A：Guangdong Harbour Smart Calculation Company, formerly a real estate company, acquired visa data focusing on AIDC business in October 2025, thus officially entering the AI circuit and completing the transformation and upgrading, becoming a AI infrastructure operator. The company changed its name to Moonlight Smart Calculation in May of the same year to reflect its development direction. The company's core business focuses on the construction and operation of high-performance servers, computing clusters and AIDC, and is committed to providing solutions for large models, biopharmaceuticals, autonomous driving and other fields, and actively expanding the global AIDC business, with the goal of becoming the world's leading operator of basic services for intellectual computing capabilities.

A：In 2025, revenue will reach 1 billion, of which 0.62 billion will come from the AI computing business, accounting for more than 60% of revenue. Infrastructure business (including IDC business) is 0.38 billion, accounting for 38%. In addition, the company's gross profit was 0.14 billion billion, a significant increase over 2024; net profit was 73 million billion, turning losses into profits; and cash at the end of the year exceeded 0.17 billion, which had increased to more than 1 billion by 2026. The company successfully obtained the investment of more than 0.9 billion RMB from Zhayin Fund and Futian State-owned Assets Investment, and was included in MSCI China Small Stock Index from 2025 to 2026. The company obtained credit of more than 30 billion RMB. By 2026, the financing cost was further reduced to 1.9, the net assets increased significantly, and the interest-bearing debt ratio also decreased significantly.

A：The size of the AI industry is expected to grow rapidly in the next few years, and is expected to grow to more than 10 trillion in 2030. The company closely follows the development trend of the industry, currently in the management of computing power has reached 42000, in the domestic ranking is relatively high, and through the construction planning of AIDC, the provision of server network tuning, computing power leasing and investment incubation related enterprises and other ways to serve the large model training, AI pharmaceutical and other fields, to promote the growth of computing power demand. Tiandun began to develop IDC business in 2014 and initially adopted a light asset model to operate data centers. With the explosive growth of AI computing power demand, Tiandun began to turn to AIDC in 2024 and established the first million card cluster. By the end of 2025, the company's management capacity reached 42000, the data center business has grown steadily.

A：The company provides services from the AIDC level, including construction planning AIDC, server network tuning, delivery of computing power, and computing power leasing. At the same time, the company has also set foot in AIGC, AI pharmaceutical and other industries through investment incubation enterprises, and applied computing power to different industries, forming a diversified business model. As a leader in the industry, Guangdong Harbour Smart has a computing power of 42000, ranking high in the industry, with more than 15 billion stable billing orders and more than 200 customers. The company has obtained a number of technical advantage certification and professional control, the management team has deep experience, the consulting team has excellent academic, published more than 400 papers, more than 100 invention patents, strong support for the company's business development.

A：Guangdong Harbor Smart uses the excess computing power of the data center to provide it to SMEs and retail customers in a scattered manner through the computing power leasing platform for large model training, reasoning or research and development purposes, which is an important source of revenue for the company. In addition, the company has efficient delivery capabilities and strong computing power scheduling capabilities, as well as top AIDC construction execution, to provide excellent services to Internet manufacturers and other customers.

A：In terms of performance, we conservatively expect that 4 billion will be needed this year (2026). We have no expectations for 2027 and 2028 for the time being, because the current demand in the AI computing industry is unlimited, and we will adjust production according to the order volume received.

A：At present, the demand for AI computing power is in the explosive growth stage, we have enough delivery experience and maintain a competitive advantage. Even if there is a peer transformation to enter, due to the strong demand for orders, they can take as many orders as they can, which has little impact on us.

A：At present, 70% of our revenue comes from IoT head companies. We hope to reduce this proportion in the future and increase the proportion of small and medium-sized enterprises and retail customers.

A：Small and medium-sized enterprises are expected to participate because their individual profits are higher and can improve the overall profit margin. Regarding the depreciation policy, we adopt five-year depreciation. The server still has a high residual value after five years.

A：Our procurement team is experienced and pays attention to the market and locks the goods in advance, such as stocking up memory modules and other components to use when purchasing machines. Before signing the contract, the supply will be locked and the price will be negotiated to ensure a stable profit margin.

A：We have both training and reasoning needs, with about half and half each. As for cards, we also have domestic cards deployed.

A：The company has previously successfully converted US dollar bonds into equity, and the US dollar bonds have been basically transferred out and will be released to investors one after another this year, when the shares will be converted according to the conversion price (5.5 yuan/share).

A：We don't have much different planning at present, mainly to ensure the advantages of the supply chain, such as locking the supply in advance to prepare for the order.

A：We currently have a self-owned data center in Shenzhen and are working with partners to build agent data centers elsewhere. We provide servers, main network tuning and computing power output, and select the appropriate computer room location to place the server according to customer needs.

A：At present, training and reasoning business orders are equally split, but we expect that the demand for reasoning business will be even greater in the future.

A：We only provide computing power to customers. How customers use computing power is their own business. We are not involved in storing their data.

A：There will be more and more reasoning orders of magnitude. With the rapid development of the industry, we hope to provide more products, but the specific order of magnitude is difficult to judge, because the industry is developing very fast.

A：Competitive advantages include financial strength, delivery experience and technical stability. In terms of capital, we rely on the platform of listed companies to obtain bank credit support. Technically, we have self-researched things with low failure rate and have successfully delivered large-scale clusters. The relative stability of the supply chain is also one of the advantages. The disadvantage may be that the popularity is not high enough compared to A- share companies. Barriers to entry are mainly related to capital requirements and technical stability. We have certain advantages in capital and technology, the supply chain is relatively stable, but the popularity is lower than some A- share companies, which may be the current disadvantage.

A：It is inconvenient to disclose which major Internet companies are specific, but there are indeed several major head companies using our services. At the same time, we also have overseas layout and received orders from different enterprises.

A：We obtain funds through bank credit and listed company financing platforms, especially after purchasing servers, we will solve our capital needs through financial leasing or bank loans. At present, we have good cooperative relations with many banks and have large credit lines.

A：We buy servers and assemble them into clusters to output computing power to customers, instead of directly buying GPU or CPU and other components to provide to customers.

A：There are many brands of server suppliers. We purchase servers of different brands for networking and tuning to stably output computing power to customers.