Why Everyone Suddenly Hates AI Data Centers — 文稿

M10:00

看起来，所有人都讨厌 data center。

Everyone, it appears, hates data centers.

民调显示，大多数美国人，不管是 Democrat 还是 Republican，都反对 data center，而政客们也开始抓住这种愤怒情绪做文章。

Polls show the majority of Americans, both Democrats and Republicans, are opposed to them, and politicians are picking up on the anger.

有一位作者开玩笑说，每一场州级选举，都不过是两个五分钟前还支持 data center 的政客，在互相指责对方支持 data center。

One writer joked that every state election is just two politicians who supported data centers five minutes ago accusing their opponent of supporting data centers.

在一些人看来，这全都是 AI 行业糟糕宣传话术带来的后果。

To some, this is all a consequence of the AI industry's terrible messaging.

在另一些人看来，这和 big tech 留下的负面遗产有关。

To others, it's about the negative legacy of big tech.

还有一些人觉得，这根本就是一场大型的 China psyop。

And to some, it's just all one big China psyop.

不管这些说法到底有多少是真的，和这些社区里的人聊下来你会发现，这件事同样重要，甚至更重要的，其实是他们在塑造自己未来时，是否拥有主动权和控制权。

However much those things may or may not be true, conversations with the folks in these communities reveal that this is as much, if not more, about their agency and control in shaping their own future.

现在，如果有人觉得外面的形势越来越灰暗，尤其是看到这么多政客都在跟着起哄，那也完全可以理解。

Now, one could be forgiven for thinking it's getting gloomy out there, especially with all these politicians piling on.

但我其实认为，不仅有一条路可以往前走，这甚至可能是有史以来最容易赢、也最能创造价值的一场政治战役。

But I actually think that not only is there a path forward, this might be the most winnable, value-creating political battle of all time.

The AI Daily Brief 是一档每日 podcast 和视频节目，内容是关于 AI 领域最重要的新闻和讨论。

The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.

M11:33

今天的话题，对经常收听节目的朋友来说应该不会陌生。

Today's topic will not be unfamiliar for you regular listeners.

围绕 AI infrastructure buildout，尤其是 AI data center 的问题，已经在这个节目边缘徘徊了一年多了，因为最近这一波 AI buildout 正变得越来越突出。

Questions around the AI infrastructure buildout and AI data centers specifically have lurked around the edges of this show for more than a year now as this most recent wave of AI buildout has come to the fore.

不过现在，这个问题已经正式不再只是小范围内的讨论了。

The issue has now officially, however, escaped containment.

所以今天的目标，是围绕这个问题做一整期 primer，算是从头到尾梳理一遍，最后再看看这里面到底还有没有任何乐观的空间。

And today the goal is to do an entire primer, kind of end to end on this issue, understanding on the other side whether there is any room for optimism at all.

那先从一个问题开始，为什么我们现在要谈这个？

So starting off, why are we talking about this right now?

简短的答案就是：中期选举。

The answer in short is the midterm elections.

这个议题，一些 pollster 已经关注了一段时间了，但随着它越来越多地进入选举话语，越来越多人开始围绕它做调查，而数据并不好看。

This is an issue that certain pollsters have been paying attention to for a while now, but as it becomes more and more a part of the election discourse, more and more people are surveying about it, and the numbers are not good.

在 Gallup 五月份的一项民调中，已经有百分之七十一的受访者反对在自己所在地区建设 data center，而且有整整百分之四十八的人是强烈反对。

In a Gallup poll from May, 71% of respondents were already opposed to data centers in their area, and a full 48% were strongly opposed.

而情况后来只变得更糟。

And things have gotten nothing but worse.

Heatmap News 过去一年一直在做关于反对本地 data center 的民调。

Heatmap News has been conducting polling on opposition to local data centers for the past year.

在他们二月份的民调中，百分之五十一的人反对在自己家附近建设 data center。

In their February poll, 51% of people were opposed to data centers being built close to their home.

而在他们本月这一版民调里，反对比例达到了百分之七十五。

And in this month's edition of their poll, opposition reached 75%.

更重要的是，强烈支持的比例从二月份的百分之九，缩水到现在只有百分之四。

What's more, strong support has shrunk from 9% in February to just 4% now.

从事民调工作的人都被这个话题变化的激烈程度震惊了。

People who work with polling are shocked at how aggressively the topic is moving.

Lobbyist Justin Slaughter 写道，我从来没见过这种情况。

Lobbyist Justin Slaughter wrote, I've never seen this.

data center 的反对者正在把支持者转化过去，而且不是转成中立，而是直接转成反对。

Data center opponents are converting supporters not just to neutral but to opposition.

这里面的趋势太疯狂了。

The trend here is crazy.

在十二个月里，对 data center 的强烈反对，从百分之二十四上升到了百分之六十一。

From 24% strong opposition to data centers to 61% strong opposition in 12 months.

上一次一个议题上的政治观点变化得这么快，是什么时候？

When's the last time political opinions on an issue shifted this fast?

而且这并不是说只有一项哗众取宠的民调是这样。

And it's not just like this is one sensationalist poll.

这周早些时候我在节目里还分享过另一个来自 Echelon 的民调，结果发现，美国人宁愿自家后院有一座核电站，也不愿意有一个给 AI 供电的 data center。

Another one that I shared on an earlier show this week from Echelon found that Americans would rather have a nuclear power plant in their backyard than a data center to power AI.

眼下，相比在附近建 data center，人们支持在附近建煤电厂的可能性大概是前者的两倍。

At the moment, people are about twice as likely to be into coal plants being built nearby them as opposed to data centers.

那么对一些人来说，这就是证据 A，说明这场讨论不只是关于用水、污染这类问题的理性争论，因为很明显，在这些维度上，煤电厂肯定会更糟。

Now, for some, this is exhibit A in why this conversation is not just a rational argument about things like water and pollution, as quite clearly a coal plant is going to be worse on those vectors.

但正如我们接下来会讨论的，这里面还有很多别的因素。

There is, as we will discuss, a lot more going on.

那我刚才也说了，民调真正重要的，不只是数字本身，还有它处在一个选举年的背景之下。

Now, as I said, what matters about the polling is not just the numbers, but the context of an election year.

而在这种环境里，这成了一个少见的两党共同关注的问题。

And in that environment, this is a rare bipartisan issue.

