Business Case for a Regional AI-Based Marketplace for Renewable Energies ^†

Abstract

The global energy sector is rapidly changing due to decentralization, renewable energy integration, and digitalization, challenging traditional energy business models. This paper explores a startup concept for an AI-assisted regional marketplace for renewable energy, specifically suited for small- and medium-sized enterprises (SMEs). Driven by advancements in artificial intelligence (AI), big data, and Internet of Things (IoT) technology, this marketplace enables efficient energy trading through real-time supply–demand matching with dynamic pricing. Decentralized energy systems, such as solar and wind power, offer benefits like enhanced energy security but also present challenges in balancing supply and demand due to volatility. This research develops and validates an AI-based pricing model to optimize regional energy consumption and incentivize efficient usage to support grid stability. Through a SWOT analysis, this study highlights the strengths, weaknesses, opportunities, and threats of such a platform. Findings indicate that, with scalability, the AI-driven marketplace could significantly support the energy transition by increasing renewable energy use and therefore reducing carbon emissions. This paper presents a viable, scalable solution for SMEs aiming to participate in a resilient, sustainable, and localized energy market.

1. Introduction

The global energy sector is going through a transformation that is driven by several interrelated factors, including the decentralization of energy production, the growing capacities of renewable energy sources, as well as the digitalization of energy systems. These trends are fundamentally changing the business models that have dominated the energy industry for decades, especially for small- and medium-sized enterprises (SMEs) in the sector. The evolving energy landscape is characterized by a shift towards sustainability, efficiency, and technological innovation, which is forcing energy companies to rethink their strategies and adapt to new challenges and opportunities.

1.1. Decentralization

Historically, energy production was centralized, with large power plants generating electricity for large geographical areas. These plants, typically fossil fuel plants, allowed for predictable, controlled energy production. However, this model is increasingly giving way to decentralized systems, where electricity is produced closer to the point of consumption, often by smaller, local producers. This shift is largely driven by the rapid growth of renewable energy technologies, such as solar and wind power, which allow private households and businesses to generate their own electricity [1].

The decentralization of energy production brings advantages, including reduced transmission losses, increased energy security, and greater local control over energy resources. However, this trend also introduces significant challenges. Renewable energy sources like solar and wind are inherently volatile, as they depend on weather conditions that can change without always being predictable. This volatility complicates the task of maintaining a stable balance between supply and demand, which has traditionally been managed through centralized power plants that could adjust their production and, by that, the output as needed [2].

1.2. The Role of Renewable Energy Within the Energy Transition

Renewable energy has emerged as a key component of the global effort to reduce carbon emissions and to combat climate change. Governments around the world have implemented ambitious targets to increase the share of renewable energy in their national energy mixes. For instance, the European Union has set a target to generate 32% of its energy from renewable sources by 2030, while individual countries like Germany aim to achieve even higher percentages [3]. Moving towards renewables is driven not only by environmental concerns but also by the desire to reduce dependency on fossil fuels, which are subject to price fluctuations and geopolitical tensions, like Germany’s reliance on Russian gas, which caused the expansion of renewables due to the invasion of Ukraine [4].

Despite those benefits, the integration of renewable energy into the grid poses significant challenges. Solar and wind power, which make up most of the renewable energy production, are intermittent sources of energy. Electricity is only produced when the sun is shining or the wind is blowing, leading to fluctuations in power supply that must be carefully managed to avoid destabilizing the grid [5]. This has led to the development of new technologies and business models, which are designed to integrate renewable energy into existing systems more effectively.

1.3. The Impact of Digitalization

Digitalization, which is a trend that has swept across various industries, is particularly impactful in the energy sector [6]. It facilitates the implementation of rapidly developing technologies such as artificial intelligence (AI), big data analytics, and the Internet of Things (IoT) into energy management systems [7]. These technologies enable more precise and accurate forecasting of energy production and consumption, optimize grid operations, and enhance customer engagement through personalized energy services. The usage of AI in particular has shown significant promise in improving the efficiency and reliability of energy systems. By analyzing large volumes of data from weather forecasts and energy markets as well as consumer behavior, AI can help predict fluctuations in energy supply and demand, allowing operators to adjust production and consumption in real time [8].

