Liangkun Technology erhält mehrere Hundert Millionen Yuan in Angel-Runde, AI4S braucht dringend quantenpräzise Daten | Exklusiv von 36Kr

Exclusive information obtained by "Hidden Currents Waves" reveals that the quantum computing company "Liangkun Technology" recently completed an Angel round and an Angel + round of financing worth hundreds of millions of yuan. This series of financings was led by the Inno Angel Fund, and several institutions such as Guoqi Investment, Beigong Investment, BV Baidu Venture Capital, Tsinghua Alumni Fund, and Mingzhi Venture Capital participated in the investment. Light Source Capital served as the exclusive financial advisor.

Behind this financing lies an increasingly clear assessment: AI for Science requires quantum computing.

AI can learn patterns, but the upper limit of model capabilities is restricted by the "resolution" of the world it has seen. In R & D scenarios such as chemistry, materials science, and pharmacy, the prediction results of the model are significantly limited if the accuracy of the underlying data is insufficient.

Quantum computing is naturally suitable for simulating molecular structures, chemical bonds, and other systems. As a high - precision solution finder, it can provide calculation results that are closer to the laws of the physical world; the high - precision quantum data obtained through calculation are also a key to improving the model performance of AI4S.

Liangkun Technology was founded in January 2026. The founder, Lü Dingshun, worked in the AI4S labs of Huawei and ByteDance for seven years and led the team in exploring the limits of quantum computing capabilities. Even earlier, he was one of the first doctoral students in the field of quantum computing at Tsinghua University and was deeply involved in the construction of an ion - trap quantum computing system.

In the past, Lü Dingshun achieved many results in large companies with the approach of "less hardware, more software first". In his view, a heterogeneous intelligent computing platform that combines quantum technology, AI, and high - performance computing can maximize the value of limited quantum computing power at the application level.

This young doctor of natural sciences has always hoped to solve real - world problems with quantum computing. When the hardware technology route of quantum computing was not yet clear, he did not rush into the hardware startup wave. Instead, he chose algorithms and software platforms above the hardware. He packaged quantum algorithms, AI models, and industry workflows into callable scientific intelligence agents to connect quantum computers with the application needs of AI4S.

Lü Dingshun speaks very fast. In an almost two - hour interview, he said "exciting" 15 times. In his first year at Huawei, the research that challenged Google's "quantum supremacy" didn't particularly excite him. But the achievement of the state - of - the - art (SOTA) in high - temperature superconductivity model calculations with AI excited him; he also gets excited when he meets people who can challenge Google and IBM and fight tenaciously.

Currently, the Liangkun Technology team has nearly 40 employees and has brought together leading talents in the fields of quantum computing, AI, and high - performance computing. In Lü Dingshun's view, "the team is a reflection of the founder's inner understanding. If you understand the quantum computing system deeply, you know how to recruit the team. The most important thing about the talents is that they are motivated."

Why does AI4S need quantum computing? How big are the business opportunities at the algorithm and operating system levels? Will quantum computing be a new solution for computing power in the future? Below is our interview with Lü Dingshun (edited):

I. The Journey So Far

"Hidden Currents": Why was the "quantum supremacy" so impressive? As one of the first quantum computing doctoral students at Tsinghua University, why did you decide to enter the industry?

Lü Dingshun: In 2019, Google released a processor with 53 usable qubits and completed a study in just 200 seconds. It was claimed that the same task would take 10,000 years with the most powerful classical computer at that time. This is the origin of the "quantum supremacy".

Later, we worked at Huawei for a year and simulated with traditional GPU clusters with hundreds of cards. Through algorithm optimization, we confirmed that a classical computer doesn't need 10,000 years to complete the task, but only a few months or even days. This research challenged Google's quantum supremacy.

But after completing this work, I wasn't particularly excited. Because quantum computers are still evolving, and scaling up (exponential expansion) is obvious. With 53 qubits, you can still keep up, but when it comes to 60 or 100 qubits, it will be difficult for classical supercomputers to keep pace.

What really matters to me is what real - world problems quantum computing can solve as the capabilities of quantum computing hardware continue to improve. Can the problems to be solved become even bigger? Quantum computing is a systematic project, so I decided to enter the industry.

"Hidden Currents": How did you look for real - world applications for quantum computing during your time at Huawei?

Lü Dingshun: A quantum computer is like a hammer, and you have to find the right nail.

Besides simulating random circuits, there were two other exploration experiences. The first was simulation in chemistry and materials science. Since a quantum computer itself is a microscopic quantum system, it makes sense to use it to simulate another quantum system, such as in materials chemistry. Before entering the industry, I hadn't studied chemistry, so I spent three months reading articles on computational chemistry, writing algorithms, and reproducing the experiments. Later, we extended the quantum chemistry simulation to 28 qubits, which was the largest simulation in the industry at that time.

The other exploration was solving combinatorial optimization problems, such as the maximum - cut problem and network traffic optimization. With the limited computing power of the quantum computer, we carried out dimensionality reduction based on the QAOA algorithm (Quantum Approximate Optimization) and finally simulated a business scale of 100,000 qubits with less than 20 qubit computing resources.

"Hidden Currents": When did you focus more on the AI4S application scenario? How did the idea of this "mixed heterogeneous computing platform" come about?