在二零二五年八月，共和党人对于在附近建一个 data center，净支持率还是正的。

In August of 2025, Republicans were net support for a data center nearby.

独立选民的净支持率是负百分之五，民主党人是负百分之八。

Independents were at net -5%, and Democrats were at -8%.

而现在，虽然独立选民和民主党人的支持率都已经断崖式下跌，分别跌到负百分之六十五和负百分之七十五，但就连共和党人，对于在自家附近建 data center，也已经是净反对百分之四十三。

Now, while independents and Democrats have both fallen stratospherically to -65% and -75%, even Republicans are net 43% against having a data center near their home.

Andrew Curran 写道，如果你想知道为什么这个议题每天都在新闻里，为什么那些之前还支持它的政客突然一脸慌张地改口，那是因为这个议题在美国公众那里已经变得极其有毒、极其敏感。

Andrew Curran wrote, if you're wondering why this issue is in the news every day and why politicians who were previously supportive are suddenly changing their tune with a panicked look in their eyes, it's because this issue has become incredibly radioactive with the American public.

而事实上，政客们确实已经注意到了。

And indeed, noticed, the politicians have.

The Bulwark 写道，政客们突然发现所有人都讨厌 data center，并指出你放眼望去，到处都是政客突然开始飞快地和它们切割。

The Bulwark writes, politicians suddenly noticed that everybody hates data centers, pointing out that everywhere you look, politicians are suddenly sprinting away from them.

The Wall Street Journal 写道，曾经力挺 data center 的政客现在开始抨击它们，并举了州长的例子，包括 Texas 的 Greg Abbott，以及 Pennsylvania 州长 Josh Shapiro。

The Wall Street Journal wrote, politicians who once championed data centers are now bashing them, pointing to the example of state governors, including Texas's Greg Abbott and Pennsylvania Governor Josh Shapiro.

所以合在一起看，你就会看到一种极其有毒的混合物：极端负面的公众情绪，被政客捕捉到之后，政客又把这个议题放到竞选的最前排，让它变得更响亮；接着媒体又进一步放大它，因为对媒体来说，不光是这种对立情绪，就连政客态度的转变本身，都是写文章时最让人兴奋的素材。

So altogether then, you have this incredibly toxic mixture of extremely negative public sentiment that becomes picked up by politicians who make the issue even louder by putting it front and center in their campaigns, who are then further amplified by media for whom not just the antagonism but also the shift among politicians is absolute red meat for their articles.

说到这里，其实甚至值得问一句，什么是 AI data center？

Now at this point, it's worth even asking what is an AI data center?

因为恕我直言，虽然很多反对者对它们的担忧是有道理的，但我不一定认为大多数人真的能说清楚，一个 data center 里面到底发生了什么，它们到底做什么，它们怎么和经济互动，以及为什么它们 überhaupt 是必要的。

Because with due respect to a lot of opponents' concerns about them, I don't necessarily think that most people could even really define what goes on inside one, or what they do, how they interact with the economy, and why they are necessary at all.

data center 基本上就是一个仓库大小的计算机，接入了工业级系统。

A data center is basically a warehouse-sized computer that is plugged into industrial systems.

电力要么通过公用事业电网线路进来，要么通过专门的发电设施进来，然后这些能源被用来驱动一排排芯片和服务器。没错，它们可以用来做训练 AI 或者提供 AI 服务这类事情，但它们也几乎是互联网上其他一切活动的基础。

Power arrives either via utility lines or dedicated generation, and that energy is used to power racks of chips and servers that, yes, can be used for things like training or serving AI, but are also essential for pretty much everything else that happens on the internet.

你的 Gmail，你看的 Netflix streaming，这一切其实都有一个物理层面的组成部分，而且都是通过 data center 运转的。

Your Gmail, your Netflix streaming, all of that does have a physical component, and it runs through the data centers.

投资人 Nic Carter 写道，我觉得 data center 遭到反弹，有一个不可忽视的原因是，人们过去想象互联网是一个没有重量、住在 cloud 里的东西。

Investor Nic Carter wrote, I think a non-negligible component of the data center backlash comes from people having imagined that the internet was a weightless thing that lived in the cloud.

他们现在才发现，互联网需要实体基础设施、冷却系统和电力，于是他们对此感到不满。

They are now learning that the internet requires physical infrastructure, cooling, and power, and they are upset about it.

而冷却系统，当然就是争议的一大来源。

And that cooling is of course a big part of where the controversy comes in.

空气、蒸发冷却塔，或者液冷循环系统，会以某种组合方式把热量从系统里带走并排到外面，然后数据再一次离开。

Some combination of air, evaporative towers, or liquid loops remove heat from the system and push it outside, and then the data leaves once again.

重点是，这些就是很大的空房间，里面有很多很多台小计算机连接在一起，组成一台大计算机。

The point is, these are big empty rooms with many, many small computers connected to make one big computer.

就是这样。

That's it.

这就是 data center。

That's what a data center is.

而它们对我们所生活的整个世界的基础设施来说，是关键的，也是绝对必不可少的。

And they're critical and absolutely essential to the entire infrastructure of the world that we live in.

那为什么人们会对 data center 这么生气呢？

So why are people so angry about data centers?

简短的答案是：所有事情加在一起。

The short answer is it's everything.

而把这些各种各样产生敌意的原因理清楚，很不幸，我认为对于弄明白接下来该怎么走，是很重要的。

And untangling all of those various reasons for the animosity is unfortunately, I think, important for figuring out how to move forward.

Puck News 最近问选民，他们认为 data centers 会怎样给自己的社区带来负担，问题是他们觉得下面这些关于 data centers 的说法是真的还是假的。

Puck News recently asked voters how they believe data centers burden their community, asking whether they thought the following statements about data centers were true or false.

你可以看到，这里排在最前面的两项，就是大家最担心的两个问题。

And you can see here two of the biggest concerns right up at the top.

当被问到他们是否认为 data centers 给当地电网带来了过大的压力时，45% 的人说“绝对是真的”，33% 的人说“可能是真的”，加起来一共是 78%。

When asked if they believed that data centers placed too much strain on the local power grid, 45% said definitely true and 33% said probably true, a total of 78%.

当被问到 data centers 是否消耗了当地过多的供水时，44% 的人说“绝对是真的”，31% 的人说“可能是真的”。

When asked if data centers consume excessive amounts of the local water supply, definitely true was at 44% and probably true was at 31%.

