"Integrating computing and electricity" is becoming a new rule for AI infrastructure: SoftBank plans to build its own battery production capacity to complete the energy foundation closed loop.

Against the backdrop of a sharp increase in artificial intelligence computing power demand and an increasingly acute bottleneck in power supply, SoftBank Group has made a decision with significant significance as a guiding light. According to sources, SoftBank's mobile communications department plans to transform part of its factory in Sakai City, Osaka, into one of Japan's largest large-scale battery production lines, with the goal of starting production within the next five years to provide stable power for its own artificial intelligence data centers. SoftBank's CEO Junichi Miyakawa is expected to officially announce this battery business when unveiling the new five-year business plan next month. The group's founder, Masayoshi Son, has expressed approval of this decision. The decision is pending final approval from the telecommunications department's board of directors. The Sakai City factory transforming into an "AI computing + energy storage" hub marks a new frontier in AI competition. From "buying electricity" to "generating electricity," SoftBank is extending its AI layout to the energy sector, which is at the bottom of the industrial chain. This move also provides a new footnote to the main AI investments transitioning from chips and software to power infrastructure. According to the plan, the total battery capacity of the battery factory is expected to reach several gigawatt-hours, making it one of the largest battery production lines in Japan once completed. The factory is located within the Sakai City liquid crystal display factory purchased from Sharp by SoftBank in 2025 for 100 billion yen (approximately 6.3 billion US dollars). The factory site covers approximately 840,000 square meters and was originally planned to be transformed into an AI data center. The initial phase of the data center project is planned to have a capacity of 150 megawatts and is expected to start operating later this year, with the ability to expand to 400 megawatts or more in the future. The addition of battery manufacturing capabilities will transform the same factory site into a composite AI infrastructure hub of "computing power + energy storage." Sources indicate that SoftBank is evaluating various new battery technologies from Japan or South Korea, which are still in the early stages of development, intending to bypass the mature lithium iron phosphate (LFP) technology route currently dominated by Chinese companies to enhance supply chain self-sufficiency. Junichi Miyakawa had previously expressed his hope for Japan to improve its manufacturing self-sufficiency to counter the continuous disruptions to the supply chain caused by geopolitical tensions. If the project is successfully implemented, the battery products will be prioritized for the protection of SoftBank's own data centers before gradually being made available to other Japanese companies. SoftBank's establishment of the factory mirrors the global AI industry's power anxiety. According to a report from the International Energy Agency (IEA), global data center electricity consumption is expected to increase from 485 terawatt-hours in 2025 to 950 terawatt-hours in 2030, with the power demand driven by AI growing even more prominently. BlackRock estimates that by the end of this decade, the United States alone will need to add approximately 148 gigawatts of generating capacity to meet the needs of data centers. However, there is a serious supply lag- Morgan Stanley estimates that from 2025 to 2028, US data centers will face an approximately 55 gigawatt power deficit, and grid expansion generally requires 5 to 7 years, with grid connection approvals taking up to 18 to 30 months. With the decreasing costs of energy storage technology, global demand for battery storage is thriving, largely to help countries cope with the rapid growth of intermittent renewable energy sources. One of the groups driving the growth in demand for green energy is data center operators facing increasing electricity demands. Additionally, due to limitations in grid capacity, the growth of renewable energy has been restricted, and battery storage can help Japan release more wind and CECEP Solar Energy generation capacity. "When the grid cannot keep up, off-grid self-generated electricity becomes the only way out," analysts point out. Recently, Oracle announced the purchase of a 2.8 gigawatt fuel cell system, and tech giants like Meta are also focusing on nuclear power. SoftBank's entry into battery manufacturing is the first time such vertical integration has extended on a large scale to the energy core of storage battery cells. This marks a shift in the standard of AI infrastructure competition from theoretical computing power scale towards deliverable and scalable stable power conditions. "Computing + electricity synergy" is becoming the new generation of infrastructure guidelines. Closing the loop on AI investments For SoftBank itself, the battery factory further closes the loop on Masayoshi Son's grand narrative that "the end of AI is energy." Through chip design (such as Arm), data centers, cloud services, and massive investments in OpenAI, SoftBank has already built a fairly complete AI stack, and now complements it with the energy foundation. The telecom department's transformation strategy spearheaded by Junichi Miyakawa transitioning from a communication channel provider to an AI infrastructure enablerthus gains key support. This new move highlights Miyakawa's long-term strategy of transforming the telecom department into an operator of AI data centers. SoftBank is also working to upgrade its mobile network coverage in Japan to meet the demands of AI equipment, including warehouse robots like Siasun Robot & Automation. This move is in line with Masayoshi Son's ambitions, as he hopes to play a more significant role in the global AI competition. However, there are also sharp challenges in reality. The construction period for the project, spanning five years, will continue to consume capital, and SoftBank is currently in the late stages of negotiations to add up to $30 billion in investments to OpenAI, highlighting the tension in capital allocation. On the technological front, the maturity of the new battery industry in Japan and South Korea remains questionable, and if they are ultimately forced to return to the LFP route, their planned several gigawatt-hour production capacity may face cost and scale disadvantages compared to the over 50 gigawatt-hour volume of leading factories in China. The top seven global energy storage battery shipments in 2025 all came from Chinese companies, collectively holding an 83.3% market share, erecting high competitive barriers. From an industry investment perspective, MSCI has made it clear that the AI investment landscape is expanding from chips and data centers to power and grid operators. SoftBank's factory establishment news will reinforce this trend of investment diffusion, further solidifying power, energy storage, and grid equipment as core beneficiaries of AI. Stocks related to AI and power have shown strong performance. GE and Vernova have more than doubled in the past year, with a 51% increase since the beginning of the year. The S&P 500 utilities index rose by 7.5% in the first quarter, while the broader S&P 500 fell by 4.6% during the same period. SoftBank's factory establishment news provides a new catalyst for this rotation. Energy storage and battery manufacturers, gas turbine and power component suppliers, and utility companies that can accommodate the growing load of data centers may all benefit from the renewed emphasis on AI power demand narratives. After Bloom Energy signed a major fuel cell deal with Oracle, the after-hours stock price surged over 15%, demonstrating the market's high sensitivity to this theme. In its latest report, Morgan Stanley warns that the market may significantly underestimate the depth of the AI revolution, with computing power demand growth rates already three times that of supply growth. BlackRock believes that natural gas turbines, CECEP Solar Energy, fuel cells, and energy storage batteries are forming a diversified solution, and the power bottleneck may be systematically alleviated. In this grand narrative, companies that can deliver practical power solutions will receive sustained structural demand support.