Name: 清潔能源能否應對人工智能熱潮？ (Can clean energy handle the AI boom?)
Uploaded: 2025-06-18T16:49:16.000Z
Duration: 9 min 18 s

情態副詞

I spent some time recently reading through a big spreadsheet of questions submitted by Vox's audience members.

最近，我花了一些時間閱讀 Vox 觀眾提交的一大堆問題。

It was from Kathy, a retired school teacher in New York City.

這是紐約市一位退休教師凱西寄來的。

What is the question that you wanted us to answer?

So the question is can green energy even begin to handle the increased demands that AI and crypto and cloud storage are going to put on our energy system?

那麼問題來了，綠色能源能否應對人工智能、加密貨幣和雲存儲對能源系統提出的更高要求？

I've done some reporting on AI, but I've never thought much of the climate impact of all the AI products we're increasingly using.

我做過一些關於人工智能的報道，但我從未想過我們越來越多使用的人工智能產品對氣候的影響。

And all of our digital belongings like photos and documents and emails getting stockpiled in servers around the world.

我們所有的數字財產，如照片、文件和電子郵件，都被儲存在世界各地的服務器中。

They need a lot of electricity and that electricity has to come from somewhere.

它們需要大量電力，而電力必須來自某個地方。

This is all happening while the climate crisis demands we use less energy, not invent new ways to use more of it.

這一切都發生在氣候危機要求我們減少使用能源，而不是發明新方法來使用更多能源的時候。

I think our climate goals already feel pretty impossible to me, but now it's almost like we haven't changed the goalposts, we've changed the entire game.

我覺得我們的氣候目標對我來說已經很不可能了，但現在幾乎就像我們沒有改變門柱，而是改變了整個遊戲規則。

Within Kathy's big question is a more basic one about how much electricity our digital lives require.

凱西的大問題中還有一個更基本的問題，即我們的數字生活需要多少電力。

At first I was thrown off that Kathy mentioned things like cloud storage and AI and cryptocurrency in one category.

起初，凱西將雲存儲、人工智能和加密貨幣等內容歸為一類，這讓我大跌眼鏡。

But then I realized that their electricity demands happen at the same place.

但後來我意識到，他們的用電需求發生在同一個地方。

Ultimately you're talking about machines loaded up in large facilities who generate computations and they need power, a lot of it, they need water, they need space.

歸根結底，你說的是裝載在大型設備中進行計算的機器，它們需要電力，需要大量電力，需要水，需要空間。

我和亞歷克斯-德弗里斯談過了。

He runs a research site called Digiconomist, where he's dug into this exact topic.

他經營著一個名為 "Digiconomist "的研究網站，專門研究這個話題。

Data centers are massive, often windowless warehouses that house thousands of servers that run virtually nonstop.

數據中心是一個巨大的、通常沒有窗戶的倉庫，裡面有成千上萬臺服務器，這些服務器幾乎不間斷地運行。

Some of the bigger data centers are as big as four football fields and use as much electricity at any given time as 80,000 households.

一些大型數據中心有四個足球場那麼大，任何時候的用電量都相當於 8 萬戶家庭的用電量。

There are more than 8,000 data centers around the world, and the U.S. has more than any other country.

全世界有 8000 多個數據中心，美國的數據中心數量超過任何其他國家。

In 2022, data centers, artificial intelligence, and cryptocurrencies made up about 2% of total global electricity demand.

2022 年，數據中心、人工智能和加密貨幣約佔全球電力總需求的 2%。

But by 2026, that number is expected to double, which is like adding the amount of electricity used by the entire country of Sweden.

但到 2026 年，這一數字預計將翻一番，相當於瑞典全國的用電量。

Support for this video comes from Klaviyo.

Klaviyo works with businesses to turn their data into meaningful connections with their customers through AI-powered email, text messages, and more.

Klaviyo 與企業合作，通過人工智能驅動的電子郵件、簡訊等方式，將企業數據轉化為與客戶之間有意義的聯繫。

According to Klaviyo, over 150,000 brands trust their data and marketing platform to build smarter digital relationships with their customers during the holiday season and beyond.

根據 Klaviyo 的數據，有超過 15 萬個品牌信任他們的數據和營銷平臺，以便在節日期間及以後與客戶建立更智能的數字關係。

Klaviyo has no editorial influence over our work, but they make videos like this possible.

Klaviyo 對我們的工作沒有任何編輯方面的影響，但他們讓這樣的視頻成為可能。

請通過以下鏈接瞭解更多資訊。

That big jump from 2022 to 2026 is thanks to rising cloud storage and cryptocurrency electricity demands, but it's also because of the AI boom.

