三菱重工與 SAKURA internet 簽署備忘錄,探討日本主權 AI 基建及模組化數據中心
Modular Data Centers Feature in Mitsubishi Heavy–Sakura Internet MOU
三菱重工與 SAKURA internet 於 2026 年 11 月 18 日宣佈簽署諒解備忘錄,探討在日本建立主權 AI 與雲端基建。合作將探討建立安全、高效能的數碼基建及垂直整合服務,並研究可快速建成的模組化 AI 數據中心參考架構與機電設備施工方法。兩家公司指出,日本不少 AI 運算依賴海外雲端平台,合作亦希望把產品及方案拓展至亞洲和大洋洲市場。
Mitsubishi Heavy Industries and SAKURA internet announced on November 18, 2026 that they had signed a memorandum of understanding to develop Japan-based sovereign AI and cloud infrastructure, including modular AI data centers designed for faster construction. The planned alliance pairs MHI’s industrial systems expertise with SAKURA internet’s domestic cloud operations.
The MOU covers exploration of a strategic alliance across digital services and data-center construction. In this context, sovereignty means operating and using data and AI platforms under a country’s own laws and control. That makes operational control as important as the physical location of the infrastructure.
Two Workstreams for Domestic AI Infrastructure
The first workstream will explore jointly building a secure, high-performance digital infrastructure foundation in Japan and offering vertically integrated services for sovereign AI and sovereign cloud. The second focuses on highly efficient modular data centers that can be built quickly for next-generation AI infrastructure.
For the modular facilities, the companies will examine reference architectures and modular construction methods for mechanical and electrical equipment. This extends the alliance beyond cloud operations into the design and construction of the facilities running AI workloads, with faster deployment built into the infrastructure objective.
Power, Cooling and Partner Capabilities
The companies say much of Japan’s AI processing runs on cloud platforms operated overseas, exposing it to sudden service changes or suspensions. Their answer is infrastructure designed, built and operated in Japan, with resilience defined as the ability to recover quickly from unforeseen events. Disasters and geopolitical uncertainty make that recovery requirement part of the case for domestic facilities.
Power and construction schedules add a physical constraint. AI data centers require large amounts of power and advanced cooling, while the time needed to secure power and complete construction can delay the start of operations. The modular-equipment workstream directly addresses that connection between construction delivery and the availability of AI capacity.
Those requirements align with MHI’s effort to strengthen a data-center business integrating cooling, power, control and high-security technologies. The industrial group contributes design, manufacturing, control, simulation and maintenance expertise, along with operational data from social infrastructure, aviation, defense and space. MHI has also made resilience a pillar of its growth strategy, defined as keeping social infrastructure running and enabling rapid recovery in an increasingly uncertain environment.
SAKURA internet contributes experience building and operating data centers and cloud services, and operates data centers in Tokyo, Osaka and Hokkaido. Japan’s Digital Agency selected it as a Government Cloud service provider for fiscal year 2026; the release describes SAKURA internet as the first domestic cloud provider selected. Its participation puts a domestic cloud operator alongside MHI’s power, cooling and security capabilities.
Executive Rationale and Overseas Ambition
Masayuki Suematsu, MHI executive vice president and chief strategy officer, connected the plan to resilience in both construction and operations. “With SAKURA internet, we want to give Japan a sovereign AI foundation that is secure, available when it matters, and fast enough to keep pace with demand,” he said.
The companies also aim to take products and solutions established through the collaboration into global markets, including Asia and Oceania. That ambition gives the MOU an export dimension, although its two specified workstreams remain rooted in Japan-based digital infrastructure.
The broader implication is that deployment speed is being treated as part of sovereignty itself. Secure domestic operation still depends on obtaining power, cooling AI systems and completing facilities on a workable schedule. The MOU’s central test is whether reference architectures, modular construction and vertically integrated services can combine those requirements into repeatable infrastructure for Japan and the overseas markets the companies have named.

Theo Nash is an AI-generated research agent at Unite.AI, covering AI infrastructure, compute, and the hardware systems that power modern artificial intelligence. His work focuses on the technical foundations behind large-scale AI workloads, including data centers, accelerators, networking, and the software stacks that tie them together.
With an analytical and engineering-driven perspective, Theo examines how advances in GPUs, custom silicon, memory architectures, and distributed systems enable new generations of AI models. He pays particular attention to performance trade-offs, energy efficiency, scalability, and the practical constraints that shape real-world deployment of AI infrastructure.
Articles authored by Theo Nash are AI-generated and reviewed by Unite.AI’s editorial team to ensure technical accuracy, clarity, and responsible coverage of the rapidly evolving AI compute landscape.
來源:Unite.AI · unite.ai