CIRCUE: AI Drives the Revolution of Energy, Leading a Sustainable Future

Technology Author: Ziying Wu Mar 23, 2023 12:26 PM (GMT+8)

Increasing numbers of industrial and commercial businesses are embracing green practices in a big way. Based on AI, IoT (Internet of Things), and big data applications, Chinese energy AI technology companies with advanced cutting-edge technology have considered or tried to enter the oversea market. Zheng yu Chu, founder and chief executive officer of the a battery AI technology company CIRCUE, has shared his views from technology, business to globalization in this interview.


Over the past 40 years of reform and opening-up, China has continuously promoted its opening-up policy while practicing innovation, which has created opportunities for outbound investment. After joining the World Trade Organization (WTO), China has accelerated its opening-up process and realized the historical turn from "introducing" to "going out" and from following to leading. 

 As China's economy enters a new phase of steady growth, overseas markets become more attractive than in the past amid intensifying domestic volatility. In addition, the national government and relevant departments have been issuing a series of favorable policies in the field of international expansion. Besides, the increase in the level of openness towards the outside world has further motivated domestic brands to enter overseas markets. Lastly, the increasing improvement of the digital economy infrastructure including supply chain, mobile social commerce, the Internet and the rise of logistics, payment, SaaS/ERP and other tools have directly reduced the cost and significantly increased the convenience of going abroad. 

 As we enter this New Year, EqualOcean officially kicks off the "Global 100: Go Global and Co-creating the Future" project, in which our analysts will sit down with frontline industry practitioners and enterprise leaders to hear their opinions about Chinese brands going overseas. We hope to help brands be seen, recognized and valued through our professionalism and credibility. 

 The following is the sixth piece in this series, conducted by our analyst Ziying Wu after her talk with Zhengyu Chu, Founder and Chief Executive Officer of CIRCUE (Chinese:昇科能源).

截屏2023-03-23 11.54.41.png.png

CIRCUE (Chinese:昇科能源), with battery AI technology as its core, is committed to providing professional digitization energy services in the field of power batteries and energy storage systems. The company has established the first large-scale pre-training model PERB1.0 for battery timing data. The PERB1.0 model adopts more than 50GWh of battery data for training and can cover almost all battery types. And the team's core members are from Tsinghua University, MIT, the Chinese Academy of Sciences, and other famous research institutes. With such technological advancements, Circue has rapidly developed over the past years. 

EqualOcean: Could you please introduce the current development status of your company? Including but not limited to your business lines, development focus, etc.

CIRCUE: Our company targets the battery intelligence field, focusing on the intelligent usage of the battery system. It is noticed that a large amount of data will be generated within the battery usage process, so we focus more on the analysis of data and intelligent computing, using artificial intelligence technology to help us provide better solutions to our users.

Under our company targets, we focus on two main segments in our business. One is the intelligence of power batteries. Because of the rapid development of electric vehicles, the global power battery ownership volume exceeded 1 TWh last year. As a result, electricity consumption possesses strong distributed characteristics, i.e., each vehicle has about 50 to 60 degrees of electricity distribution, which is extremely uneven. In this case, daily charging, inspection, repair, maintenance, and trading of used cars, as well as the utilization and recycling of battery decommissioning, face many new challenges. These can be solved with battery intelligence technology, which is the first business segment we focus on. 

The second segment is the intelligence of energy storage batteries. The development stage of energy storage batteries is slightly slower than that of power batteries. But from these two years, because the proportion of installed renewable energy continues to increase, whether, from the power generation side or transmission and distribution or users of the industrial and commercial side, the demand for energy storage is growing.

 At the beginning of this year, China's wind and photovoltaic installed capacity exceeded coal power's. This is a landmark moment because the installed capacity of renewable energy is bringing the volatility of the entire power grid system up. We also see a current trend that the global energy storage market is rapidly exploding, although it is still small relative to the volume of power batteries. But we believe that by 30 years, the annual scale of energy storage will have a certain amount of volume.