For smaller-sized energy companies, the adoption of digital technologies offers an opportunity to differentiate themselves from larger competitors. By using AI and big data analytics to optimize their operations, SMEs can reduce costs, improve their service quality, and offer innovative products such as dynamic pricing and personalized energy plans. However, the transition to a digital business model also requires significant investment in technology and expertise, which can be a barrier for smaller companies with limited resources, especially in terms of AI.

1.4. The Emergence of AI-Based Platforms in the Energy Market

In response to those challenges posed by decentralization and the integration of renewable energy, a new class of business models is emerging in the form of AI-based digital platforms. These platforms connect energy producers with consumers, which allows for more efficient trading of electricity based on real-time supply and demand data [9]. By leveraging AI, these platforms can dynamically adjust prices to reflect the current availability and demand of renewable energy, incentivizing consumers to use electricity when it is most abundant and reducing the strain on the grid during periods of low production [10]. One of the most promising applications of AI within the energy sector is the development of regional energy marketplaces, where local producers and consumers can trade electricity directly with each other. This model not only supports the decentralization of energy production but also encourages greater community involvement in energy management. By keeping energy production and consumption within a regional area and timewise closely connected, these platforms can reduce transmission losses and increase the use of renewable energy.

This paper focuses on the development of a startup concept for a regional AI-based marketplace for renewable energy sources. The marketplace is designed to connect local producers, so-called prosumers and prostorers of photovoltaic (PV) electricity, with consumers in the same region, using dynamic pricing models facilitated by AI to balance supply and demand.

1.5. Research Objectives

The specific objectives of this research include the following:

• The development of a digital platform that facilitates direct transactions between local energy producers and consumers.
• Implementing AI-based predictions as a foundation for dynamic pricing models to optimize energy use.

This paper contributes to the ongoing discourse on how SMEs can contribute to the energy transition and harness the opportunities presented by digitalization. The findings have broader implications for the design of energy markets that need to be resilient, sustainable, and inclusive, particularly in the context of regional energy systems.

2. Materials and Methods

2.1. Research Design

The methodology for developing the AI-assisted marketplace for regionally generated renewable sources of electricity (RES-E) is based on a structured approach that combines qualitative and quantitative methods. This research project follows a comprehensive design that integrates theoretical model development with practical implementation in a real-world setting by utilizing the business model builder while including literature research that was carried out. A fundamental hypothesis on which this paper is based states that the energy consumption will increase globally, ironically due to the high computational costs of AI, as well as due to the technological improvements in third world countries and decarbonization in first world countries. The main hypothesis states that a competitive business model can be created utilizing the business model builder.

2.2. The Business Model Builder Approach

Based on the findings of a systematic literature review and practical experience in researching and developing business models, Karlheinz Bozem and Anna Nagl developed the business model builder as shown in Figure 1 [11]. The development of a business model is made up of many interrelated subcomponents, which join up to several components, which again build modules. The business model builder describes the single subcomponents and the logic behind a business model. It supports the logical and efficient definition and description of the single subcomponents. The business model builder offers a structured, result-oriented method for developing a business. It contains different methods with the goal of being able to evaluate a business model before it is implemented. It utilizes methods such as design thinking workshops and SWOT analysis and includes a detailed qualitative description of the business.

The business model builder consists of three key modules:

Module 1: Business Idea Development

This module involves identifying and defining the core business idea, which in this case is the creation of a regional marketplace for renewable energy. The identification process includes trend analysis focusing on energy transition, regionality, and digitalization.

Module 2: Elaboration of Business Models

In this phase, the qualitative aspects of the business model are elaborated. This includes defining the value proposition, key activities, customer segments, and revenue streams. The model considers the dual role of participants as both consumers and producers (prosumers) of energy.

Module 3: Quantitative Business Case

The final module involves the quantitative assessment of the business model, including financial projections, risk analysis, and potential for scalability [11].

The core of the methodology revolves around the ‘business model builder’ framework, which has been selected due to its iterative and robust nature, allowing for continuous refinement and testing of the business model throughout the project. This approach is particularly well-suited for startups in changing and uncertain environments, such as the renewable energy market [12,13,14]. The ideas and concepts for the business case for the regional AI-based marketplace for renewable energies that emerged as part of the design thinking workshops with the project partners were developed, expanded, refined, and iteratively adapted. This paper focuses on individual elements of the business case.

2.3. Data Collection and Analysis

Data for this project were collected from multiple sources, including regional energy producers as well as prosumers. The data collection process was designed to capture the diverse aspects of the energy market, including consumption patterns as well as production capabilities.

