Hitachi, the Japanese industrial conglomerate, is taking its advance into digital technology to the next stage with an “AI factory” based on Nvidia processors and a tie-up with OpenAI. The company is also expanding its manufacturing operations in the US.
On October 2, Hitachi wrote on X: “Hitachi CEO Toshiaki Tokunaga met @OpenAI CEO Sam Altman in Tokyo today. We signed a landmark MoU to collaborate on sustainable and plentiful energy to power the growth of AI in our society. This is a significant development, a big step forward for the future of #genAI @sama [the honorable Sam Altman].”
Under the agreement, Hitachi will supply OpenAI data centers with energy-efficient electric power transmission and distribution equipment, along with water cooling and air conditioning systems. OpenAI will provide its large language model (LLM) technology to Hitachi, which will incorporate it into its digital services portfolio.
High energy consumption is one of the most serious challenges confronting OpenAI, Microsoft and other builders of AI data centers.
According to a Bloomberg study, US wholesale electricity prices have more than doubled since 2020 in areas near data centers, and the situation will only get worse as electric power demand from data centers is expected to double to 9% of total US electricity demand by 2035.
That is a conservative forecast. The Lawrence Berkeley National Laboratory projects data center energy consumption will rise to 6.7% to 12% of total US electricity demand by 2028.
This is already having serious knock-on effects in the retail market. Energy analysts, research organizations and other sources report annual increases in residential electricity rates near major data center clusters ranging from 6% to 15% per year.
This is a very unwelcome development from the point of view of ordinary citizens and the politicians they elect to local government.
Pointing the way ahead, the state of Texas has passed a law allowing data centers to be cut off when blackouts caused by storms or other events put ordinary citizens at risk. This idea is gaining traction in other regions, such as the mid-Atlantic states, where data center construction is overwhelming the power grids.
It is anathema to data center operators, who require a steady supply of electricity, but good news for Hitachi and other providers of energy-saving equipment.
OpenAI’s Stargate project, a partnership with Softbank and Oracle, is planning to build multiple data centers in the US, Europe, the Middle East and Asia-Pacific. Initially, Stargate is targeting AI workload processing capacity of 7GW (gigawatts) in the US and 10GW worldwide, roughly the output of seven to 10 nuclear power plants.
On September 25, Hitachi announced the establishment of a Hitachi AI Factory based on its iQ data analytics platform and Nvidia’s AI Factory reference architecture.
It will feature Nvidia’s HGX B200 high-performance computing platform with Blackwell GPUs and Omniverse software libraries for simulation, industrial scale, and physically accurate digital twins. Hitachi iQ is built on Nvidia RTX PRO Servers.
According to Hitachi, “This infrastructure allows for the rapid development and deployment of advanced physical AI models. These models will acquire and interpret information from physical environments via cameras and sensors; determine the next steps to take; and execute actions based on that data.”
The goal is to create and deploy AI that enables Hitachi to expand its Lumada portfolio of digital solutions for the mobility, energy, industrial and technology sectors.
The AI Factory will be distributed internationally so Hitachi engineers can collaborate from wherever they happen to be located. Lumada is derived from “Illuminating Data.”
Hitachi management views the AI Factory as key to realizing Lumada 3.0, which aims to “evolve social infrastructure through AI strengthened with domain knowledge.” It follows Lumada 1.0, an IoT (Internet of Things) platform for customer operations, and Lumada 2.0, which provides digital solutions for entire value chains.
On September 10, Hitachi announced three new projects in the US in response to President Donald Trump’s tariffs and demands for investment in manufacturing, but also as part of a long-term strategy of expansion in what is now the company’s second-largest overseas market. The three projects are:
Electric Power Infrastructure: About $1 billion to be invested in what the company says will be the nation’s largest transformer manufacturing plant in Virginia, along with expansions of existing transformer production facilities in Virginia and Tennessee and increased production of high-voltage power grid components in Pennsylvania. These investments target rising power demand from data centers, renewable energy and manufacturing.
Railcar Factory: About $100 million will be invested in a new manufacturing plant in Maryland, which opened earlier this month. Including more than $30 million in “digital enhancements,” it can produce up to 20 railcars per month for customers, including the Washington, DC, Baltimore and Philadelphia metros.
The digital enhancements include Lumada smart manufacturing technology, a private 5G factory network developed with Ericsson, and logistics software developed with subsidiary GlobalLogic, which Hitachi acquired in 2021.
There is also a demonstration center to familiarize customers with Hitachi Rail’s HMAX digital asset management platform, which enables railway operators to predict potential problems and increase efficiency across train fleets, signaling systems and other infrastructure.
New Automation Headquarters: Hitachi’s US subsidiary JR Automation (acquired in 2019) plans to build a $72.8 million global headquarters in Michigan to develop and demonstrate new solutions, collaborate with customers and promote the use of digital twins, edge computing and other automation technologies marketed by Hitachi.
Hitachi Ltd CEO Toshiaki Tokunaga told Japan’s Jiji Press the company had invested more than $12 billion in the US in the past five years, and “We want to spend the same amount or more” by around 2030.
Follow this writer on X: @ScottFo83517667
