LogoLOGO

TOPICS & NEWS

Obayashi Corporation Announces Entry into Urban Data Center Development Business with New Subsidiary

The data center market continues to experience rapid growth due to advancements in artificial intelligence (AI) technology and the expanding use of cloud services. This article will highlight a noteworthy announcement from major Japanese construction company Obayashi Corporation regarding its domestic data center business.

 

Announcement of Entry into Urban Data Center Development Business in Central Tokyo

 

Obayashi Corporation has officially announced its full-scale entry into the urban data center development business. At the end of November, it established a new subsidiary, MiTASUN Corporation, to undertake the development and operation of urban data centers in central Tokyo.

 

Leveraging its expertise cultivated through previous construction and development projects, Obayashi will act as the project owner, renovating and rebuilding office buildings in urban areas into small-scale data centers. It plans to establish multiple small-scale data centers within an 8km radius and connect them with dedicated optical fiber lines to create a data center cluster. The aim is to secure a scale equivalent to a large-scale (hyperscale) data center with the combined capacity of this cluster.

 

Project Aiming for a Sustainable Society

 

MiTASUN has announced plans to invest approximately 100 billion yen in data center development in central Tokyo within the next 10 years. As its first project in fiscal year 2028, it will construct a new small-scale data center in Mita, Tokyo, with a receiving capacity of approximately 5.5MW, a common metric for data center size. Furthermore, it has already secured land for its second project in Tokyo, approximately 6km away from the first, with the goal of establishing a 40MW-class data center cluster by fiscal year 2031.

 

The company also plans to actively pursue alliances with other companies and implement the following measures:

 

  • Converting buildings owned by other companies into data centers
  • Interconnecting with other data center operators
  • Continuously expanding facility scale
     

The Obayashi Group aims to contribute to the advancement of urban infrastructure and realize a sustainable society through this new business venture.

 

Expectations for Urban Data Center Business Development

 

Demand for data centers is expected to continue growing, especially in central urban areas. Will Obayashi Corporation’s urban data center business become a new model case for urban development? There are high expectations for its future development, leveraging the technical capabilities cultivated in its construction business and the know-how gained from its development projects.

TOPICS & NEWS

2024.12.19

AI’s Growing Power Consumption Pushes the US to Promote Nuclear Energy

We’ve previously discussed how the rapid proliferation of artificial intelligence (AI) has significantly increased the global demand for data centers.

 

In recent years, both semiconductor factories and new data centers have been actively constructed domestically [in Japan]. However, the government aims to achieve net-zero greenhouse gas emissions, including CO2, by 2050.

 

As companies grapple with the challenge of securing the massive amounts of electricity consumed by data centers, they are devising innovative ways to meet demand while reducing carbon dioxide emissions through the use of renewable energy.

 

In this context, I’d like to introduce an announcement from Google related to this power challenge.

 

Google Announces Purchase of Power from Small Modular Reactors

 

Google has announced plans to purchase electricity from small modular reactors (SMRs) developed and installed by Kairos Power, a US next-generation nuclear power company.

 

The company aims to have the first SMR operational by 2030 and deploy additional reactors by 2035, with a total of seven SMRs providing 500 megawatts of power.

 

SMRs can be assembled in factories and then transported to the installation site as a complete unit (module), reducing construction time and costs compared to conventional nuclear power plants. Their smaller size also allows for more flexible site selection and easier cooling, making them safer.

 

Due to these advantages, SMRs are being developed worldwide, including in the United States, as a next-generation decarbonization power source. However, one of the challenges in implementing SMRs is finding buyers for relatively high-cost electricity until SMRs achieve economies of scale.

 

For Google, this move aligns with its goal of addressing the power needs of its AI data centers while advancing its decarbonization objectives.

 

The government is also supporting private initiatives. The US government has set a goal of tripling nuclear capacity by 2050 and is working to promote SMRs through initiatives such as supporting demonstration projects and building a fuel supply chain.

 

The US Department of Energy (DOE) has announced up to $900 million in support for next-generation SMR development, and public-private efforts to promote SMR development and construction are intensifying.

 

Power Challenges and the Need for Global Action

 

In the United States, IT giants are investing in nuclear energy one after another. In contrast, the trend toward phasing out nuclear power is accelerating in Europe, particularly in Germany and Italy. Japan is also not as proactive in nuclear power.

 

As the demand for AI continues to grow, countries around the world will be forced to address power challenges.

TOPICS & NEWS

2024.11.27

DOJ Launches Antitrust Investigation into NVIDIA’s Dominance in AI Chip Market.

NVIDIA Under Antitrust Scrutiny: Implications for the AI Industry

 

The U.S. Department of Justice (DOJ) has launched a formal investigation into NVIDIA, the world’s leading semiconductor company, raising concerns about potential antitrust violations. The DOJ has issued subpoenas, signaling a potential antitrust lawsuit against the company.