The data were then analyzed using a combination of qualitative content analysis and quantitative statistical methods. Machine learning models were developed to forecast energy production and consumption, utilizing historical data combined with weather forecasts from regional photovoltaic (PV) systems and consumption patterns [15].

2.4. Implementation of AI and Dynamic Pricing Models

A significant aspect of the methodology was the integration of AI to predict energy production and consumption [16]. The AI models, primarily based on machine learning algorithms, were designed to forecast day-ahead energy production and consumption, allowing for the implementation of dynamic pricing models.

Two distinct AI models were developed. The first model predicts PV electricity generation based on weather forecasts and differentiates between different PV plants by including system specifications in the dataset. The second model forecasts the electricity demand of each prosumer, enabling the future platform to predict an offerable amount of electricity for each prosumer.

The dynamic pricing model then adjusts electricity prices based on real-time data from the AI models. This pricing strategy will incentivize users to consume electricity during periods of high production and low demand, thus optimizing the use of renewable energy and reducing the need for grid expansion.

2.5. Validation and Iteration

The initial versions of the business model and AI models were subjected to validation through pilot testing. Feedback from these tests was used to iteratively refine the models and the overall business strategy. This iterative process ensured that the final business model will be both economically viable and technically sound and will be in the future by constantly being able to adjust the business model.

2.6. SWOT Analysis

The SWOT analysis is characterized in a framework for identifying and analyzing the external factors (opportunities and threats) and internal factors (strengths and weaknesses) that have an impact on the business case for the regional AI-based marketplace. Fundamental for the informational basis of the SWOT analysis were the design thinking workshops with the project partners. This analysis provided insight into the internal and external factors that influence the success of the AI-based market.

The following steps were used for this SWOT analysis within the design thinking workshops with the project partners:

• Identification and analysis of opportunities and threats of the AI-based marketplace for project partners.
• Identification and analysis of the strengths and weaknesses of the project partner.
• Determination of the SWOT analysis.

2.7. Ethical Considerations

Throughout this research, ethical aspects were considered, particularly in the collection and use of data. All participants in workshops were informed of the purpose of this research.

3. Results and Discussion

3.1. Data Collection and Training of the AI Models

In the initial phase of the project, AI models were trained to predict photovoltaic (PV) electricity production and regional electricity consumption of each consumer within the project. These models showed a high degree of precision, with a prediction error of less than 5% in most cases. This precision is crucial as a foundational input variable for the dynamic pricing model, which relies on precise predictions to set prices that reflect the actual availability of renewable energy [17,18].

Data were provided within the research project ‘AI-REN marketplace’ within the EU Lighthouse project ‘AI Factory SME’ by the companies involved in the project, a dealership and a logistics and freight-forwarding company. First, the data of PV power generation and power consumption were collected and pre-processed into a useable format, summing the kW values within a 15 min timeframe. The AI models integrated into the platform for forecasting PV electricity generation and consumption are regression models, which, compared to deep learning models with neural networks and complex structures, have the decisive advantage that they have a significantly shorter run time both when training the models and when creating forecasts, which is much more user-friendly. In addition, AI models are easier to interpret, which means that results can be discussed in detail with project partners, and the quality of AI forecast models can be evaluated more easily.

The model to predict PV electricity generation is based on ‘random forest regression’, in which the AI is trained with multiple uncorrelated decision trees with random subsets of the training data in parallel and independently, giving each tree a different perspective on the training data. The AI model for forecasting electricity consumption is a time series model, in which an ‘LGBM regressor’ (light gradient boosting machine) is trained, in which individual weak models are trained one after the other using decision trees, each of which learns from the error value (the deviation of the estimate from the actual result) of the previous model and then is combined into a strong model.

Based on the importance assessments, it was determined that for the AI model, the generation of PV electricity, the size of the system (‘rating’ factor), and local/regional weather data such as the sun and humidity (‘sun’ and ‘humid’ factors) are the most important for influencing a precise forecast. For the model to forecast electricity consumption, it is noticeable that information from the timestamp, such as the hour or day of the week (‘hour’ and ‘weekday’ factors), is crucial, which led to the use of a time series model for electricity consumption. Suitable AIs were trained and selected; for PV production, weather data were added as machine learning features, and a tree-based algorithm was selected. For power consumption, a time series model was utilized.