当被问到他们是否认为 data centers 会拉低周边社区的房产价值时，27% 的人说“绝对会”，32% 的人说“可能会”。

When asked whether they believed that data centers would decrease property values in the surrounding community, 27% said definitely, 32% said probably.

当被问到 data centers 是否会产生持续不断、让人受不了的噪音时，25% 的人说“绝对是真的”，另外 34% 的人说“可能是真的”。

When asked if data centers generate constant disruptive noise, 25% said definitely true, with 34% adding probably.

当被问到 data centers 是否没有创造足够多的本地就业，因而不足以证明建设它们是合理的时，24% 的人说“绝对是真的”，30% 的人说“可能是真的”。

When asked if they do not create enough local jobs to justify their construction, 24% said definitely true, 30% said probably true.

而当被问到 data centers 是否会提高地方税时，老实说，这件事到底会通过什么机制发生，其实都不是特别清楚。

And when asked if data centers increase local taxes, which it isn't even exactly clear by what mechanism that would happen.

但即便如此，总体上仍然有 48% 的人说这要么“绝对是真的”，要么“可能是真的”。

Still, 48% overall said it was either definitely or probably true.

所以从这项民调里，你能看到人们反感 data centers 的一大批原因，而这些原因都跟住在附近的人真实的生活质量有关。

So with that poll, you see a big cross-section of reasons for data center antipathy that are about actual quality of life for people near them.

但当然，这还不是全部。

But of course, that is not the whole story.

那说到这一切愤怒到底该怪谁，对一些人来说，很多批评里那些泛滥的事实错误，指向的是某些有意制造冲突和争议的明确行动。

Now, when it comes to who is to blame for all of this anger, for some, the rampant factual inaccuracies in a lot of the critiques point to specific deliberate attempts to sow strife and controversy.

反 data center 的这些说法简直疯了，而且清楚证明我们已经进入了一个后真相时代，在这个时代里，很多人会被说服，往往是被恶意行为者说服，去抛弃现实，让阴谋论和谎言来左右政策偏好。

The anti-data center stuff is insane and clear evidence that we have entered a post-truth era where many people can be convinced, often by malicious actors, to abandon reality and allow conspiracies and falsehoods to drive policy preferences.

Frank Fleming 提出了一个越来越常被问到的问题：所有这些反 data center 的歇斯底里，背后是 China 吗？还是说我们只是因为完全本土自产的愚蠢，在搬起石头砸自己的脚？

Frank Fleming asked a question that is coming up increasingly: Is China behind all the anti-data center hysteria, or are we shooting ourselves in the foot over completely homegrown stupidity?

他转发了一条 President Trump 的帖子，里面 Trump 说，如果他是一个镇的市长，或者一个州的州长，而且有机会引进一个大型工厂，一个 AI 工厂或者一个 data center，他绝对会想要。

He retweeted a post of President Trump talking about how if he were the mayor of a town or governor of a state and had a chance to get a big plant in, an AI plant or a data center, he would absolutely want it.

Trey 也把矛头指向 China，说，这可真奇怪。

Trey also pointed towards China saying, that's weird.

这几乎就像是没有封禁 TikTok 真的产生了现实后果，因为美国民众还在继续被 CCP propaganda 毒害。

It's almost like not banning TikTok has had real-world consequences as the US population continues being poisoned by CCP propaganda.

谁能想到呢？

Who would have guessed?

Rob Leclerc 也跟着加码写道，我们过去三十年一直在抱怨，说我们失去了所有薪水不错的制造业工作。

Piling on, Rob Leclerc writes, we spent the last 30 years whining about how we lost all our good-paying manufacturing jobs.

现在我们终于有机会重新开始建设了。

We finally have an opportunity to start building again.

Tech companies 愿意在 CapEx 和劳动力上花费数万亿美元，可现在两党都说不行，就因为一个假的 Chinese psyop，说 data centers 的用水量是实际用水量的三万倍。

Tech companies are willing to spend trillions on CapEx and labor, and now both parties are saying no because of a fake Chinese psyop that data centers use 30,000x more water than they really do.

这种“China 在背后操纵”的指控甚至已经进入了法庭，因为 Shark Tank 的 Kevin O'Leary 高调宣称，反对他在 Utah 推进的那个 data center 的本地活动人士，背后事实上就是 China。

This accusation that China's behind this has even made it into the courts, as Shark Tank's Kevin O'Leary loudly proclaimed that it was in fact China behind local activists who opposed the data center that he was working on in Utah.

我的总体看法是，psyop 这个说法大概至少有一部分是真的，意思是说，在社交媒体上制造分裂，现在已经是美国所有对手在几乎所有议题上的一种有意策略。

My general take is that the psyop argument probably has at least some truth in the sense that sowing social media discord is just now an intentional strategy of all America's adversaries around everything.

所以，就算并不存在一个专门针对 data centers 的大型行动，只要它是社交媒体上的热门争议话题，那几乎可以肯定，某个地方的 bot farms 就在参与推波助澜。

And so even if there's not some giant campaign specifically about data centers, by virtue of the fact that it is a hot-button topic on social media, it is almost absolutely the case that bot farms from somewhere are contributing.

但我也觉得，把这当成借口实在太方便了；如果我们把它当成全部原因，那肯定会让我们停下来，不去真正处理那些现实问题，而这些问题如果不解决，可能会造成更多麻烦。

But I also think it's way too convenient as an excuse, and treating it like it's all that will absolutely halt us from actually dealing with the real issues that could cause many more problems if they're not addressed.

在回应有人指责整件事都是 psyop 的时候，Connecticut Senator Chris Murphy 提到了另一个让人们变得愤怒的原因，也就是 AI industry 谈论它可能如何影响人们生活的方式。

Responding to someone accusing the whole thing of being a psyop, Connecticut Senator Chris Murphy got at another reason that people have gotten angry, which is about the way that the AI industry talks about its likely impact on their lives.

Murphy 写道，我不知道啊，AI industry 基本上一直在承诺要大规模摧毁工作岗位，还要用让人上瘾的产品毒害我们的孩子，而现在感觉人们只是相信了他们说的话。

Murphy wrote, I don't know, the AI industry has been basically promising to destroy jobs at scale and poison our kids with addictive products, and it feels like people are just believing them.

