Foxconn Chairman: Each 1GW Vera Rubin data center has a capital expenditure as high as $47 billion, with an annual electricity cost of $1.3 billion.

```

The construction costs of AI computing infrastructure are drawing intense market attention.

On June 19, according to Taiwan's "Commercial Times," Foxconn Chairman Liu Yangwei recently publicly disclosed a set of industry-shaking data: Building an AI data center (AIDC) with Vera Rubin at its core and a scale of 1GW requires capital expenditure as high as $47 billion, with annual electricity costs alone reaching $1.3 billion, while hardware depreciation is six times the electricity costs. These figures vividly reveal the astonishing capital threshold behind the AI infrastructure arms race.

Liu Yangwei stated, the world is rapidly entering the era of the "AI factory." It is expected that by 2030, total investment in global data centers will reach $1.6 trillion. Computing load will surge from about 68GW in 2024 to 174GW, meaning that from 2025 to 2030, an additional 106GW of electricity will be needed, averaging nearly 18GW of new supply annually.

At the enterprise application level, Liu Yangwei pointed out that most companies are currently in the AI empowerment phase, but the future goal should be transforming into "AI native" organizations, and he proposed seven major transformation characteristics.

1GW Data Center: $47 Billion Capital Threshold

Liu Yangwei broke down the construction cost structure of AI data centers with specific figures, providing the market with a valuable benchmark.

According to the data he cited, building a 1GW AIDC with Vera Rubin at its core requires about 3,557 cabinets, and a single Vera Rubin cabinet sells for $9.1 million, resulting in a total capital expenditure as high as $47 billion.

As for operating costs, a 1GW-scale AIDC spends $1.3 billion annually on electricity, while hardware depreciation is six times the electricity costs, meaning an annual depreciation burden of about $7.8 billion.

This cost structure shows that for investors and companies intending to develop AI computing infrastructure, the pressure of ongoing capital investment will far surpass the initial construction phase.

Global Computing Power Demand in 2030: 106GW New Addition

Liu Yangwei cited data showing that the scale of global data centers will reach $1.6 trillion by 2030, with computing load jumping significantly from about 68GW in 2024 to 174GW, an increase in electricity demand of 106GW over six years, averaging nearly 18GW of new electricity supply each year.

Regarding sources of computing power demand, Liu Yangwei categorized the four main buyers as: model developers (Model Makers), cloud service providers (CSP), government, and enterprises.

Among them, model developers and CSPs have clear business models and are currently the largest consumers of computing power and the group with the most robust demand; governments are still in the exploratory phase but have huge potential; enterprises are considered the future "blue ocean market."

On the issue of how enterprises apply AI, Liu Yangwei distinguishes two stages: Most companies are currently in the "AI empowerment" phase, introducing AI into existing organizational structures and processes to enhance efficiency; the future goal is to become "AI native" organizations—where all processes operate with AI at the core, and humans are only responsible for setting objectives and governance oversight.

Liu Yangwei emphasized that AI native organizations must possess seven major characteristics: data must be effectively utilized by AI; processes must be redesigned and equipped with agent functions; AI's role is to assist decision makers in avoiding blind decisions, not to make final decisions directly; organizational roles must be restructured; talent capabilities must be upgraded; and controllable, traceable, and automated governance mechanisms must be established.

Risk Warning and DisclaimerThe market is risky, and investment should be cautious. This article does not constitute personal investment advice and does not consider the special investment goals, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article suit their specific circumstances. Investment based on this is at your own risk. ```