Jul 7, 2026, 17:01

Problems with New AI Data Centers

新 AI 數據中心面臨的問題

Introduction

Companies want to build more AI centers. But they have problems with power, laws, and materials.

許多公司希望建造更多 AI 中心，但在電力、法律和材料方面遇到了問題。

Main Body

Many big AI projects are slow or stop. Some projects in the US and Malaysia stopped. This happens because the land is special or the law says no. Google and other companies cannot grow their AI services fast.

許多大型 AI 項目進度緩慢或已停工。美國和馬來西亞的部分項目已停止。這是因為土地特殊或法律禁止。Google 及其他公司無法快速擴展其 AI 服務。

AI centers need a lot of electricity. In the US, power use will grow by 2027. Old power lines cannot give enough energy. Some buildings in California and Amsterdam are empty because they have no power.

AI 中心需要大量電力。在美國，電力使用量到 2027 年將會成長。舊的電線無法提供足夠的能量。加州和阿姆斯特丹的一些建築物因為沒有電力而處於空置狀態。

Some people think the problem will end. They want to make their own power with batteries. Other companies want to use gas to make a lot of energy for their big centers.

有些人認為問題將會解決。他們希望利用電池自行發電。其他公司則希望使用天然氣為其大型中心提供大量能源。

Conclusion

People want AI, but they need more power and better laws to build these centers.

人們需要 AI，但需要更多電力和更好的法律來建造這些中心。

Vocabulary Learning

⚡ The Power of 'BECAUSE'

In this text, we see a very important word for A2 students: because. We use it to explain why something happens.

Look at the pattern: Result $\rightarrow$ because $\rightarrow$ Reason

Examples from the text:

Projects stopped $\rightarrow$ because $\rightarrow$ the land is special.
Buildings are empty $\rightarrow$ because $\rightarrow$ they have no power.

🛠️ Word Swap: 'Many' vs 'Some'

Notice how the writer changes the amount of things:

Many (A lot/Big number) $\rightarrow$ Many big AI projects are slow.
Some (A few/Not all) $\rightarrow$ Some projects in the US stopped.

Quick Tip: Use 'Many' for a crowd and 'Some' for a small group.

Vocabulary Learning

materials (n.)

Things used to make something

Example:We need wood and metal materials to build the house.

services (n.)

Work that a company does for people

Example:The internet service is very slow today.

electricity (n.)

The power used to make lights and machines work

Example:My phone needs electricity to charge.

energy (n.)

The power that comes from electricity, gas, or sun

Example:Solar panels create clean energy from the sun.

batteries (n.)

Objects that store electricity

Example:I put new batteries in my remote control.

Analysis of Global Data Center Growth and Energy Demand

全球數據中心增長與能源需求分析

Introduction

The global expansion of artificial intelligence (AI) infrastructure is currently being slowed by energy shortages, legal problems, and supply chain limitations.

全球人工智慧（AI）基礎設施的擴張，目前正受到能源短缺、法律問題及供應鏈限制的影響而放緩。

Main Body

The growth of high-capacity data centers is facing a major problem: AI models are evolving quickly, but the physical buildings and power systems are being built too slowly. According to the Uptime Institute, about 50% of 250 large global projects announced between 2021 and 2024 are likely to be delayed or cancelled. For example, a project in Virginia was stopped by a court because it was too close to a historic battlefield, while other projects in Arizona and Malaysia were also cancelled. Consequently, companies like Google are facing 'compute-constrained' environments, which limits how well their AI services can operate.

高容量數據中心的增長面臨一個重大問題：AI 模型的演進速度很快，但實體建築與電力系統的建設速度太慢。根據 Uptime Institute 的數據，2021 年至 2024 年間公布的 250 個全球大型項目中，約有 50% 可能會被延期或取消。例如，維吉尼亞州的一個項目因過於靠近歷史戰場而被法院叫停，而亞利桑那州與馬來西亞的其他項目也遭到取消。因此，像 Google 這樣的公司面臨著「計算受限」的環境，限制了其 AI 服務的運行表現。

Furthermore, these problems are made worse by a serious lack of energy. The Energy Information Administration (EIA) predicts that U.S. power demand will rise from 4,195 billion kWh in 2025 to 4,399 billion kWh by 2027, mainly because of AI and cryptocurrency. This increase is especially strong in the commercial sector, which is expected to use more power than residential areas by 2026. The Uptime Institute emphasized that current power grids, particularly in North America, cannot support this growth. For instance, some facilities in California remain empty, and there are legal fights over power connections in Amsterdam. Similarly, the UK government has been criticized for not carefully checking if they have enough electricity to meet their AI goals.