这里“用让人上瘾的产品毒害我们的孩子”这一部分，感觉像是他把反 AI 和反 social media 混在一起了。顺便说一句，我觉得这其实也是其中很大一部分原因。

Now, the poison our kids with addictive products part feels like he's conflating anti-AI with anti-social media, which, by the way, I think is actually a big part of this.

但他绝不是唯一一个认为 AI industry 应该为此承担部分责任的人，原因就在于这个行业的领导者谈论失业问题的方式。

But he is far from alone in thinking that the AI industry bears some of the responsibility here for the way that its leaders have talked about job loss.

因为别搞错了，虽然人们总体上不喜欢 data centers，但他们非常明确地不喜欢的是 AI data centers。

Because make no mistake, while people don't like data centers in general, they very specifically don't like AI data centers.

而在那些把两者区分开的民调里，你可以看到，反对声音最大的，明确来说就是 AI data centers。

And in polls that differentiate between the two, you can see that it is AI data centers specifically that have the most opposition.

Brendan Steinhauser 写道，是的，data centers 已经变成了政治上的烫手山芋。

Brendan Steinhauser wrote, yes, data centers have become politically toxic.

只要你看看民调，或者跟普通美国人聊聊这件事，就会很清楚。

It's clear when you review the polling or meet with regular Americans about this.

而且这不只是对能源成本、用水量、外观，或者噪音的担忧。

And it's not just concerns about energy costs, water usage, aesthetics, or noise.

人们担心 AI 会取代人类。

People are worried about AI replacing human beings.

他们不希望这些公司去打造 superintelligence。

They don't want the companies to build superintelligence.

Zarathustra 写道，我总体上是支持 data center 的，但如果你是一个普通人，那过去十年里，你从 Silicon Valley 听到的故事大概是这样的。

Zarathustra writes, I'm generally pro-data center, but if you're a normie, here's the story you've been hearing from Silicon Valley over the last decade.

AI 很可能会抢走你的工作。

AI is probably going to take your job.

它可能会杀了你。

It might kill you.

你很可能会变成永久底层阶级的一部分。

You will probably become part of the permanent underclass.

过去十年里，那些大型科技亿万富翁和公司，比世界上任何其他商业集团都更敌视、更恶意地对待保守派，而他们会变得富到难以想象，同时你和所有你爱的人都会失去工作。

The very big tech billionaires and companies who have been more hostile and malicious to conservatives than any other business bloc in the world over the last decade will become unimaginably rich while you and everyone you love will lose your jobs.

然后变穷。

And become poor.

主要的发言人，或者说台前负责的面孔，Dario 和 Sam Altman，都公开敌视你，敌视你的价值观、你的政治立场、你的生活方式，以及你所代表的一切。

The main spokesmen or faces in charge, Dario and Sam Altman, are openly hostile to you, your values, your politics, your way of life, and everything you stand for.

科技圈那些毫无品味的 Philistines，会在你家后院建起一个丑到极点的工业怪物，而且很可能一分钱都不会给你。

The tasteless Philistines in tech are going to plant an utterly hideous industrial monstrosity in your backyard and probably pay you nothing.

而你付出这一切之后，换来的是什么呢？问号，问号，问号。

And for all this, in return, you get question mark, question mark, question mark.

他们的结论是，要理解为什么普通人会有所保留，这一点一点都不难。

It's not remotely difficult, they conclude, to understand why normies have reservations.

投资人 Chamath Palihapitiya 写道，frontier AI 的集体领导层，在做好基本功这件事上失败得很惨，比如说，描绘出 AI 可能带来的积极前景。

Investor Chamath Palihapitiya writes, the collective leadership of frontier AI has failed miserably in doing the basics, like painting a positive picture of what is possible with AI.

他们没有多花笔墨去讲那些半杯水还满着的故事，反而到处都是紧张兮兮、灾难化的叙事，或者晦涩难懂的技术黑话，这些东西加在一起，只会滋生不信任、恐惧和怀疑。

Instead of spilling the ink on the glass-half-full, stories abound of pearl-clutching, catastrophizing, or arcane technological jargon that together breed mistrust, fear, and skepticism.

但对一些人来说，反 AI 的情绪，远不只是营销做得不好，或者还没有展示出价值那么简单。

But for some, the anti-AI sentiment goes a lot deeper than marketing and a lack of demonstrated value.

T. Greer 写道，更好的信息传达，也就是说，永远不要谈 AI 抢工作，这并不是问题所在。

T. Greer writes, better messaging, i.e., never talking about AI taking jobs, is not the problem.

问题在于，AI 到现在还没有给整个国家带来太多实实在在的好处。

The problem is that AI has just not delivered too many tangible benefits to the country writ large.

普通人想到 LLMs，想到的就是他们孩子用来在学校作业上作弊的东西。

The average person thinks of LLMs as the thing their kids use to cheat on school assignments.

AI 也许在边边角角上让他们的生活更方便了一点，但那也只是边边角角。

AI may have made their life more convenient on the margins, but that is the margins.

所以到最后，你是在要求社区接受这些 data centers，认可一项技术伸到本地的一条触手；这项技术已经让远方的一些人变得非常富有，但却没有给整个国家提供任何实实在在的东西。

So you end up asking communities to accept these data centers and sanction the local tentacle of a technology that is already making people far away very rich, but which has not provided the country writ large with any tangible goods.

问题在于实质，而不是话术。

The problem is the substance, not the messaging.

沿着这个思路，我也看到有人提出另一个观点：即使是 AI 做得特别好的事情，比如提高生产力，也不一定是普通人想要的。

Another point along these lines that I've seen some make is that even the things that AI does really well, i.e., increasing productivity, aren't necessarily things that the average person wants.

在最近一次 podcast 露面中，作者 Ben Thompson 说，每隔十年，Silicon Valley 就会重新学一遍：消费者不想为 software 付钱，而且消费者并不在乎提高生产力。

In a recent podcast appearance, writer Ben Thompson said that every 10 years Silicon Valley relearns that consumers do not want to pay for software and that consumers do not care about being productive.

我觉得说到 AI，问题不只是消费者不在乎自己变得更高效；而是当你把这件事和整个 AI 行业都在说 AI 要来抢你的工作、再加上一大堆 CEO 拿它当借口来裁掉他们本来就打算裁的人时，所谓生产力就只是另一种说法：你要让我交出更多我并不拥有的劳动。

I think when it comes to AI, it's not just that consumers don't care about being more productive, it's that when you combine it with the entire AI industry saying that AI is going to come for your jobs and tons of CEOs using it as an excuse for layoffs that they were going to make anyway, productivity is just another way of saying, you're gonna make me give away more of my labor that I don't own.