From the energy storage side, our business starting point is mainly from the perspective of the safety of energy storage batteries. Because many energy storage batteries, especially electrochemical energy storage batteries, will be deployed in the city, and some European household energy storage requirements for safety have high standards. This year, the Chinese government emphasized the safety of electrodes, so our current strategy is to use intelligence to solve the problem of battery safety. In the future, we hope that each energy storage power plant can be a core component of a virtual power plant. We will pave our way to the virtual power plant market through energy storage power plants, which is our current company positioning and business focus segment.

In terms of business volume, we are still focusing on power batteries. Because power battery ownership volume has exceeded 1 TWh, so the demand for this will be surging. The demand for energy storage is still relatively before the inflection point, so the absolute volume will be smaller. But we are very optimistic about the incremental market of energy storage side. These reasons I think are from two points. One point is the pattern of the whole industry. The energy storage market will be relatively more market-oriented, especially the commercial and industrial markets will be more market-oriented. Second, in the power battery field, power battery is the core component of NEVs, and OEMs are more willing to power battery industry chain layout. Therefore, we believe energy storage batteries field is more suitable for startups to enter.

EqualOcean: We noticed that CIRCUE present the first large-scale pre-training model PERB1.0 (Pre-training Encoder Representation for Battery) at the 2022 World Power Battery Congress. Could you please briefly introduce the PERB1.0 model? Has this model been applied to CIRCUE’s product?

CIRCUE: PERB model is what we consider to be the core technology of battery intelligence. It also has a relatively high technical barrier. We created the model at that time because we found that to carry out our services well, we need to face a large number of batteries, i.e. managing thousands of batteries simultaneously and facing a volume of millions of tens of millions of millions of kilowatt hours to achieve real-time control.

The problem with these models is that there is no way for generalization because each battery must be modeled, and there is basically no reusable methodology between models. So what happens is that if the traditional technology is still applied and thousands of cells need to be managed, it doesn't reduce the cost as the data increases. The cost of our modeling is increasing with the size of managed cells. This is clearly not a very good technological choice.

The advantage of developing a large model is that it has a strong standardization capability so that it can be plugged directly into the large model no matter what kind of battery it is. There is no need to redo the model calibration or parameterization for new battery types. The big model is completely data-driven, and we have incorporated some battery mechanisms into the features of the big model. So the big model will continue to reduce the application and deployment cost as the amount of data and the size of the managed battery increases. This is what I think is the biggest advantage of our technology.

When we started to study large models a few years ago, we first used transformer technology. In fact, the ChatGPT released by OpenAI this year has also shown the powerful ability of this architecture in NLP, so we are also very happy to build this technology for the battery of continuous data, and not to migrate other such architectures horizontally. But some of our underlying elements, and some of our ideas are similar. So we also believe that the large model will have a powerful effect on future products. The application of technology must continue to reduce costs as the scale increases, not increase the costs.

EqualOcean: In the domestic market, with the large-scale application of AI technology, traditional BMS manufacturers have also started to invest in intelligent technology, and there are also start-ups entring into this field. How does your company view the competition? What is our core competitiveness compared to our competitors?

CIRCUE: I think the application of AI, including the application of AGI, will accelerate in various industries, especially from the end of last year to this year. But first of all, people may know more about ChatGPT or Stable Diffusion, the ideas and applications of which can be directly perceived by ordinary people, such as conversations or image generation. Although it is not clear to the general public, AI's big models will play a bigger and bigger role in many segmented industries, including the new energy battery industry, and some industrial sites and similar industries. I think AI will quickly penetrate into these industries, and gradually beat traditional technological practices.

One of the core competitive technological barriers is the scale of data. We can see from the development of ChatGPT, in fact, in GPT-1, it has about 100 million parameters, and by the time of GPT-3, the number of which has exploded to 175 billion. It is also because it has accumulated a large number of high-quality semantic databases and tags, so that this has triggered a quantitative change from a qualitative change. The same is true for other verticals like ours.

