Nvidia and several major tech companies are often seen as representatives of the AI boom. In reality, the global data center construction boom has driven a surge in demand for various equipment such as transformers, advanced cooling systems, and suppliers have greatly benefited as a result.
According to a report by Reuters, consulting firm McKinsey estimates that by 2030, global AI data center investments will approach nearly $70 trillion. Nvidia also mentioned last week that AI expenditures are expected to remain robust in the coming years.
Many large AI data centers require substantial power support, making transformers, cooling systems, and other equipment increasingly crucial.
Transformer suppliers like South Korea’s Hyundai Electric and China’s Hainan Jinpan Smart Technology have reported significant growth in their performance for the first half of 2026, benefiting from AI infrastructure projects in North America, Europe, and the Middle East.
Hyundai Electric’s order backlog as of the end of June increased by 23% compared to six months prior, reaching $8.5 billion; while Jinpan saw a more than fourfold increase in new data center orders in the first half of the year compared to the same period last year, with related backlogs nearly tripling.
The construction speed of more AI data centers is falling behind expectations. Consultancy firm Pivotale AI points out that many large-scale operators aim for completion within six months, but grid connections in some emerging markets may take 24 months, while major developed markets may need to wait over 96 months.
CEO of digital infrastructure service provider BodaData, Wing Kin Cheung, stated, “People outside the industry talk about GPUs, but insiders are discussing delivery cycles of generators and transformers.”
As the power and water consumption of AI data centers continue to rise, the Bank of America cited Nvidia’s technical blueprint estimating that by the end of 2030, the power consumption of each AI cabinet may exceed 1.5 million watts (MW), nearly 100 times that of traditional cabinets.
One of the technologies currently attracting attention from equipment suppliers is the Solid-State Transformer (SST), which uses semiconductors to replace traditional bulky magnetic coils and copper windings to convert and transmit power; UBS estimates that SST can improve power efficiency by about 4% and reduce costs.
Although commercial applications of SST are still in the early stages, UBS forecasts that its penetration rate will rise to 40% by 2030. Hyundai Electric and Jinpan have both expressed plans to increase investment in SST research and development; while Delta Electronics stated that one small data center has already adopted its SST.
In addition to power consumption issues, the large amount of heat generated by AI data center operations has made cooling systems another growing area. Currently, the mainstream cooling methods consist of air cooling and liquid cooling.
Matty Zhao, Director of Infrastructure Materials and Oil & Gas Research at Bank of America Asia-Pacific, stated that the power consumption and cooling requirements of AI data centers are directly proportional, “The higher the power usage, the more heat is generated, and the larger the cooling system required.”
Bank of America estimates that by 2030, liquid cooling will account for 70% of new AI data center installations (currently around 30%), far exceeding the installation volume of air cooling. McKinsey believes that liquid cooling technology can reduce energy consumption by 27%.
Furthermore, some operators are exploring non-traditional cooling methods, hoping to build AI data centers underwater, underground, or utilizing floating facilities, creating new opportunities for a wider range of suppliers, driving rapid growth for companies like Delta Electronics and its peers Asia Vital Components, Auras Technology, and China’s Shenzhen Envicool Technology.
While the strong demand for cooling management has led to rapid development for many companies, the upward trend in supplier stock prices has begun to slow down, as investors are starting to question whether current valuations are too high amid increasing competition.
Delta Electronics’ stock price has risen by over 90% this year, Hyundai Electric has remained relatively stable, while Jinpan and Envicool have seen declines of nearly 30% and 20% respectively.
Chairman of Delta Electronics, Ping Cheng, stated in July of this year, “Even with revenue growth, the gross profit margin will probably remain at current levels. There are many variables in the market, including new product platforms, deployment delays, and component shortages, which may become more severe in the second half of the year.”
Matty Zhao cautioned that investors need to be discerning to identify the real “winners” who can truly secure customers.