從 2022 年到 2026 年的大幅躍升歸功於雲存儲和加密貨幣電力需求的增長，但也是因為人工智能的蓬勃發展。

We know AI requires a ton of computational power, but it turns out that the amount of electricity it uses is a really difficult question to answer.

我們知道，人工智能需要大量的計算能力，但事實證明，其耗電量確實是一個難以回答的問題。

AI is a huge umbrella term that includes everything from basic statistical models that detect patterns in data to generative AI that creates text and images and videos.

人工智能是一個龐大的總括術語，包括從檢測數據模式的基本統計模型到創建文本、影像和視頻的生成式人工智能。

That's the most computationally intensive kind.

The thing is, the handful of private tech companies that dominate the AI field don't really disclose how much of their energy use is dedicated to AI specifically.

問題是，在人工智能領域佔據主導地位的少數幾家私營科技公司，並沒有真正披露它們的能源使用中有多少是專門用於人工智能的。

If you look at Google's latest environmental report, it clearly states they absolutely don't want to make a distinction between regular workloads and AI-specific workloads.

如果你看一下谷歌最新的環境報告，其中明確指出他們絕對不想區分常規工作負載和人工智能特定工作負載。

And these companies' AI models are mostly closed-source, meaning no one knows exactly how they are built.

而這些公司的人工智能模型大多是閉源的，這意味著沒有人確切知道它們是如何構建的。

This has left some researchers to try to piece it together on their own.

這使得一些研究人員不得不嘗試自己去拼湊。

Researchers looked at an open-source large-language model called Bloom that has roughly the same amount of parameters as GPT-3, and found that training something like GPT-3 required almost 1,300 megawatt-hours of electricity, about as much power as consumed by 130 homes in the U.S. for one year.

研究人員研究了一個名為布魯姆（Bloom）的開源大型語言模型，該模型的參數量與 GPT-3 大致相同，研究人員發現，訓練類似 GPT-3 的模型需要近 1,300 兆瓦時的電力，大約相當於美國 130 個家庭一年的耗電量。

Today, large-language models like GPT-4 have hundreds of billions of parameters, if not a trillion.

如今，像 GPT-4 這樣的大型語言模型擁有數千億甚至上萬億個參數。

And researchers say that the computational power required to train these models is expected to double every nine months.

研究人員表示，訓練這些模型所需的計算能力預計每九個月就會翻一番。

So far, it has mostly been large-language models driving the AI energy boom.

迄今為止，推動人工智能能源熱潮的主要是大型語言模型。

Of course, that could change going forward.

當然，未來情況可能會發生變化。

Now we see AI on the rise for image generation and also specifically video generation.

現在，我們看到人工智能在影像生成，特別是視頻生成方面的應用日益增多。

So far, we talked about training a large-language model.

到目前為止，我們已經討論瞭如何訓練大型語言模型。

Researchers also looked at energy use from people actually using it.

研究人員還調查了實際使用能源的人的能源使用情況。

It's been estimated by myself and others that a single Jet-GPT interaction would take like three watt-hours, which is comparable to running a low-lumen LED bulb for one hour.

據我自己和其他人估計，一次 Jet-GPT 互動大約需要 3 瓦時，相當於低流明 LED 燈泡運行 1 小時。

So on itself, it doesn't sound like a whole lot, but of course, hey, it's the volume that matters.

是以，就其本身而言，這聽起來並不算什麼，當然，嘿，重要的是音量。

This is 10 times more than a standard Google search.

這比標準的谷歌搜索多出 10 倍。

And of course, if you're talking about millions or billions of interactions, the numbers start to stack up quickly.

當然，如果你說的是數百萬或數十億次的互動，那麼數字就會迅速堆積起來。

Alex took another research approach by looking at the hardware used for AI training and use.

亞歷克斯採用了另一種研究方法，即研究用於人工智能訓練和使用的硬件。

Over 95% of the AI industry uses servers made by the company NVIDIA.

95% 以上的人工智能行業都在使用英偉達公司生產的服務器。

They could sell 1.5 million of their servers by 2027.

到 2027 年，他們可以售出 150 萬臺服務器。

He multiplied that by the information NVIDIA publicizes about each of their servers' energy demand.

他將這一數字乘以英偉達公佈的每臺服務器的能源需求資訊。

He found that data centers devoted to AI alone could consume around 100 terawatt-hours of electricity per year, or about the same as his home country of the Netherlands.

他發現，僅用於人工智能的數據中心每年就能消耗約 100 太瓦時的電力，與他的祖國荷蘭差不多。

There's a big part of Kathy's question I haven't gotten to yet.

凱西的問題中還有一大部分我還沒回答。

Can renewable energy meet the surging demand from the world's data centers?

可再生能源能否滿足全球數據中心激增的需求？

The good news is that using green energy is the stated goal of a lot of these companies.