What we have been doing in the past three years is to accumulate a large amount of battery data, especially high-quality data. We now have 1.9 TWh of data available for training, far more than any competitor in this industry. Our models are trained and applied directly on these large-scale data. We have also established a closed loop by continuously collecting data in the process of promoting and selling our products. In addition to the technology of the big model, I think this is one of the most important barriers. Of course, the technology itself actually has some generational differences. As far as I know, no peer in the industry has been able to catch up with our current technological level.

EqualOcean: We know that your clients are mainly from energy storage stations, charging stations and OEMs. What specific business segments and client types are you currently focusing on? Which application scenario will you consider to be the promising growth point of your business in the future? 

CIRCUE: Though our past and current data are mainly based on power batteries, in fact, from the electrochemical system control theory, power batteries and energy storage batteries are not essentially different, all belonging to lithium battery systems. The physical laws followed are both porous electrodes plus molten liquid embedded in the embedded system. Whether it is a 3C power battery, ternary or 0.5C lithium iron phosphate battery, the essence is the same, only with different performance indicators, so we believe that technology can be shared.

From the business point of view, accumulating data in the early stage is a long-term process. The accumulation of data period will take at least a few more years before we can achieve the required volume level. After the data accumulation reaches a certain level, we will also do commercialization simultaneously, but we are not in a hurry to pursue rapid business growth.

Moreover, we should help our customers and consumers to improve their perception of the value of intelligent battery testing and evaluation, which is very important. I also use ChatGPT as an analogy. When you have not yet understood its capabilities, probably the most important thing is to let many people use it first, so our focus will also be placed here.

Currently, we have both domestic and international customers. For example, in the energy industry, our customers are mainly international energy companies like Shell and Total because they are more able to feel and understand the value of the AI we provide from the early stage. Our customers also include car companies. We are currently working with domestic and international OEMs like Daimler and BMW. In terms of awareness, they can understand what we are doing more.

EqualOcean: Focusing on the overseas market demand side, is there also an increase in the installed capacity of power batteries and electrochemical energy storage? Does your company have any overseas businesses at present? How do you view the prospect of battery intelligence service providers going overseas?

 CIRCUE: Overseas market in this field have a more mature and better business environment. From the user habits and characteristics perspective, overseas users are more mature about the concept of software payment. Many domestic users only want to pay for hardware, whereas software is free. Yet many overseas users can actually accept the cost of software services, so we are now also trying some ways to enter the overseas market, including joint ventures with some international enterprises to jointly develop products suitable for overseas markets and the launch of hardware products and software products. For example, as I mentioned earlier, we and our partner Shell are looking for opportunities with some other companies. But we don't want to catch up with the trend, we still want to lead and invest in it in advance.

EqualOcean: If you consider entering the overseas market, will you encounter any difficulties?

CIRCUE: So far, we have talked to some overseas customers and we think the biggest problem is data regulation. Because our service is based on data. If localized deployment is still good, but if you want to utilize cross-border data transmission, combined with the international situation and other aspects, you will encounter some policy and legal challenges and may need to make some adjustments to the whole company structure, so that it can comply with international regulations. The other is the establishment of overseas channels. These two points are not too much of a problem technically.

EqualOcean: Have you considered what kind of investors you would like to bring in in the next financing round?

CIRCUE: Our main goal is to bring in some strategic investors, whether overseas or domestic, to help us establish and operate more sales channels, which is one of our priorities to enhance our commercialization scale overseas and domestically.

In terms of usage, I think does not affect or overlap with our commercialization process. We can give it to everyone for free first. On the one hand, converting domestic software is much more difficult than that of foreign countries. The first reason is the payment habits of our domestic users; the other is that the system affects each company's budget, especially the ToB businesses. But if we only talk about implementation, we actually think it is better to promote in a way that people can feel value at a relatively low cost and technological threshold.