不过，如果这只是讨厌 AI，那哪怕这样，也和现实情况不一样。

Still, if it were just dislike of AI, even that would be different than what it actually is.

AI 继承的，是过去十五年来人们对整个科技行业看法一路下滑的那顶王冠。

AI is the inheritor of the mantle of a 15-year decline of people's perception of tech in general.

我的感觉是，对 AI 的这些负面情绪里，至少有百分之七十五，其实是针对整个 Big Tech 的。

My feeling is that at least 75% of this negativity against AI is about Big Tech at large.

过去十年里，Big Tech 出过下面这些非常糟糕的事情，而且还远不止这些。

Over the last 10 years, Big Tech has had the following very bad stories and many more.

比如操纵选举、孩子们的屏幕时间让人担心、让人上瘾的内容黑洞把人一步步激进化、把 TV、电影和新闻这样的媒体行业搞垮、让不同社区彼此对立、公司赚取巨额利润而本地社区却越来越穷、woke 公司做出各种文化选择，等等等等。

Election manipulation, screen time worries amongst kids, addictive content rabbit holes that radicalize people, killing the media business like TV, movies, and news, polarizing communities against one another, massive profits versus local communities getting poorer, woke companies making cultural choices, and on and on and on.

我觉得 AI，尤其是数据中心，终于让这种积压已久的愤怒有了现实世界里的目标——在数据中心这件事上，甚至是字面意义上的现实世界目标。

I think AI and data centers specifically has finally let a long-simmering anger have real-world— in data centers, literal real-world— targets.

在我前面读过的同一篇帖子里，Chamath 又写道：数据中心很不幸地成了一种象征，代表着一个非常狭窄的科技精英群体和一个不值得信任的阶层，享受着不对称的巨大上行收益。

In the same post I was reading from before, Chamath again writes, data centers have unfortunately become the symbolic representation of the asymmetric upside for a very narrow tech elite and a class that are untrustworthy.

不过要记住，美国人现在对精英和机构的信任，正处在历史最低点。

Now, keep in mind, Americans' trust of elites and institutions right now is at an all-time low.

根据最近一项 Gallup 民调，在美国成年人当中，对九个主要美国机构表示非常有信心或者相当有信心的平均比例，已经降到了有史以来最低的百分之二十七。这九个机构包括教会、军队、最高法院、银行、公立学校、报纸、国会、工会组织，以及大企业。

According to a recent Gallup poll, the average percentage of US adults that express a great deal or quite a lot of confidence in nine major US institutions, including the church, the military, the Supreme Court, banks, public schools, newspapers, Congress, organized labor, and big business, is down to its lowest point ever at 27%.

只有四分之一的成年人，对美国生活中的主要机构表示有信心。

Just 1 in 4 adults express confidence in the major institutions of American life.

而说到大企业，这个数字只有百分之十七。

The number when it comes to big business is just 17%.

而当你本来就不信任机构的时候，数据中心做生意的方式，也就是关起门来、不让公众看到，用保密协议之类的工具，几乎就是专门为了加深这种不信任而设计的。

And when you're already mistrustful of institutions, the way that data centers have done business, i.e., behind closed doors outside the eye of the public, using tools like nondisclosure agreements, is almost tailor-made to increase that mistrust.

NBC 最近发表了一篇文章，标题是《NDA 如何让 AI 数据中心的细节对美国人保持隐藏》。我觉得，这正是眼下这个问题里一个非常非常重要、重要到怎么强调都不为过的部分。

NBC recently published a piece called How NDAs Keep AI Data Center Details Hidden from Americans, and this, I think, is a massive, massive, hard-to-overstate part of this problem right now.

技术不断向前推进，代表着一种你无法控制的变化。

Technology moving ahead represents change you can't control.

而你无法控制的变化，会让人觉得自己对生活的掌控感更低。

Change you can't control makes people feel less agency in their lives.

所以，本来人们就已经觉得掌控感在下降，然后他们甚至连正在发生什么、这些决定是怎么做出来的都没机会看到。这种事，任何人都不会乐意接受，尤其是美国人。

And so the fact that already people are feeling less agency, and then they don't even have a chance to see what's going on and how these decisions are being made, that's not something that anyone, but particularly Americans, is going to take kindly to.

Jasmine Sun 最近围绕这个问题做了一些很出色的报道。在她最近一篇长篇文章里，她提出这其实是问题中非常大的一部分；这篇文章也让她出现在很多主要新闻媒体和 podcast 上。

Jasmine Sun has done some great reporting around this recently, and in a recent long-form piece that has gotten her featured on numerous major news outlets and podcasts, she argued that this is actually a huge part of the problem.

在那篇六千字长文里，我认为最重要的一段是她写的这段：我的这些谈话让我开始怀疑，数据中心交易里的技术内容，大多可能并不是重点。

In what I think is the most important part of that entire 6,000-word tome, she writes, my conversations made me wonder if the technical content of data center deals is mostly beside the point.

闭环冷却系统？

Closed-loop cooling systems?

我不信。

I don't believe them.

缴纳数百万美元税款？

Paying millions in taxes?

我不信。

I don't believe them.

创造一千个工作岗位？

Creating 1,000 jobs?

我不信。

I don't believe them.

我听到的是一种对每一个说法的本能怀疑。

I hear a reflexive skepticism of every claim.

更确切地说，AI 公司和社区打交道的方式——强迫签 NDA，拿几十亿美元的承诺来诱惑人，把环境外部性远远推离 AI 富裕用户群体的视线——很像政治中关于黑钱的经典故事。

Rather, the way that AI companies engage with communities—forcing NDAs, dangling billion-dollar promises, pushing environmental externalities far away from AI's wealthy user base—resembles a classic story about dark money in politics.

你的政客被亿万富翁和巨型公司收买了，好让他们来坑你。

Your politicians are being bought by billionaires and gigantic corporations to screw you over.

这不是 Skynet，这是寡头统治。

It's not Skynet, it's the oligarchy.

数据中心成了一种替代物，承载着人们对科技的那些很难说清楚的担忧，也承载着一种越来越明显的可能性：AI 将会像互联网当年那样，彻底改变我们的生活方式。而这种担忧，至少部分是由两种民粹式直觉推动的。

Data centers are a proxy for largely inarticulable concerns over tech and the emerging likelihood that AI is going to radically change our way of life, much in the same way the internet has, and driven at least in part by two populist intuitions.