好消息是，使用綠色能源是許多這些公司的既定目標。

Both Google and Microsoft have made pledges to be net zero by 2030, but there are signs that AI is disrupting those plans.

谷歌和微軟都承諾到 2030 年實現淨零排放，但有跡象表明，人工智能正在擾亂這些計劃。

That's because solar and wind energy can't produce electricity all of the time.

這是因為太陽能和風能不能一直髮電。

And these data centers need to be running all of the time.

而這些數據中心需要一直運行。

In most cases, they will just have a backup connection to the power grid, which will have fossil fuels on it.

在大多數情況下，它們只是與電網有一個備用連接，而電網上會有化石燃料。

Data centers are being built at a rate that renewable energy infrastructure can't keep up with.

數據中心的建設速度遠遠跟不上可再生能源基礎設施的建設速度。

It can take a year to build a data center, but many more years to get a solar or wind farm on an electrical grid.

建造一個數據中心可能只需要一年時間，而將太陽能或風能發電場接入電網則需要更長的時間。

Google's 2024 sustainability report showed that the company's emissions rose by 48% from 2019 to 2023, in large part due to its data center energy consumption, suggesting that integrating AI into their products could make reducing their emissions challenging.

谷歌 2024 年的可持續發展報告顯示，從 2019 年到 2023 年，該公司的排放量增加了 48%，這在很大程度上是由於其數據中心的能耗造成的。

There's already evidence in the U.S. that coal plants that were meant to close are staying open because of data centers' electricity demands, and that state utilities are building new natural gas plants for the same reason.

在美國，已經有證據表明，由於數據中心的用電需求，本應關閉的煤電廠仍在繼續營運，而各州的公用事業公司也出於同樣的原因正在建設新的天然氣發電廠。

But even if these tech companies can look good on their sustainability reports and get to net zero, there's still a problem.

但是，即使這些科技公司能在可持續發展報告中表現良好，實現淨零排放，仍然存在問題。

The thing is that our renewable energy supply globally is limited.

問題是，我們在全球範圍內的可再生能源供應是有限的。

So if we're attributing an increasing part of that to the data center industry, the consequence is that there's less renewables available for everything else.

是以，如果我們將其中越來越多的部分歸因於數據中心行業，那麼後果就是，可用於其他方面的可再生能源越來越少。

That probably will mean that on the whole, we will end up using more fossil fuels anyway.

這可能意味著，無論如何，我們最終都會使用更多的化石燃料。

With all this context, the answer to Kathy's question is that for right now, we aren't prepared for renewable energy to meet the increasing demand of the world's data centers.

綜上所述，凱西問題的答案是：目前，我們還沒有準備好用可再生能源來滿足全球數據中心日益增長的需求。

As users, it would be extremely difficult to opt out of backing up our data on the cloud, or even refrain from using AI.

作為用戶，我們很難選擇不在雲端備份數據，甚至不使用人工智能。

I think AI is embedded in so many things that I'm not sure I will have the option to say, I'm not using it, you know, I'm out.

我認為，人工智能已經融入了很多東西，我不確定自己是否能選擇說 "我不使用它了，我退出"。

Researchers like Alex say the best place to start is to force more transparency from these tech companies.

像亞歷克斯這樣的研究人員說，最好的起點是迫使這些科技公司提高透明度。

In the EU, the AI Act doesn't really force tech companies to disclose anything with regard to the environment.

在歐盟，《人工智能法》並沒有真正強制科技公司披露任何與環境有關的資訊。

And that's the EU, not even talking about the U.S. yet, which is lagging behind a bit on this matter.

這還只是歐盟的情況，美國的情況就更不用說了，美國在這個問題上有些落後。

Some environmental organizations and local communities are calling for moratoriums on And some researchers have proposed the idea of an energy efficiency rating so companies and consumers can choose data centers that are the most sustainable.

一些環保組織和當地社區呼籲暫停使用數據中心，一些研究人員提出了能效評級的想法，以便公司和消費者可以選擇最具可持續性的數據中心。

We could also hope that the servers and data centers will keep getting more energy efficient.

我們還希望服務器和數據中心能不斷提高能效。

But more than anything, this issue emphasizes how desperately we need to be scaling up renewable energy, and fast.

但更重要的是，這個問題強調了我們是多麼迫切需要快速擴大可再生能源的規模。

Not only to meet the ever-increasing data center demands, but so there's plenty of renewable energy to go around.

這不僅是為了滿足不斷增長的數據中心需求，也是為了有充足的可再生能源可供使用。

字幕列表影片播放

清潔能源能否應對人工智能熱潮？ (Can clean energy handle the AI boom?)

entire

force

surge

specifically