Through the focused training of the AI models on the important influencing factors and regular consultations with the project partners, further optimization was possible. The following figures illustrate collected data on the power consumption and the PV power generation of three days of a consumer, each in blue, and the corresponding AI prediction in red:

-

An AI model for forecasting electricity consumption, which shows small deviations in the forecasts (see Figure 2);

-

An AI model to forecast PV electricity generation with a very small deviation from the live data (see Figure 3).

3.2. AI-Driven Dynamic Pricing Model

The dynamic pricing model developed as part of this project utilizes AI predictions to adjust electricity prices in real-time. During periods of high PV production, prices are lowered to encourage consumption, while prices are increased during periods of low production to discourage excessive use and maintain grid stability. This approach not only optimizes energy consumption but also provides financial incentives for consumers to adjust their usage patterns, leading to a more balanced and sustainable energy system. Below, significant influence variables for a dynamic pricing model are defined.

• Electricity supply and demand

• PV electricity generation: The amount of PV electricity generated varies depending on the time of day, weather conditions, and seasonal factors. The tariffs should take into account the availability of PV electricity (live and, if applicable, from storage) and be dynamically adjusted.
• Demand profile: Electricity demand varies throughout the day and can be optimized through the use of load management. Peak load times and phases of low demand should be included in the tariff design.

• Market prices and grid status

• Spot market prices: Current electricity prices on the spot market can serve as a reference for pricing. Dynamic tariffs should react to short-term market price fluctuations.
• Grid utilization: Local/regional grid conditions, including grid congestion or grid bottlenecks, should be taken into account to ensure grid stability. Variable tariffs can provide incentives to shift electricity consumption during periods of low grid utilization.

• Localization and conditions

• Local/regional characteristics: Regional differences in electricity generation and demand should be taken into account. For example, rural areas with high PV density may require different tariffs than urban areas.
• Weather conditions: Weather forecasts can influence the expected PV electricity generation and therefore the tariff design. A sunny weather forecast can mean lower tariffs for the coming hours.

• Customer segments and usage patterns

• User behavior: Different customer profiles (e.g., households, commercial industrial companies) have different usage patterns. Tariffs should offer flexible options that are adapted to the specific needs of customers.
• Electric vehicles: Prosumers with their own charging stations for electric vehicles should have special tariff options to optimize charging and create incentives for charging during periods of high PV generation.

• Incentive mechanisms and reward systems

• Peak shaving: Incentives to reduce electricity use during peak times and increase use during times of excess PV electricity.
• Net metering and feed-in tariffs: Consideration of feed-in tariffs for electricity fed into the grid and the offsetting of self-generated and consumed electricity.

• Technological integration and data management

• Smart metering: Use of smart meters to monitor electricity consumption in real time and apply dynamic tariffs.
• Data analysis and forecasting: Use of data analysis and AI-supported forecasts to predict generation and consumption and dynamically adjust tariffs.
• Balancing capability.

• Regulatory framework conditions

• Legal requirements: Compliance with all relevant legal as well as regulatory requirements/guidelines; this includes data protection regulations and grid charges.
• Subsidy programs: Use of government support programs and subsidies for renewable energy.

• Example of a dynamic tariff for PV electricity and charging stations

• Basic fee: A monthly basic fee for the use of the platform and services.
• Variable consumption price: A variable price per kWh based on real-time market prices and current PV power generation.
• High PV generation: Lower tariffs during periods of high PV electricity generation (e.g., during the day when the sun is shining).
• Low PV generation: Higher tariffs during periods of low PV electricity generation (e.g., at night).
• Grid load-dependent prices: Additional adjustments based on current grid utilization to promote grid stability.
• Reward systems: Discounts or credits for charging electric vehicles during periods of high PV power generation or low grid load.
• Feed-in tariff: Remuneration for surplus electricity fed into the grid at variable market prices.

3.3. Environmental Impact

The environmental impact of the AI-based marketplace is another key area of investigation. By increasing the usage of renewable energy and reducing reliance on the grid during peak times, the platform would contribute to lowering overall carbon emissions [19]. The dynamic pricing model would also optimize the usage of renewable energy as customers are more incentivized to consume electricity when a lot of it is available as prices are cheaper in those times. Furthermore, the platform would also be an incentive for some customers to buy and use electricity storage to lessen the need for buying electricity in expensive timeframes, i.e., during dark doldrums [20]. Most of the renewable energy sources cannot be produced on demand, which is why energy stores are crucial for transitioning towards less fossil energy sources.