第一种直觉是：人的价值越是从有形物品的生产中被抽象出去，情况就越糟。另一种是：如果某个东西的价值不容易解释，如果它的运作机制不透明，甚至根本无法被了解，那我这个普通人很可能就在某种程度上被坑了。

One is that the further human value is abstracted out from the production of tangible things, the worse it is, and also that if the value of that thing isn't easily explained, if its mechanics are opaque or even unknowable, then I, the average guy, am probably somehow getting screwed.

这并不是说这些担忧是假的，或者没有正当理由，只是说它们常常指错了方向。

This doesn't mean that the concerns are fake or unjustified, only that they are often misdirected.

你们大概也听得出来，虽然在这些批评和担忧里，有不少部分我并不同意，或者我认为完全是基于错误信息，但我觉得，要想找到前进的路，唯一现实的起点就是先理解那些真实存在的担忧，然后试着为这些具体问题找到解决办法。

As you can probably tell, while there are entire sections of the critiques and concerns that I either disagree with, or that I think are just based on total misinformation, I think that the only realistic starting point for finding a way forward is understanding the real concerns that exist, and trying to find solutions for those specific problems.

不幸的是，到目前为止，提出的解决方案大多都是暂停令这种形式。

Unfortunately, the solutions that have come up so far are largely in the form of moratoriums.

Interconnected Capital 有一张数字地图，用来追踪这些暂停令，显示有九十七个县级禁令、九十三个市级禁令，以及二十八个镇级禁令。

Interconnected Capital has a digital map that tracks moratoriums, showing 97 county bans, 93 city bans, and 28 town bans.

最近，我所在的 New York 州成了第一个正式对 data center 实施暂停令的州，但现在这个趋势正在迅速蔓延。

Recently, my state of New York became the first state to officially impose a data center moratorium, but now the tide is moving quickly.

就在过去几周里，我们看到了一些很典型的例子：原本极其支持 data center 的州长，转而改变立场，基本上开始比赛看谁能制定全国最严格的规则。

In the last couple of weeks, we've had great examples of formerly extremely data center-bullish governors reversing their opinions and basically competing to have the strictest rules in the country.

我当然说的是 Pennsylvania 州长 Josh Shapiro，以及 Texas 州长 Greg Abbott。

I'm referring, of course, to Pennsylvania Governor Josh Shapiro and Texas Governor Greg Abbott.

还有，National Republican Senatorial Committee 最近发了一份关于 data center 政治风险的备忘录，特别提到了 Ohio，但也涉及美国其他地区。这份备忘录似乎一下子让整个 Republican Party 都站到了反对 data center 的一边。

And a recent memo from the National Republican Senatorial Committee about the liability of data centers, specifically in Ohio, but in other parts of the country as well, seems to have gotten the entire Republican Party on board with being anti-data center in one fell swoop.

Michigan 的 Republican Tom Barrett 正在投放一条广告，用他的话说，是谴责那些政客和 big tech 亿万富翁走得太近，强行把 data center 塞进我们的社区，尽管他以前的投票记录其实是支持 data center 的。

Michigan Republican Tom Barrett is running an ad decrying, in his words, politicians cozying up with big tech billionaires to force data centers on our communities, despite his voting record in favor of data centers before.

Wisconsin 州长的 Republican 候选人 Tom Tiffany 也在特意把他的 Democratic 对手 David Crowley 描绘成 data center 的伟大拥护者。

Republican candidate for Wisconsin Governor Tom Tiffany is going out of his way to paint his Democratic counterpart, David Crowley, as the great champion of data centers.

有意思的是，Tiffany 似乎想一口气把 culture war、反 data center 情绪和反环保主义全都揉在一起。他这周在 X 上发帖说：

Interestingly, Tiffany seems to be trying to combine the culture war, anti-data center sentiment, and anti-environmentalism all in one fell swoop, posting on X this week.

David Crowley 的 data center 计划会毁掉 Wisconsin 至少十万英亩土地。

David Crowley's data center plan would destroy at least 100,000 acres of Wisconsin land.

他的 Oprah 式方案要求 data center 百分之百由太阳能、风能和电池储能供电。

His Oprah approach calls for data centers powered by 100% solar, wind, and battery storage.

你家后院会分到一个 data center，再加上成千上万英亩的太阳能板。

You get a data center and thousands of acres of solar panels in your backyard.

Heatmap 的 Matthew Zeitlin 总结得特别到位：现在每一场全州选举，都是两个五分钟前还支持 data center 的人在说，对方支持 data center。

Heatmap's Matthew Zeitlin perfectly sums it up: every statewide election now is two people who supported data centers five minutes ago saying that their opponent supports data centers.

M121:55

那接下来我们该往哪里走呢？

So where do we go from here?

首先，确实还有很多工作要做，而且也已经有人在认真做事，试图对抗这些错误信息。

Firstly, there is work to be done and good work being done on trying to combat the misinformation.

比如关于用水量的很多梗，要么是脱离了语境，也就是说，没有理解这些巨大的用水加仑数，和其他所有用途相比到底意味着什么；要么就是彻头彻尾的事实错误，结果变成了梗。

A lot of the memes about water usage, for example, are somewhere between out of context, i.e., understanding what these big numbers of gallons of water used actually mean relative to all other uses, or just straight-up factual inaccuracies that become memes.

比如在 Karen Hao 的书 Empire of AI 里，她把用水量算错了，差了整整一千倍。

For example, in Karen Hao's book Empire of AI, she miscalculated the amount of water used by a factor of 1,000.

虽然她已经承认了这个错误，但这个印刷错误仍然在流传，而且正在影响人们对这件事的看法。

And while she has admitted the error, the misprint remains in circulation and is contributing to people's takes on this.

围绕另一个担忧，也在发生类似的过程。Bloomberg 去年的一篇文章是这样说的：AI data center 正在让电费飙升。

A similar process is happening around the concern that, as Bloomberg put it in an article last year, AI data centers are sending power bills soaring.

但当人们真正深入去看时，情况似乎是，电价上涨最厉害的地方，和 data center 所在的地方之间，其实并没有真正的统计相关性。

But when people actually dig in, it appears like there's actually no real statistical correlation between where electricity prices are going up the most and where data centers are located.

