36Kr Exclusive | Former Huawei VP Launches AI-Native Power Asset Operation Startup, Secures Tens of Millions of US Dollars Series A Financing From the Fund Under CATL
Author | Qiao Yujie
Editor | Yuan Silai
This article is approximately 3600 words, with an estimated reading time of 9 minutes
Hard Krypton learned that Green Voltis, a Sweden-headquartered AI-native power asset operator, recently announced the completion of a multi-million-dollar Series A financing, exclusively invested by Lochpine Capital, a USD fund established with the support of CATL.
The funds from this round of financing will be mainly used for three aspects: continuing to increase R&D investment in the Aether AI power trading agent and the Chronos grid-connected control system, expanding business in key European markets, and making layouts in the AI infrastructure sector.
Yao Hong Shi, founder of Green Voltis, worked at Huawei for 16 years before starting his business. He once served as Vice President of Huawei, successively holding positions including President of the Middle East Region, President of the Software Product Line, Executive Vice President of the Government and Enterprise Department of China Region, and President of Cloud & Computing China Region, and was long responsible for Huawei's overseas business expansion and cloud computing business. After leaving Huawei in 2023, he devoted himself to the European power market, operating flexible assets with AI trading and energy management systems.
The core team of Green Voltis also brings together talents from leading ICT enterprises such as Huawei, Ericsson and T-Mobile, as well as professionals from power system and power trading specialist companies such as Rolls-Royce, Brady and Varberg Energi.
Nowadays, the competition for AI computing power is pushing the power industry to a new bottleneck. In the United States, some data centers choose to obtain power by building self-owned gas power stations and other methods. In Europe, where natural gas is highly dependent on imports, carbon pricing and climate legislation make new fossil-fueled power sources neither economically viable nor easy to obtain approval, while the expansion speed of power grids is far behind the growth of computing power demand. In core hubs such as Frankfurt, Amsterdam and Dublin, the queuing period for data centers to connect to the grid can be as long as 7 to 10 years.
Against this backdrop, whether a data center can become a regulable flexible asset for the power grid has begun to affect its efficiency of obtaining power. For flexible assets such as energy storage, how to increase revenue through trading and control systems is also directly related to the economic viability and financing capacity of the project.
The entry point of Green Voltis is to provide one-stop services from grid connection to trading for independent energy storage and wind-solar integrated storage projects. The company's local engineering team is responsible for grid connection commissioning and auxiliary service qualification certification, helping projects to meet operating conditions as soon as possible. The self-developed Chronos Energy Management System (EMS) is in charge of on-site power station control and supports grid-forming operation. The company hopes to transform energy storage equipment from a pure hardware asset into an operating asset that can participate in the power market and continuously generate revenue through the collaboration between the control system and the trading platform.
Among them, the Aether AI power trading agent is the company's core R&D investment direction. Different from traditional process-based software that mainly executes transactions according to preset rules, Green Voltis chose an AI-native architecture from the very beginning of its establishment, hoping to connect market forecasting, trading decision-making and execution, and dynamically optimize revenue opportunities in different markets by leveraging electricity price forecasting and quantitative strategies.
The trading mechanism of the European power market provides application space for this system. According to Yao Hong Shi's introduction to Hard Krypton, the intraday market in Europe generally adopts a continuous trading mechanism, where prices fluctuate continuously with changes in supply and demand. Meanwhile, on the premise of meeting market rules and delivery constraints, energy storage assets can participate in multiple markets and products such as day-ahead, intraday and frequency regulation. How to allocate the limited energy storage power and capacity across different markets has become the key factor affecting revenue.
The strategy adopted by Green Voltis is called Revenue and Capacity Stacking. In simple terms, instead of choosing only one market in each time period, the system comprehensively predicts the price and capacity demand of different markets, and collaboratively configures multiple trading products on the premise of meeting all delivery constraints. To implement this strategy, the company not only develops its own trading platform, but also independently develops EMS, so as to combine the power curves of different products and coordinate trading decisions and on-site control.
