TECH NEWS – More than $1 billion must be accounted for in annual energy consumption, which is no small sum.

The Vera Rubin era has arrived. The first systems are being shipped to major cloud service providers for testing and evaluation before large-scale deployment. Since full-scale production is underway, Nvidia hopes for greater success than with Blackwell. However, Foxconn Chairman Young Liu stated that costs will rise exponentially as gigawatt-scale data centers are built. Vera Rubin will be a significant breakthrough in agent-based AI, providing unparalleled computational performance beyond what Blackwell could have imagined.

Nvidia Vera Rubin-based AI data centers cost US$47 billion per 1GW (gigawatt) data center holding 3,557 server racks – $9.1M each – with an annual $1.3B electric bill and hardware depreciation 6x higher than power costs, said Young Liu, chairman of Foxconn, the world’s biggest AI…

– Dan Nystedt (@dnystedt) June 19, 2026

Foxconn estimates that the cost of AI data centers based on Nvidia’s Vera Rubin architecture could reach $47 billion per 1 GW deployment. Each data center will house up to 3,557 server racks, costing approximately $9.1 million each. According to Morgan Stanley Research’s latest bill of materials, or BOM, VR200 NVL72 servers cost around $8 million each. Construction costs are also very high. For example, Young Liu stated that building a Vera Rubin-based 1 GW AI data center, or AIDC, would require up to $47 billion in capital expenditures and approximately 3,557 racks. Furthermore, a single Vera Rubin rack costs $9.1 million, and the annual electricity cost for a 1 GW AIDC is $1.3 billion. The depreciation cost of the hardware is six times the annual electricity cost.

We are already seeing the construction of AIDCs with capacities of several gigawatts, or GW. The global data center market is expected to reach $1.6 trillion by 2030, when global computing power is expected to consume 174 GW of energy, more than double the 68 GW required in 2024. Therefore, between 2025 and 2030, 18 GW of new electricity generation capacity must be built annually to keep pace with the growing demand. The four main customer segments requiring this computing capacity are AI model developers, cloud service providers, or CSPs, governments, and large enterprises. Most of these customers are in the early stages of AI adoption, but their future goal is to create AI-based organizations where every process is driven by AI and people are needed only to set and manage objectives and oversee workflows and results.

To this end, Foxconn’s chairman has proposed establishing “Taiwan-style” science and technology parks in the United States, primarily in Arizona and Texas. Efforts are already underway to make this a reality by the end of the year. The Vera Rubin era ushers in a new era of AI by providing autonomous systems with unprecedented computing power, albeit at an extraordinary cost. As the world moves toward AI-based organizations and multi-gigawatt deployments, the coming years will reveal how quickly energy capacity, capital, and innovation can grow together. The winners will be those who master the technology and the economic considerations involved in building the AI infrastructure of the future.

Source: WCCFTech, CTEE, EC