystem to Software and Hardware PartnersAlipay's AI Assistant Abao Adds Fliggy Flight and Ticket BookingQuantum Computing Is Not a Supermachine for Everything, Chinese Scientist Says at Bund Conference
Study Finds Climate Feedback Loops Could Add 0.2C to 0.4C by 2100iRobot’s Roomba Duo Concept Puts a Tiny Robot Vacuum in Its HatAnt Lingbo, Amap and Daxiao Robotics Push World Models Toward Single GPUs, 3D Cities and Robot SkillsTencent's WorkBuddy Opens AI Office Ecosystem to Software and Hardware PartnersAlipay's AI Assistant Abao Adds Fliggy Flight and Ticket BookingQuantum Computing Is Not a Supermachine for Everything, Chinese Scientist Says at Bund Conference
Research

Quantum Computing Is Not a Supermachine for Everything, Chinese Scientist Says at Bund Conference

At the 2026 Inclusion Bund Conference in Shanghai, USTC professor Lu Chaoyang said quantum computing remains far from broad practical use and lacks accepted evidence of practical advantage in finance or large-model training, while posing a long-term threat to public-key cryptography.

AI News Feed · 2026-09-10 17:58

Speaking on the main forum, Lu said quantum computing is important, but it is not a supermachine that speeds up everything. It can offer genuine speedups only for a small number of problems with special mathematical structures, he said. He described the field as highly competitive while still limited in practical capability, saying, 'Quantum computing, although very competitive, is currently still quite weak.'

Lu explained why quantum computing is often imagined as universally powerful. A quantum bit can exist in a superposition of 0 and 1, and combining multiple qubits expands the state space rapidly. But he cautioned that quantum computing is not simply parallel computing. Although operations can be applied to a superposition as a whole, measurement yields only one result. Quantum speedup does not happen automatically; it requires a problem with special mathematical structure and a suitable quantum algorithm.

The problems widely recognized as suited to quantum computers remain few, he said. They include cryptographic problems such as large-number factorization and discrete logarithms, and simulation of quantum systems themselves, including quantum many-body systems and problems related to materials and chemistry.

That helps explain why 'quantum finance' remains more hype than solid science, Lu said. 'So far, for the financial sector, there is no scientifically recognized evidence that a clear quantum advantage can be obtained under practical conditions,' he said. He similarly rejected the idea that quantum computing is a 'super GPU' for today's large models. 'For an area as important as today's large models, the scientific community still does not have a recognized quantum algorithm that proves a clear quantum speedup for large-model training under practical conditions,' he said.

In Lu's view, the most important direction for quantum computing is not financial modeling or large-model training but information security. If a sufficiently powerful fault-tolerant quantum computer appears, widely used public-key cryptosystems such as RSA could face challenges. The elliptic-curve digital signature algorithm that Bitcoin relies on would also be threatened by Shor's algorithm, he said.

Lu also reviewed a landmark moment in the field. In 2019, Google used its 53-qubit Sycamore processor to complete a random circuit sampling experiment, claiming the quantum processor needed about 200 seconds for a task that the strongest classical supercomputer at the time would need about 10,000 years to finish. The result was seen as a representative moment of 'quantum supremacy.'

As classical algorithms advanced, Lu's team at USTC and the Shanghai Artificial Intelligence Laboratory developed a new classical simulation algorithm that used GPUs to compress the calculation time for a similar task to about 17 seconds, with lower energy consumption than the quantum experiment at the time, according to Leiphone. Chinese teams also advanced the 'Jiuzhang' series of photonic prototypes, achieving new milestones on specific tasks. Lu said this meant China not only participated in the quantum computing competition but also broke Google's original 'quantum supremacy' narrative at a key node.

Still, Lu repeatedly emphasized that today's quantum computers are far from practical use. The problem is not only the number of qubits but also their quality and stability. As qubit counts rise and quantum circuits deepen, errors accumulate rapidly. Without effective quantum error correction, a larger system may produce less reliable results. 'If we want to build a skyscraper, we need a more solid material. This is called quantum error correction,' Lu said.

He noted that one of the world's strongest quantum computing teams is still soliciting genuinely valuable quantum algorithms from around the world. When the strongest teams are still searching for applications, the market should be cautious about claims that quantum computing has already been fully deployed, he said. For some companies' claims of using quantum computing for new drug discovery, quantum finance, or even 'quantum pig farming,' Lu said many such application scenarios look quite absurd to academia.

In his view, quantum computing companies worth supporting are not those that tell the best stories, but those that build real quantum hardware and rigorously seek real application scenarios. The value of quantum computing is not that it will immediately change everything, but that it will suddenly become very strong on a small number of key problems, and those problems may prove critical.

The 2026 Inclusion Bund Conference was held in Shanghai from Sept 9 to 12 under the theme 'Co-creating AI New Economy,' exploring how AI can promote more sustainable economic growth and broader social value.