此外，嚴重的能源短缺讓這些問題更加惡化。美國能源資訊署（EIA）預測，受 AI 與加密貨幣影響，美國電力需求將從 2025 年的 4,195 十億千瓦時增加到 2027 年的 4,399 十億千瓦時。這種增長在商業部門尤為強勁，預計到 2026 年，商業部門的用電量將超過住宅區。Uptime Institute 強調，目前的電網（特別是在北美地區）無法支持這樣的增長。例如，加州的部分設施仍處於空置狀態，而阿姆斯特丹則在電力連接問題上陷入法律爭議。同樣地，英國政府也因未能仔細核查電力供應是否足以達成 AI 目標而受到批評。

Despite these difficulties, experts have different views on the future. JLL predicts that 1,200 data centers will be completed by 2030, suggesting that using onsite power generators and better batteries could reduce the reliance on the main power grid. On the other hand, the scale of current plans is enormous. Six projects announced last year aim for at least 5GW each, and the seven largest planned facilities propose a total of 45GW of onsite power, mostly using natural gas.

儘管面臨這些困難，專家對未來的看法不一。JLL 預測到 2030 年將有 1,200 個數據中心完工，認為利用現場發電機與更高效的電池可減少對主電網的依賴。另一方面，目前計劃的規模極其龐大。去年公布的六個項目 each 目標均至少為 5GW，而七個最大計劃設施共提出 45GW 的現場電力，主要使用天然氣。

Conclusion

Although the demand for AI infrastructure remains high, the success of these projects depends on overcoming significant energy, legal, and logistical barriers.

雖然 AI 基礎設施的需求依然很高，但這些項目的成功與否，取決於能否克服重大的能源、法律及物流障礙。

Vocabulary Learning

🚀 The 'Logic Bridge': Mastering Transition Words

At an A2 level, you likely use and, but, and because. To reach B2, you need to show a professional relationship between your ideas. The text uses specific "bridges" to move from a problem to a result or a contrasting opinion.

🛠 The Connector Upgrade

A2 Word (Simple)	B2 Upgrade (from text)	How it changes the feel
So	Consequently	Moves from a casual chat to a formal report.
Also	Furthermore	Signals that you are adding a stronger point, not just another one.
But	Despite these difficulties	Shows you can acknowledge a problem before introducing a solution.
But	On the other hand	Perfectly balances two different professional perspectives.

🔍 Linguistic Analysis: The "Contrast Shift"

Look at this sequence from the text:

"Despite these difficulties... JLL predicts that 1,200 data centers will be completed... On the other hand, the scale of current plans is enormous."

Why this is B2 level: The writer doesn't just say "things are bad but some are good." They use a three-step logical chain:

Acknowledgement (Despite...) $\rightarrow$ 2. Positive Prediction (JLL predicts...) $\rightarrow$ 3. Reality Check (On the other hand...).

💡 Pro-Tip for your Writing

Instead of starting every sentence with the subject (e.g., "The company is...", "The power is..."), start with a transition phrase followed by a comma.

Incorrect/A2: The power grid is old. But the demand is growing.
B2 Level: The power grid is old. Consequently, the demand is growing faster than the system can handle.

By using Consequently or Furthermore, you aren't just giving information—you are guiding the reader's brain through your logic.

Vocabulary Learning

infrastructure (n.)

The basic physical and organizational structures and facilities needed for the operation of a society or enterprise.

Example:The government is investing heavily in digital infrastructure to improve internet access in rural areas.

evolving (v.)

Developing gradually over time, especially from a simple to a more complex form.

Example:The way we communicate is constantly evolving due to new social media platforms.

consequently (adv.)

As a result of something that has happened.

Example:The company failed to innovate; consequently, it lost its market share to smaller competitors.

constrained (adj.)

Limited or restricted in a particular way.