当然，这并不意味着这不是问题，但它是一个具体 data center 设计，以及周边环境设计的问题，而不是某种自然规律，说 data center 一定会把你的电价推高。

Which of course doesn't mean that it's a non-issue, but it's a question of specific data center design and the environmental design around it, not some rule of nature that data centers are going to jack up your electricity prices.

不幸的是，我们前面讨论过的那种基础信任鸿沟，限制了这些理性论证真正被人接受的能力。

Unfortunately, that baseline trust gap that we discussed limits the ability for these rational arguments to find purchase.

还是来自 Puck News 的一项民调。当受访者被问到，他们觉得某些支持 data center 的论点有多有说服力时，那些事实纠正式的说法排名都相当低。

Again, from a Puck News poll, when asked how convincing the respondents found certain arguments in favor of data centers to be, the factual fixes ranked pretty low.

比如，他们专门问到了“用水过多”这个说法是基于那个数学错误，并说明这个错误后来已经被纠正了。结果只有百分之八的人觉得这非常有说服力，只有百分之二十的人觉得这有点说服力。

For example, they asked specifically about the excessive water claim being based on that math error, acknowledging that it has since been corrected, and only 8% of people found that very convincing and only 20% of people found that somewhat convincing.

再比如那个论点，说 data center 会为当地社区居民创造数千个高薪工作。

Around the argument that data centers would create thousands of high-paying jobs for local community members.

同样，人们就是不相信。

Again, people just don't believe it.

只有百分之十二的人觉得这非常有说服力，百分之二十三的人觉得这有点说服力。

Only 12% found that very convincing and 23% found that somewhat convincing.

你不得不想，这并不是因为人们不在乎高薪工作的价值，而是在说，行吧，等我真的看到这些工作的时候再说。

One has to think that that's not people being dismissive of the value of high-paying jobs, it's saying, yeah, right, we'll talk when I actually see those jobs.

他们能找到的最有说服力的论点，是 data center 会给当地带来税收，用来支持学校、道路和其他服务。

The most convincing argument that they could find was data centers generating money in local tax revenue for schools, roads, and other services.

但这也没有形成压倒性优势，只有百分之十四的人觉得非常有说服力，另外百分之二十八的人觉得有点说服力。

This still wasn't a knockout, with only 14% finding it very convincing, and another 28% finding it somewhat convincing.

那说到这里面可以归因于围绕 AI 和就业的宣传话术的那一部分，谢天谢地，最近这方面已经有变化了。

Now, when it comes to the part of this that could be attributed to the messaging around AI and jobs, thankfully that has shifted lately.

我倾向于认为，这种变化有点太少也太晚了，但比话术改变更重要的是，证据显示，所谓 AI 会造成大规模失业的说法，从一开始就是错的。

I tend to think it's on the too little, too late side of things, but what's more important than just the messaging changing is the fact that the evidence shows that the argument that AI was going to cause massive job loss was always wrong in the first place.

换句话说，Ford 决定重新雇用他们所谓的 graybeard engineers，这件事的说服力，比 CEO 们说“哎呀，我之前对失业问题判断错了”要大得多。

In other words, Ford deciding to rehire what they call graybeard engineers does a lot more than CEOs saying, oops, I was wrong about job loss, ever could.

另外，对于某些类型的支持数据中心的信息来说，也有比科技公司或者数据中心建设方自己更好的信息来源。

There are also better sources for certain types of pro-data center messaging than the tech companies or data center builders themselves.

当然，如果说所有工会都站在数据中心这一边，那肯定是夸大其词了；但如果你要找一个它们立刻产生巨大正面影响的领域，那就是技工行业。

It would certainly be overstating it to say that all labor unions were on the side of the data centers, but if you are looking for one area where they have had an enormous positive effect right away, it's in the trades.

坦白说，这些工会既看到了这些好处，同时他们自己也是这些社区的成员，所以他们处在一个很独特的位置，能够处理这个问题里的细微差别和复杂性，同时又不否认这些东西建起来所带来的真实价值。

And frankly, the unions who are both seeing those benefits, but who are also members of those communities, are uniquely positioned to navigate the nuance and complexity of the situation without surrendering the realness of the value that is coming from these things being built.

而随着暂停令的威胁越来越大，这些群体的声音也越来越响。

And as the threat of moratoriums gets bigger, these groups are getting louder.

International Brotherhood of Electrical Workers 最近表示，新英格兰地区提出的数据中心禁令和暂停令，引用原话，“威胁到数十亿美元潜在投资的落地，并将消除 IBEW 成员关键的就业机会。”

The International Brotherhood of Electrical Workers recently said that proposed data center bans and moratoriums in New England, quote, threaten to block billions of dollars in potential investment and will eliminate critical job opportunities for IBEW members.

如果实施全州范围的禁令，新英格兰将失去无数由当前 AI 技术热潮带来的工作岗位和基础设施投资。

If statewide bans are implemented, New England will lose countless jobs and infrastructure investments created by the current boom in AI technology.

他们接着说，为什么这对 IBEW 很重要：数据中心会在与 AI 和能源建设相关的行业里，为电力工人创造大量工作机会。

Going on, they said, why this matters to the IBEW: Data centers create substantial work opportunities for electrical workers in industries tied to AI and energy buildout.

对数据中心实行强有力的劳工标准，将提高工资、改善工作条件，并保护本地工人。

Strong labor standards on data centers will raise wages, improve working conditions, and protect local workers.

对工会学徒制和培训项目的投资，将有助于培养下一代熟练的 IBEW 工人。

Investment in union apprenticeship and training programs will help build the next generation of skilled IBEW workers.

数据中心项目可以带来可观的地方税收收入和基础设施投资。

Data center projects can generate significant local tax revenue and infrastructure investment.

全面禁令会打击投资，并把好工作送到其他地区。

Total bans discourage investment and send good jobs to other areas.

而当你拨开围绕这个议题、像评论区那种愤怒情绪的时候，你会发现，也有人不愿意向这场讨论中任何一方的任何极端立场屈服。

And when you peel back the comment-section-style anger that surrounds this issue, there are folks out there who are unwilling to surrender to any extremes of any side in the conversation.

作家 Noah Rothman 曾问：“难道数据中心恐慌就没有解药吗？”

In a response to author Noah Rothman asking, "Is there no cure for the data center hysteria?"

Nebraska 国会议员 Don Bacon 回应说：“有，说真话。”

Nebraska Congressman Don Bacon said, "Yes, speak the truth."