Licensing is another threshold. Green Voltis chooses to hold key power market licenses on its own, instead of relying entirely on third-party licensed institutions to carry out transactions. According to Yao Hong Shi, there are only 2 VPP companies holding BRP licenses across Sweden at present. The company has already obtained access to the power markets in 10 European countries.
In terms of revenue performance, the data provided by the company shows that the revenue of its managed energy storage assets continues to outperform the Clean Horizon energy storage revenue index, while the industry average level in the same period is about 75% of this index.
In addition to flexible asset operation, Green Voltis is extending the same set of capabilities to the AI infrastructure sector. Its judgment is that the core problems faced by European data centers are not only building computer rooms and purchasing GPUs, but also how to obtain power and how to put computing power into operation as soon as possible while meeting the requirements of the power grid.
Therefore, Green Voltis positions itself as a Power-to-Compute operator rather than a traditional data center developer: it connects the power grid on one end and computing power on the other, trying to convert scarce grid-connected capacity into quickly deliverable computing power resources. The company plans to cooperate with partners to promote data center projects, starting from sites that meet construction conditions, to open up links such as medium and high voltage power access, power supply and distribution, energy storage and GPU power control, so that data centers can cooperate with the power grid to adjust loads when necessary.
Yao Hong Shi revealed that the company has currently reserved about 120 megawatts of data center land (Powered Land) with grid connection conditions in Northern Europe, and the company hopes to further integrate power access conditions with power supply and distribution, computing power deployment and flexibility control capabilities in subsequent projects.
The following is an excerpt from the conversation between Hard Krypton and Yao Hong Shi:
Hard Krypton: Why did you choose to enter the European power market after leaving Huawei?
Yao Hong Shi: I was in charge of cloud computing related business during my time at Huawei. Back then, we hoped to empower thousands of industries through cloud and AI, but we gradually realized that without in-depth understanding of the industry, it would be difficult to truly create value for customers. Therefore, after leaving Huawei, I wanted to bring the technologies and experience I accumulated in the ICT field to specific industries, and deeply explore the opportunities of integrating AI with real industries.
The energy market has a large volume and many changes, with both complex industry rules and a large number of opportunities worth exploring. At the same time, many of the businesses we did in the past are software-oriented and operation-oriented, which share certain common underlying logic with power trading and new energy asset operation. Of course, the professional threshold of the energy industry is very high, so we have attached great importance to cross-domain team building since the very beginning of our entrepreneurship. The co-founder and head of R&D are from Huawei, and the core personnel related to power solutions are mainly recruited in Europe. We hope to combine ICT technology accumulation with local power industry experience to solve practical problems in the energy market.
Hard Krypton: What are the core competitive barriers of the company? How can they be converted into the revenue advantage of energy storage assets?
Yao Hong Shi: Our competitive advantages are mainly reflected in three aspects.
The first is licensing. We chose to hold key licenses on our own from the very beginning of our entrepreneurship, including BRP (Balancing Responsible Party), BSP (Balancing Service Provider) and power retail related licenses. In this way, we can independently formulate and execute trading strategies without being restricted by other licensed parties. Licensing determines whether we can give full play to our trading capabilities. In the Nordic market where we have carried out power trading, there are not many start-ups in the VPP (Virtual Power Plant) field that can hold relevant licenses on their own.
The second is technical architecture. We chose to adopt an AI-native technical architecture to solve power trading problems as early as the start of our business in 2024. Compared with traditional process-based software, we hope to process market information, predict price changes and execute transactions in a more timely manner through AI and quantitative technology, thus supporting higher-frequency trading decisions. From the market situation we have observed so far, among enterprises that have reached a certain scale, the number of those using traditional process-based software is still relatively large.
