Tatung (2371.TW), a century-old power and electromechanical giant, formally unveiled its new AI data center modular solution “TTG AIDC CUBE” in Taipei on the 30th, declaring a full transformation from a traditional power equipment manufacturer into an AI infrastructure integrator. Chairman Chang Jung-hua said Tatung is extending its century of accumulated power and electromechanical integration capabilities into the compute infrastructure sector. Through standardized and prefabricated architecture, the company aims to dramatically compress traditional data center construction timelines of 2–3 years down to completing compute deployment within 180 days.
At the launch event, Chang Jung-hua noted that Tatung has spent the past century delivering power to every corner of Taiwan, and now aims to go a step further by delivering compute to wherever industry needs it. He emphasized that AI development brooks no delay, and that traditional data center construction speeds can no longer respond to market demand. Tatung’s 180-day compute deployment target is precisely designed to shorten the timeline from infrastructure construction to compute going live through standardized and flexibly scalable design. He also mentioned that Tatung has emphasized returning to its core business in recent years, but “returning to core business does not mean returning to the past” — rather, it means transforming a century of power expertise to cross from the power era into the compute era, and bringing this rapidly deployable AI infrastructure solution to global markets.
Building-Block Architecture: On-Site Assembly in Four Weeks
President Chang Sung-pin elaborated on product details. He pointed out that traditional AI data centers often take 1–2 years from civil engineering planning to facility construction, whereas the core concept of “TTG AIDC CUBE” lies in modularity, standardization, and rapid deployment — pre-integrating key functions such as compute racks, power distribution, cooling, and intelligent monitoring like building blocks. Equipment is shipped in standard 20-foot and 40-foot high-cube containers, and on-site piping and rack assembly can be completed in under four weeks after arrival, with the goal of delivering a fully operational solution to the owner within 180 days.
In terms of performance metrics, the solution guarantees a Power Usage Effectiveness (PUE) below 1.3 and meets Tier 2 or higher reliability standards, featuring a fully redundant system architecture that allows equipment replacement without downtime. Chang Sung-pin said Tatung’s demonstration site has already installed two cooling systems, with medium-voltage power supply, transformers, and switchgear all being Tatung’s own products. The demonstration site covers both air-cooled and liquid-cooled racks: air cooling is supplied by external chillers feeding chilled water, paired with cold-aisle and hot-aisle designs within the compute racks; liquid cooling uses room-temperature water to directly cool compute chips, and has already factored in the liquid cooling requirements of NVIDIA’s next-generation Vera Rubin platform.
Targeting Mid-to-Large Scale Compute Demand, Courting ODMs
On market positioning, Chang Sung-pin noted that the solution primarily targets mid-to-large scale enterprise self-built compute and data center operators in the 0.5MW to 50MW range; for projects exceeding 50MW, traditional civil engineering approaches are still recommended. Furthermore, Tatung is directly reaching out to major Taiwanese server ODM manufacturers for collaboration, emphasizing that after ODMs complete L10 server assembly, they often face gaps in high-voltage power supply and liquid cooling technology during subsequent L13 rack-level integration. Tatung possesses complete capabilities in power and thermal management, and can help Taiwanese server makers accelerate their global expansion.
This strategy reflects the reshuffling of roles within Taiwan’s hardware supply chain amid the explosive growth in AI compute demand. Server ODMs have long focused on manufacturing and assembling compute units, but when customers demand full-rack delivery or even rapid deployment of entire data centers, the integration capabilities for power, cooling, monitoring, and other peripheral infrastructure become critical bottlenecks. Tatung is leveraging its century of power equipment manufacturing experience to fill this gap, seeking to occupy a position spanning the full value chain from the grid to the rack in the AI infrastructure wave.
Government Resources Ramp Up in Tandem
Taiwan’s Minister of Economic Affairs Kung Ming-hsin, who attended the same event, said the AI era is reshaping national competitiveness and energy dynamics, and that Tatung has broken the traditional manufacturing mold to demonstrate new industrial value. In response to surging compute demand, the government is advancing the “New Top Ten AI Infrastructure Projects,” and the president has announced an additional NT$100 billion in budget, of which approximately NT$25 billion (approximately $784.2 million) will be fully dedicated to helping small and medium-sized enterprises adopt AI compute transformation. On the power supply front, 25 large natural gas generation units are planned to come online over the next decade, releasing a total of 26GW of stable power. He emphasized that with Tatung’s infrastructure capabilities and government resources both in place, public-private collaboration will build a solid foundation for Taiwan’s AI compute development.
Share Price Reaction and Market Observations
News of Tatung’s AI infrastructure product launch boosted its share price, which opened higher at NT$29.4 on the 30th and climbed to an intraday high of NT$30.25, up more than 3%, reclaiming the NT$30 level. As of 10:30 a.m., the stock was trading at NT$29.9, up approximately 1.87%, with trading volume exceeding 8,000 lots. However, institutional investors remained cautious, with the three major institutional categories having been net sellers for four consecutive trading sessions, totaling 4,245 lots in net sales.
From an industry perspective, global AI data center construction demand continues to climb, and the challenges posed by high-density computing — power supply, cooling, construction speed, and energy efficiency — are becoming increasingly severe. Market intelligence indicates that the long cycle from planning to construction of traditional data centers has become one of the bottlenecks in AI application deployment, and modular and prefabricated architectures are gradually becoming the mainstream trend in the new wave of data center construction. Tatung’s 180-day compute deployment proposition directly responds to the market’s urgent demand for speed, though actual order intake and delivery capability remain to be validated.
Chang Jung-hua emphasized at the event that Tatung is moving from providing individual equipment to providing complete solutions, from power infrastructure to compute infrastructure, starting from Taiwan and expanding outward, with plans to export replicable, scalable, and rapidly deployable AI infrastructure to the world. This vision outlines Tatung’s long-term transformation blueprint, and whether “TTG AIDC CUBE” can gain traction in the fiercely competitive AI infrastructure market will be Tatung’s most critical test going forward.