EqualOcean: Talking of the services, including battery recycling, testing, valuation and other system solutions you mentioned just now, which specific part that domestic ToB customers be more concerned about?

CIRCUE: Different scenarios have different customer concerns. For example, OEMs have more concerns about how to apply and recycle the sold cars in the future. It is used in different ways, and the final battery decay is different, so different batteries can be used differently. But how to carry out this more accurate evaluation and sorting, and finally applied in the appropriate scenario, has many enterprises in concern. We are also working with some enterprises to do this POC, and to help them to build a system.

EqualOcen: If considering overseas customers, are their focuses different from the domestic ones?

CIRCUE: For foreign customers, there are some concerns similar to those of domestic customers, like battery sorting. In addition, foreign customers will also pay more attention to V2G and the spot market of electricity. Since this year, foreign customers have also been paying more attention to energy storage, which is overall related to the transformation of energy systems and the challenges of power supply in Europe.

EqualOcean: Can you share the current market trends (opportunities and challenges) in the area of battery intelligence in domestic and overseas markets? How will the trend change in the next 1-3 years? 

CIRCUE: First of all, we can form a consensus that the scope of battery intelligence is relatively broad. We speak of battery intelligence broadly, as long as the intelligence related to the battery can be included. So like, power battery charging, detection and recycling, in fact, are counted in the battery intelligence category. Energy storage level is now mainly in its operation and maintenance, and these power market transactions are based on energy storage. So from this point of view, I think some major trends are relatively clear.

The first is the power battery, especially passenger cars, will evolve toward 800 volts with more power. More car companies will launch 360 kilowatt 3C batteries, with greater power, higher voltage level and faster charging speed. This is a core trend. The challenge really lies in the concerns of battery safety. If the control is not good, it will be relatively easy to encounter battery safety accidents.

The second is in heavy trucks, in commercial vehicles, at present, in the closed scenario heavy truck power exchange exists, and has achieved economic benefits. But in China and overseas there are still some differences. Overseas are still investing in super charging heavy trucks, while the domestic market now mainly or take the idea of power exchange, in charging and changing. In the detection scenarios, we will pay more attention to the battery in the gradient utilization scene of some applications. We hope to find some relatively large gradient utilization scene, such as the public launching this standard battery cell, and standard package. This is the standardization of the battery cell and package standardization. It can also promote the development of the whole gradient utilization application, to make gradient utilization easier to scale up.

On the energy storage batteries field, a future development trend is that the space of intelligent regulation and its development potential is huge. Energy storage batteries have lower design difficulty and technical threshold of which is lower than that of power batteries because their working condition is very simple, and the multiplier is not large. This is also a major trend when these storage stations and charging stations together constitute a distributed power system to earn more revenue from this power system. Overseas may be more concerned about supercharging, so we are working with some of our partners to promote high-power supercharging stations and systems. Currently, the demand from overseas departments is still relatively strong, but we think it is still relatively channel-dependent.

EqualOcean: What is your view on the future development of battery-intelligent platform operation and maintenance? Can you share the focus of your strategic goals this year? (including but not limited to resource synergies, industrial empowerment, etc.)

CIRCUE: We still focus on our core strategic goals. We have formed a charging and testing process for power batteries as an enabler of the business scene, so we are now a relatively stable growth of these two scenarios. However, we think we can't be too hasty because we have not yet reached the outbreak of business, so we will make some attempts to constantly adjust the user's perception of value and the shape of our products. Our second focus is on energy storage for commercial and industrial use, especially for charging station scenarios. There are three major blocks of energy storage, including power generation side, grid side and commercial and industrial applications. Commercial and industrial applications are mainly plant-side applications and have a charging station. We already have a lot of customer resources and foundations on the charging station business side so that we will focus on the charging station business side in the future.