Example:The project was constrained by a very tight budget and a short deadline.

emphasized (v.)

To give special importance or prominence to something in speaking or writing.

Example:The teacher emphasized the importance of reviewing the vocabulary before the final exam.

reliance (n.)

Dependence on or trust in someone or something.

Example:The city is trying to reduce its reliance on fossil fuels by installing solar panels.

enormous (adj.)

Very large in size, quantity, or extent.

Example:The cost of repairing the bridge after the flood was enormous.

logistical (adj.)

Relating to the detailed coordination of a complex operation involving many people, facilities, or supplies.

Example:Moving the entire office to a new city presented several logistical challenges.

Analysis of Global Data Center Expansion Constraints and Energy Demand Projections

全球數據中心擴展限制與能源需求預測分析

Introduction

The global expansion of artificial intelligence (AI) infrastructure is currently impeded by systemic energy shortages, regulatory hurdles, and supply chain limitations.

目前全球人工智慧 (AI) 基礎設施的擴展，正受到系統性能源短缺、監管障礙及供應鏈限制的阻礙。

Main Body

The proliferation of high-capacity data centers is characterized by a significant discrepancy between the rapid evolution of AI models and the slower pace of physical infrastructure deployment. The Uptime Institute reports that approximately 50% of 250 global projects exceeding 100MW announced between 2021 and 2024 are expected to face delays or cancellation. This trend is exemplified by the cessation of the Prince William Digital Gateway project in Virginia, which encountered judicial intervention due to its proximity to a historic battlefield, as well as the cancellation of Project Range in Arizona and the Cyberjaya campus in Malaysia. Such impediments result in 'compute-constrained' environments for providers like Google, limiting the operational capacity of AI services.

高容量數據中心的普及，呈現出 AI 模型快速演進與實體基礎設施部署緩慢之間的明顯差異。Uptime Institute 報告指出，在 2021 年至 2024 年間宣布的 250 個超過 100MW 的全球項目中，約 50% 預計將面臨延遲或取消。這一趨勢在維吉尼亞州的 Prince William Digital Gateway 項目中得到體現，該項目因鄰近歷史戰場而遭到司法干預而停止，此外亞利桑那州的 Project Range 與馬來西亞的 Cyberjaya 校區亦被取消。此類障礙導致 Google 等供應商處於「運算受限」的環境，限制了 AI 服務的運作能力。

Institutional constraints are further compounded by severe energy deficits. The Energy Information Administration (EIA) projects that U.S. power demand will escalate from 4,195 billion kWh in 2025 to 4,399 billion kWh by 2027, driven primarily by AI and cryptocurrency infrastructure. This surge is particularly evident in the commercial sector, which is forecasted to surpass residential demand in 2026. The Uptime Institute posits that existing power grids, especially in North America, are insufficient to sustain this trajectory, citing instances of vacant facilities in California and legal disputes over grid connectivity in Amsterdam. Furthermore, the UK government's AI ambitions have been criticized for a perceived lack of rigorous auditing regarding resource allocation and electrical viability.

制度性限制進一步被嚴重的能源短缺所加劇。美國能源資訊局 (EIA) 預測，在 AI 與加密貨幣基礎設施的推動下，美國電力需求將從 2025 年的 4,195 億度電上升至 2027 年的 4,399 億度電。這一增長在商業領域尤為明顯，預計 2026 年將超過住宅需求。Uptime Institute 認為，現有的電網（尤其是北美）不足以維持此趨勢，並舉例提及加州有設施空置，以及阿姆斯特丹關於電網連接的法律糾紛。此外，英國政府的 AI 雄心被批評在資源分配與電力可行性方面缺乏嚴格的審計。

Despite these systemic frictions, divergent perspectives exist regarding the industry's trajectory. JLL anticipates the completion of 1,200 data centers by 2030, suggesting that the implementation of onsite power generation and enhanced battery storage may mitigate grid dependency. Conversely, the scale of current ambitions is immense; six projects announced last year aim for at least 5GW each, with the seven largest planned facilities proposing a combined 45GW of onsite power, predominantly sourced from natural gas.