Lü Dingshun: At ByteDance, we still initially followed the logical principle of "the feasibility of quantum computing". If a quantum computer only has 20 - 50 qubits in the long run, how can we solve real, large - scale problems?

Later, I found that the "quantum embedding approach" is a good idea. Simply put, it means using the best material in the most important places. By decomposing the computing tasks, the quantum computer solves the most central and complex problems, while the classical computer calculates the rest. In this way, a balance can be achieved between computing scale, accuracy, and cost.

For example: There are two computers on the conference table, representing the most important features, while the rest are similar. Then we use the quantum computer to calculate the "computers". Regarding specific application scenarios, we chose materials with complex electronic structures and traditionally difficult - to - solve problems, such as transition metal oxides like nickel oxide.

With the rise of the capabilities of large AI language models, the team has more strongly oriented its strategy towards applications. In the past, we used the quantum computer as a hammer to look for the right nail; now we use AI, quantum computing, and classical algorithms together as long as we can solve scientific problems.

In the fields of chemistry and materials science, we explored three approaches: The multi - scale quantum chemistry simulation, which reduces a problem that originally requires thousands of qubits to 20 qubits; the quantum computer as a high - precision solution finder, which provides high - quality data for AI4S models. The GPU - based quantum embedding algorithm, which does not depend on the improvement of quantum computing hardware; and the solution of physical problems purely based on neural networks and quantum states, which functions both as a solution finder and a data synthesizer.

"Hidden Currents": You emphasize that the problems to be solved are "big" enough. What is the most important thing when exploring these applications?

Lü Dingshun: The most important thing is "topic selection". You have to find a problem that has a big enough impact.

Later, we chose "high - temperature superconductivity", a problem that attracts a lot of attention in the field of condensed matter physics and is also noticed by ordinary people. With the help of artificial neural networks, we achieved the SOTA in the calculation of the Hubbard model for high - temperature superconductivity.

This really excited me. Compared with the traditional computing paradigm, our algorithm already showed an advantage in the second decimal place, while the previous scientific community was competing in the fourth decimal place.

This AI model is not a traditional data - driven algorithm. Essentially, it solves the extremely complex Schrödinger equation based on the "variational principle" and finds the real ground - state solution by continuously minimizing the loss. From the perspective of the first principle, it can be extended to many problems in chemistry and materials science.

At first, this method consumed a lot of computing resources. Then we improved the algorithm and the framework, which greatly reduced the computing power requirement and enabled more research teams to participate.

II. The Present

"Hidden Currents": How do you understand your position in this systematic project of quantum computing?

Lü Dingshun: In the field of quantum computing, many companies are working on the development of quantum computer hardware to solve the problem of basic computing power. The demand at the top - level application level is also very high. Users want to use quantum computing and AI to solve real - world problems, such as in the development of semiconductor materials, chemical materials, and new drug molecules.

But there is a lack of algorithms and software tools between the hardware computing power level and the application level. The operating system for quantum computing power is exactly the position we want to take.

Source: Liangkun Technology

"Hidden Currents": How do you understand the technological barriers in the development of algorithms and tools at the middle level? Why did you choose the business opportunities at the algorithm and operating system levels?

Lü Dingshun: The middle level is not simply programming existing algorithms. Especially when the hardware resources of quantum computers are still limited.

The limited resources mean that not all algorithm paths can fulfill the task. Since errors in quantum computing accumulate, you can only maximize the limited quantum computing power if you optimize the algorithms sufficiently and make the path as short as possible.

This is different from running algorithms on a GPU. On a GPU, a worse algorithm may run slower, but it still works, albeit at a higher cost. In quantum computing, an algorithm that is five times slower may not run at all. This is the difference between 0 and 1.

The barrier in algorithms lies in whether you can design and modify the algorithms skillfully. After these algorithms and the operating system platform are built, the functions can be continuously expanded and gradually develop into an algorithm and tool platform.

"Hidden Currents": Which users do you want to serve in the current quantum computing industry landscape?

Lü Dingshun: The first group are customers who already have a demand for quantum computing, such as state - owned enterprises and research institutions. They need to develop the capabilities of quantum computing and improve quantum algorithms. This group usually starts with tools, decomposes the problem into quantum algorithms, and then runs them on the corresponding quantum computer.

The second group are industrial customers with clear R & D needs, such as companies in semiconductor and pharmaceutical development. Users don't care whether the computing power comes from a quantum computer. Instead, they are interested in solving the problem and cost - efficiency. In the solution method, they can use AI algorithms, quantum algorithms, or mixed algorithms of quantum computing and classical methods. (The mixed algorithm entrusts the most difficult and central problem to quantum computing, while the rest is handled with neural networks, classical algorithms, or other accurate methods)

Quantum computer manufacturers are also our partners and customers. Many companies focus on hardware development, while the operating system, algorithms, and application ecosystems require a professional team and long - term investment. In cooperation, we can, for example, sell the operating system and the algorithm platform together with the hardware, offer computing power, or the entire computer with the operating system.

"Hidden Currents": There are many AI4S companies, and the financing is also very hot. Why is quantum computing absolutely necessary?

Lü Dingshun: From the perspective of pure AI for Science, AI is a solution, and quantum computing is also a solution. Besides the fast computing speed (quantum acceleration), accuracy is also an advantage of quantum computing.