NVIDIA holds a dominant market share in AI semiconductors, with approximately 80% of the data center market in 2023, according to Omdia. The DOJ’s investigation focuses on whether NVIDIA has engaged in anticompetitive practices that hinder other companies from entering the market.

Beyond hardware, NVIDIA provides software optimized for AI development. The DOJ is examining whether the company has bundled hardware and software, offering preferential treatment to customers who purchase both.

CUDA: The Industry Standard

NVIDIA’s CUDA, a software development environment tailored for its GPUs, has become the industry standard for large-scale computing tasks, including training generative AI models. This dominance is due to NVIDIA’s extensive promotion and the close integration of CUDA with its hardware.

If found guilty of antitrust violations, NVIDIA could face significant penalties and be required to change its business practices. However, the widespread adoption of CUDA presents a challenge, as it would be difficult to quickly replace with alternative solutions.

Implications for the AI Industry

The DOJ’s investigation into NVIDIA highlights the growing scrutiny of Big Tech companies. Recent antitrust cases against Google and Apple set a precedent for increased regulatory oversight.

The outcome of this investigation could have far-reaching implications for the AI industry. If NVIDIA is found to have engaged in anticompetitive practices, it could lead to increased competition in the AI semiconductor market, potentially benefiting AI developers and end users. However, it could also disrupt the current AI ecosystem and lead to uncertainty in the market.

Conclusion

The DOJ’s investigation into NVIDIA marks a significant development in the AI industry. As the investigation unfolds, it will be essential to monitor the potential impact on AI innovation, competition, and the broader technology landscape.

TOPICS & NEWS

2024.11.19

Sakura Internet Takes an Offensive Stance, Sets Sights on Overseas Markets

In this article, we delve into the recent developments of Sakura Internet, a domestic company that has been at the forefront of GPU cloud services in Japan.

 

Founded in 1996 by current president Kunihiro Tanaka while he was still a student, Sakura Internet has grown significantly since its inception. In 2011, the company constructed the Ishikari Data Center, one of Japan’s largest suburban data centers, and strengthened its physical infrastructure and cloud services. From the dawn of the internet, Sakura Internet has adapted its services to meet the evolving needs of its customers. 

Competition with Foreign Services and Market Changes

 

As discussed in a previous article, Japanese companies have become increasingly reliant on foreign services for AI development. Sakura Internet has been no exception, facing intense competition from foreign IT giants over the past five years.

 

The impact of foreign services is substantial on a national level. Japan’s “digital deficit,” the outflow of funds due to payments for overseas digital services, has more than doubled in the past decade, reaching 5.5 trillion yen. With the limited presence of domestic services, the situation has led to concerns that “the more advanced cloud services become, the more Japan’s trade deficit will increase.” The recent depreciation of the yen has further exacerbated this issue.

 

However, the market environment surrounding Sakura Internet is undergoing significant changes. With the digital transformation (DX) initiative gaining momentum, all companies are aspiring to become IT companies. The COVID-19 pandemic accelerated the digitalization of society, leading to a substantial expansion of the market.

 

Moreover, the rising costs of foreign cloud services due to the weakening yen and the growing expectations for the government cloud market have increased demand for domestic public clouds like Sakura Internet.

 

Against this backdrop, Sakura Internet was selected by the Digital Agency as the sole domestic provider qualified to offer government cloud services, albeit with a condition to meet certain functional requirements by the end of 2025. Prior to this certification in November 2023, only foreign providers such as AWS, Microsoft, Google, and Oracle were eligible. Sakura Internet’s certification marked a significant milestone as the first domestic company to achieve this recognition.

Expansion into Asia through Partnership with Equinix

 

Sakura Internet is further expanding its business operations. On October 1st, the company announced a partnership with Equinix, a leading data center operator, aiming to enter the Asian cloud services market. Sakura Internet plans to lease space in Equinix’s data centers and install its own servers, enabling it to provide cloud services overseas. The company also intends to strengthen its cloud services in Japan using a similar approach.

Equinix operates data centers in countries such as Singapore and Indonesia. While the specific countries and timelines for Sakura Internet’s services have yet to be finalized, the company is moving forward with detailed plans.

A Growing Sense of Expectation

 

With its aggressive stance and focus on expanding into the Asian cloud market, where US IT giants hold a significant share, Sakura Internet is generating increasing anticipation. We will continue to follow the company’s progress and provide updates on future developments.

TOPICS & NEWS

2024.10.28

Highreso: A Promising New Player in Japan’s AI Data Center Development

We’ve previously reported on Sakura Internet’s active involvement in the AI data center business within Japan. Now, another domestic company has made a significant move in this space.