儘管存在這些系統性摩擦，但業界對發展軌跡持有分歧看法。JLL 預計到 2030 年將完成 1,200 個數據中心，認為實施 onsite 自主發電與強化電池儲能可減輕對電網的依賴。相反，目前的野心規模巨大；去年宣布的六個項目每個目標至少為 5GW，而七個最大的計劃設施合計建議 45GW 的 onsite 電力，且主要來源為天然氣。

Conclusion

While demand for AI infrastructure remains high, the realization of these projects is contingent upon overcoming substantial energy, legal, and logistical barriers.

雖然對 AI 基礎設施的需求依然高漲，但這些項目的實現將取決於能否克服重大的能源、法律及物流障礙。

Vocabulary Learning

The Architecture of 'Nominalization' and Abstract Density

To bridge the gap from B2 to C2, a student must move beyond describing actions and start encoding concepts. The provided text is a masterclass in Nominalization—the process of turning verbs (actions) or adjectives (qualities) into nouns. This is the hallmark of high-level academic and professional English, as it allows the writer to treat complex processes as single 'objects' that can be manipulated grammatically.

🔍 The Linguistic Pivot

Observe the transition from a B2-style sentence to the C2 density found in the text:

B2 (Action-oriented): AI infrastructure is expanding globally, but it is being stopped because there isn't enough energy and there are too many regulations.
C2 (Concept-oriented): *"The global expansion of artificial intelligence (AI) infrastructure is currently impeded by systemic energy shortages, regulatory hurdles, and supply chain limitations."

What happened here? Instead of using verbs like "expanding" or "stopping," the author uses nouns: Expansion, Shortages, Hurdles, and Limitations.

🛠️ Deconstructing the C2 Mechanism

The Semantic Compression: By using "regulatory hurdles" instead of "the fact that it is hard to follow regulations," the author compresses a whole clause into a single noun phrase. This creates a higher information density.
The Agency Shift: Notice the phrase "The proliferation of high-capacity data centers is characterized by a significant discrepancy..." Here, the subject isn't a person or a company, but a phenomenon (proliferation). C2 English often removes the human agent to emphasize the systemic logic.
Collocational Sophistication: Pair the nominalized nouns with high-tier verbs.
- Impediments $\rightarrow$ result in
- Constraints $\rightarrow$ are compounded by
- Ambitions $\rightarrow$ have been criticized for

⚡ C2 Application: The 'Nominalization' Formula

To elevate your writing, apply this transformation: [Verb/Adjective] $\rightarrow$ [Abstract Noun] $\rightarrow$ [Insert Precision Modifier]

Example: "The grid is not strong enough" $\rightarrow$ "The insufficiency of the existing power grids."
Example: "They disagree on where the industry is going" $\rightarrow$ "Divergent perspectives regarding the industry's trajectory."

Scholarly Note: Overusing this can lead to 'wooden' prose, but in the context of white papers, legal briefs, and C2 Proficiency exams, it is the primary tool for achieving an objective, authoritative tone.

Vocabulary Learning

impeded (v.)

Delayed or obstructed the progress or movement of something.

Example:The project's progress was severely impeded by a lack of funding and bureaucratic red tape.

proliferation (n.)

A rapid increase in the number or amount of something.

Example:The proliferation of smartphones has fundamentally changed how the world consumes information.

discrepancy (n.)

An illogical or unexpected lack of compatibility or similarity between two or more facts.

Example:The auditor found a significant discrepancy between the company's reported earnings and its actual bank balance.

cessation (n.)

The process of ending or being brought to an end.

Example:The cessation of hostilities was welcomed by both nations after years of brutal conflict.

compounded (v.)

To make a bad or difficult situation worse by adding further complications.

Example:The economic crisis was compounded by a sudden surge in inflation and unemployment.

posits (v.)

To put forward as a basis of argument; to suggest or assume the existence of something.

Example:The scientist posits that the observed phenomenon is a result of quantum entanglement.

divergent (adj.)

Tending to be different or develop in different directions.

Example:The two political parties held divergent views on how to handle the national healthcare crisis.

mitigate (v.)

To make something less severe, serious, or painful.

Example:The government implemented new zoning laws to mitigate the effects of urban sprawl.

contingent (adj.)

Subject to chance; dependent on certain circumstances being met.

Example:The success of the merger is contingent upon the approval of the regulatory commission.

Practice All words in a crossword→