3.4. Economic Viability

The economic viability of the renewables marketplace is strongly linked to its ability to scale. However, achieving a high scale requires targeted marketing efforts and possibly incentives to attract more users on both sides to the platform [21]. This strategy should be individualized, depending on the regional differences of each customer, e.g., whether they live in rural or urban areas, as rural people have a higher likelihood of having an energy storage on average. The network effect, where the value of the platform increases as more users join, is therefore critical for its long-term success. As more users participate, the platform becomes more reliable, and the pricing model can operate more effectively [22]. Figure 4 and Figure 5 illustrate the most important tasks differentiated between the different channels and the value creation network.

3.5. SWOT Analysis

A SWOT analysis was conducted to evaluate the strengths, weaknesses, opportunities, and threats associated with the AI-REN marketplace. The key findings are summarized below:

• Strengths

• Innovative technology: Use of advanced AI algorithms to forecast electricity consumption and generation [23].
• Dynamic pricing: Flexible pricing model as a control tool that adapts to supply and demand in real time.
• Sustainability: Promoting the use of renewable energies and supporting the energy transition.

• Weaknesses

• Dependence on data quality: The accuracy of the forecasts depends on the availability and quality of the data.
• Regulatory hurdles: Compliance with regulatory requirements in the energy market can be complex, as well as time-consuming and costly.
• Technological integration: Integrating the platform into existing energy systems can pose challenges.

• Opportunities

• Growing market: Increasing demand for sustainable energy solutions and smart grid technologies.
• Political support: Incentives and political support for renewable energy.
• Potential integration of blockchain in order to enhance security via peer-to-peer trading.
• Creating public–private partnerships to shorten the distance to policymakers.
• Technological advances: Further development of digitalization and AI technologies and integration of new renewable energy sources [12].

• Threats

• Market competition: Competition from other providers of energy forecasts and digital marketplaces.
• Data protection: Strict data protection regulations and potential security threats.
• AI: risk of system failures and potential biases, e.g., regional biases, weakening the prediction quality.
• Market regulation: Changes in market regulation, e.g., with regard to the balancing of electricity volumes, can influence business conditions.

3.6. Future Research and Development

The results from this pilot project provide a strong foundation for further development and expansion of an AI-REN marketplace. Future research could focus on refining the AI models to further improve prediction accuracy, exploring additional features to enhance user engagement, and developing strategies to scale the platform across different regions. Additionally, the integration of automated load-shifting technologies could be a key area of focus to address the challenges identified in the pilot. Furthermore, integrating other renewable energy sources such as wind power into the platform could lead to more grid stability. The emerging role of formic acid functioning as a hydrogen carrier could be utilized as another form of transportable and tradable energy in the future.

4. Conclusions

The AI-assisted marketplace for regionally generated renewable electricity represents a significant innovation within the energy sector, particularly for small- and medium-sized enterprises (SMEs) looking to navigate the challenges of the energy transition [24]. The dynamic pricing model, which adjusts energy prices in real-time based on AI predictions of supply and demand, can incentivize consumers to adjust their energy consumption patterns according to price signals. By those price signals, the platform not only increases grid stability but would also contribute to a reduction in carbon emissions. The option for user feedback highlighted the importance of a seamless integration with certain smart home technologies, which can automate energy management and further enhance the user experience. This integration is important for the realization and maximizing the benefits of dynamic pricing and ensuring that the platform is accessible and valuable to a diverse range of users. The SWOT analysis conducted as part of this study identified several strengths, including the platform’s transparency, user engagement features, and scalability. However, challenges remain, particularly in achieving widespread adoption and integrating advanced smart home systems. The analysis also highlighted significant opportunities for expanding the platform to other regions and enhancing its functionality through partnerships, i.e., with technology providers. Looking forward, the continued development and refinement of the AI models, along with strategic efforts to scale the platform, will be essential for realizing the full potential of an AI-REN marketplace. The platform’s ability to grow and adapt to the changing energy landscape will determine its long-term success and its contribution to the broader goals of the energy transition towards more and more renewables [25]. In conclusion, the AI-REN marketplace offers a promising solution for regional energy markets, allowing SMEs to leverage digital technologies in order to stay competitive and also contribute towards a more sustainable energy future while still including business incentives. As the platform evolves, it has the potential to become a key player within the decentralized energy landscape, driving innovation and sustainability in the energy sector.