 

Japan Policy Investment Bank and Mizuho Bank Invest in Highreso

 

The Japan Policy Investment Bank and Mizuho Bank, among others, have announced a combined investment of 10 billion yen in the startup Highreso. Japanese companies have been insufficiently providing the high-performance computing infrastructure necessary for AI development, leading to a heavy reliance on overseas services. This investment, which is unprecedented in scale for a startup, aims to bolster domestic AI development.

Highreso operates a GPU-dedicated data center and provides a cloud service called “GPUSOROBAN.” The company has established Highreso Kagawa Co., Ltd. in February 2024 to open a generative AI data center in Kagawa Prefecture.

Highreso Kagawa is a special purpose company established to open a generative AI data center in Kagawa Prefecture. The company plans to develop the data center by partially renovating existing research facilities like “RIST Kagawa” and the gymnasium of the former Ayakami Middle School.

Amid concerns about the excessive concentration of data centers in metropolitan areas, this project aims to reduce development costs by utilizing existing regional facilities and provide services to a wide range of domestic businesses.

The project has been certified under the Ministry of Economy, Trade and Industry’s cloud program and is expected to utilize approximately 7.7 billion yen in subsidies from NEDO (New Energy and Industrial Technology Development Organization).

Highreso CEO Yoshiyuki Shikura expressed his enthusiasm, stating, “We will build a solid domestic infrastructure from the regions and revolutionize the AI industry.”

Moving Away from Overseas Dependence

Until now, Sakura Internet has been the most prominent domestic company actively involved in the GPU cloud service business. The announcement of this new domestic company’s plans has garnered significant attention. It remains to be seen whether Highreso’s project will be a game-changer in reducing dependence on overseas services. We will continue to follow the progress of this project and also provide updates on Sakura Internet’s efforts.

TOPICS & NEWS

2024.10.15

NVIDIA Earnings Announcement and the Trend of Domestic Semiconductor Industry

At the end of August, the world’s attention was drawn to the earnings announcement of US semiconductor giant NVIDIA. Around the same time, there were new announcements regarding the domestic semiconductor industry in Japan. In this article, we will introduce these two topics.

 

1. NVIDIA’s Q2 Earnings Announcement

 

NVIDIA, a global leader in AI semiconductors, announced its earnings for the second quarter (April-June) on August 28th, reporting record-high revenue and net income.

 

Revenue surged 2.2 times year-on-year to $30.04 billion. In particular, the data center segment, which houses most of its AI products, generated $26.272 billion, accounting for 87% of the total.

 

The importance of data centers has increased significantly with the advent of generative AI. It is said that generative AI requires 10 times more power than a typical search to answer user queries. Additionally, data centers consume a vast amount of electricity by performing calculations on massive amounts of data using numerous servers in a short period.

 

NVIDIA plans to start mass production of its AI-dedicated GPU, “Blackwell,” later this year.

 

Although production delays have been reported, it is said that generative AI will become 30 times faster and 25 times more energy-efficient.

 

Once the shipment of Blackwell begins in earnest, interest in data centers is likely to further increase.

  

2. Expectations for Preferred Networks (PFN), Developing Domestic Semiconductors (for Generative AI)

  

SBI Holdings, a major internet financial company, announced a capital and business alliance with Preferred Networks (PFN), a startup that provides AI services.

 

The two companies will jointly develop AI semiconductors by incorporating PFN’s AI semiconductor design know-how.

 

Founded in 2014, PFN is a startup that vertically integrates the development and provision of hardware and software necessary for the practical application of AI technology. The company handles everything from the design of AI semiconductors and development of peripheral software to the development of supercomputers using its own AI semiconductors, construction of generative AI foundation models, and development of applications that utilize them. Since 2023, it has been providing computational platforms using its own AI semiconductors to external users.

 

Recently, the global demand for AI semiconductors has been rapidly increasing due to the advancement of generative AI technology, leading to a supply shortage. Furthermore, it is expected that power consumption from AI development and utilization will continue to increase, necessitating the development of high-performance and low-power consumption AI semiconductors.

 

Through this collaboration with PFN, the SBI Group expresses its enthusiasm in strongly supporting the domestic dissemination of next-generation AI semiconductors and promoting PFN’s business expansion, thereby contributing to the development of the semiconductor industry in Japan.

 

Additionally, by manufacturing PFN’s next-generation AI semiconductors, the SBI Group expects to increase its recognition in the manufacturing industry and create a semiconductor ecosystem in Miyagi Prefecture, where its planned business site is located, leading to high economic effects within the region. This, the company believes, will enable a significant increase in the added value of the SBI Group and establish a role model for industrial creation in local areas starting with the semiconductor industry.

 

We will continue to follow the developments of how the collaboration between the SBI Group and PFN will contribute to the revitalization of Japanese industries and the semiconductor industry.

 

TOPICS & NEWS

2024.09.25

1 2 3 8