The third is trading strategy. We adopt a multi-market and multi-product collaborative trading strategy, namely Revenue and Capacity Stacking. In simple terms, within the same time period, we participate in transactions in multiple power markets at the same time according to the prices, capacities and delivery rules of different markets, instead of choosing only one market in each time period. To implement this strategy, in addition to the capabilities of the trading platform itself, the cooperation of the underlying Energy Management System (EMS) is also required. One important reason why we chose to develop EMS independently is that the common local EMS products in the European market at that time generally did not support the power superposition function we needed. Trading products in different markets have their own power curves and delivery requirements, which require combining multiple curves while meeting all constraints, placing very high demands on the system's control and optimization capabilities.
Hard Krypton: The company has obtained access to power markets in 10 European countries. How does the AI trading system balance the cross-market reuse of underlying capabilities and localized adaptation for different markets?
Yao Hong Shi: The power markets of European countries are gradually moving towards unification. Institutions such as the European Network of Transmission System Operators for Electricity (ENTSO-E) have promoted a series of common standards, and the day-ahead and intraday markets have also continuously strengthened linkage through corresponding market coupling mechanisms. In general, a considerable part of the rules and technical requirements in the European market are common.
However, due to the different histories and energy structures of power systems in various countries, specific market rules still vary. Therefore, we do not directly cover all countries with a completely unchanged model, but adapt to the rules of different markets on the basis of a unified technical platform and core capabilities. For regions with similar rules, the adjustment cost is relatively low; for markets with large differences, more targeted training and algorithm iteration are required.
Among them, the differences in the intraday market are particularly worthy of attention. Different countries have different proportions of fossil energy, nuclear power and renewable energy, and the fluctuation characteristics of power supply and demand will also be different, which will affect the data required for price forecasting, as well as the way of model training and iteration. Our goal is to adjust according to the specific situation of each market while reusing the underlying technology and trading capabilities as much as possible.
Hard Krypton: You are not a traditional data center developer, why do you want to make layouts for Power-to-Compute? What does the current project reserve of about 120 megawatts mean?
Yao Hong Shi: We have always hoped to carry out business around the intersection of AI and power. Initially, we mainly deployed flexible assets such as energy storage, and later expanded to projects such as photovoltaic power and wind power. With the development of AI infrastructure, data centers have become an increasingly important variable in the power market. In Europe, data centers not only need green power, but also face grid access restrictions. If a certain degree of flexibility cannot be provided, grid connection approval and power access may require a very long waiting period. We hope to apply capabilities such as grid connection pre-certification, power trading and self-developed EMS to the AIDC scenario, so that data centers can become flexible assets that can collaborate with the power grid.
Our positioning is Power-to-Compute, which means opening up the chain from power to computing power, and working with partners to promote the integration of medium and high voltage power access, power supply and distribution, and GPU power control. The about 120 megawatts is the scale of projects with land and grid connection conditions that we have reserved in Northern Europe at present, which does not mean that all these data centers have been completed or put into operation. The key next step is to convert these conditions into computing power projects that can be actually delivered.
Hard Krypton: This round of financing is exclusively invested by the fund under CATL? How will the cooperation between the two sides form complementarity?
Yao Hong Shi: The European energy storage market is transforming from simply selling equipment to end-to-end services such as system integration, engineering delivery, operation and maintenance, and power trading. For customers, to evaluate the value of an energy storage system, in the end, we cannot only look at the performance of batteries or equipment, but also see whether it can truly improve the revenue of the project.
If an enterprise can integrate equipment, engineering delivery, operation and maintenance, and power trading, it can be more directly responsible for the actual revenue of customers. However, for many energy storage integrators, it is not easy to build their own trading capabilities in Europe, so cooperating with professional power trading companies is a feasible path. Our cooperation with CATL is built on the basis of complementary capabilities. CATL has stronger industrial resources, customer base and project reserves, while our advantages lie in power trading, AI forecasting and energy storage asset operation, which can help customers further tap project revenue.