数据中心可以用聪明的方式建设，既节约用水，又扩大能源生产。

Data centers can be built smartly to conserve water and expand energy production.

如果我们要对数据中心开战，那我们就不可能赢得 AI 战争。

We can't win the AI war if we're going to wage war against data centers.

数据中心已经对我居住的 Sarpy County 产生了积极影响。

Data centers have made a positive impact in Sarpy County, where I live.

而对科技公司来说，议程上非常重要的一项，必须是不只是解释本地社区如何能够从承接数据中心中受益，而是要说明它们在现实中已经如何受益。

And very high on the agenda for the tech companies has to be actually not just explaining how local communities could benefit from hosting data centers, but how they are in practice.

比如，最近一些研究显示，在某些地区，例如 Virginia 的 Dominion Energy 用户，实际上因为数据中心以及相关的负荷增长，电价还略有下降。

For example, some recent research is showing areas where, for example, Dominion Energy customers in Virginia actually saw slightly lowered electricity rates because of data centers and the associated load growth.

Quincy 是 Washington 的一个小农业镇，从 2006 年就开始吸引数据中心入驻。

Quincy is a small farming town in Washington that started attracting data centers back in 2006.

在接下来的二十年里，他们持续增加数据中心，目前这个市镇里有三十个数据中心。

They continued adding data centers over the subsequent two decades and currently have 30 in the municipality.

大约有九百个工程和建筑岗位直接服务于这些数据中心；而根据 Washington State 的分析，这九百个数据中心岗位中每一个岗位，还会带来另外四到六个建筑、零售和本地服务领域的工作岗位。

There are around 900 engineering and construction jobs that work directly on the data centers, and according to analysis from Washington State, another four to six jobs in construction, retail, and local services for every one of those 900 data center jobs.

这些数据中心贡献了 Quincy 房产税的百分之五十七，并资助了公共建筑，包括一个造价一千五百万美元的六泳道游泳池、一所造价一亿两千万美元的先进高中，以及一个目前正在建设中的大型室内体育综合体。

The data centers contribute 57% of Quincy's property taxes and have funded public buildings, including a $15 million, six-lane swimming pool, a $120 million state-of-the-art high school, and a large indoor sports complex that is currently under construction.

Quincy 的贫困率已经从 2012 年的百分之二十九点四，降到了 2024 年的百分之六点二。

The poverty rate in Quincy has fallen from 29.4% in 2012 to 6.2% in 2024.

前市政管理员 Pat Haley 把这称为 Quincy Miracle。

Pat Haley, a former city administrator, called it the Quincy Miracle.

他说，Quincy 曾经是一个以农业和农产业为主的小镇。

He said Quincy was a farming and ag town.

如果你打算留下来，那就是你的职业道路。

That was your career path if you were going to stay.

Data centers 把一个原本可能机会很少的小镇，变成了一个什么都是全新的地方。

Data centers opened up what could have been a low-opportunity town into this place with brand new everything.

另一个例子是 Northern Virginia 的 Loudoun County。

Another example is Loudoun County in Northern Virginia.

这里是 Data Center Alley 的所在地，也就是全球 data centers 最密集的地区。

It's home to Data Center Alley, which is the region with the highest concentration of data centers on Earth.

这里有超过两百个设施，占全球容量的大约百分之十四。

It houses over 200 facilities representing about 14% of global capacity.

Data centers 占用了这个 county 大约百分之三的土地，却贡献了它大约百分之四十的预算。

Data centers take up about 3% of the county's land and account for around 40% of its budget.

不过，data center 行业对经济来说是一个巨大的净贡献者。

However, the data center industry is a massive net contributor to the economy.

根据一份州审计报告，大约七万四千名 Virginia 居民在这个行业工作，带来了五十五亿美元的劳动收入。

According to a state audit, around 74,000 Virginia residents work in the industry, representing $5.5 billion in labor income.

这里也是美国最富有的 county。

It is also the wealthiest county in the US.

不过，和其他超级富裕的 county 不同，它的住宅 property tax 税率也是全国最低的之一。

However, unlike other ultra-wealthy counties, it also has one of the lowest residential property tax rates in the country.

Loudoun County 在过去十年里每年都在降低 property tax 税率，现在只收百分之零点八。

Loudoun County has lowered its property tax rate every year over the past 10 years and now charges just 0.8%.

而这很大程度上是因为 data centers 在替大家买单。

And that's largely because data centers are footing the bill.

它们现在贡献了当地税收资金的大约百分之四十二，高于二零二四年的百分之三十二点八。

They now contribute around 42% of local tax funding, up from 32.8% in 2024.

而且，在最近一项研究中，Chamber of Progress 发现，和 Northern Virginia 另外三个 county 相比，这个 county 在学校、图书馆和公共安全上的支出都高于平均水平。

What's more, in a recent study, the Chamber of Progress found that compared to the other three counties in Northern Virginia, the county delivers above-average spending on schools, libraries, and public safety.

比如 Fairfax County，它收的 property tax 大约高出百分之三十五，但在公共服务上的支出大约少了百分之四十五。

Fairfax County, for example, charges around 35% more in property taxes while spending around 45% less on public services.

确实，对一些人来说，data center 问题的显而易见答案不是怎么做宣传，而是要拿出行动。

Indeed, for some, the obvious answer to the data center problem is not about messaging, it's about action.

X 上的 Polymath 写道，科技公司需要停止试图用理性和逻辑来说服人们接受 data centers。

Polymath on X writes, the tech companies need to stop trying to use reason and logic to convince people about data centers.

它们应该开始建公园。

They need to start building parks.

Lulu Cheng Meservey 写道，最久经考验、最可靠的 PR 手段，就是给人们东西。

Lulu Cheng Meservey writes, the most tried and true PR tactic is just giving people stuff.

Julius Caesar 的宴会，Rockefeller 的一角硬币，Marshall Plan，Oprah's Favorite Things，United 上的 Starlink。

Julius Caesar's feasts, Rockefeller's dimes, the Marshall Plan, Oprah's Favorite Things, Starlink on United.

每次都管用。

Works every time.

Labs 和 data centers 应该建公园，补贴账单，办烧烤聚会，赞助实地参观。

Labs and data centers should build parks, subsidize bills, hold a cookout, sponsor field trips.

公司在 marketing 和 lobbying 上花的钱多得多，而这些做法有一半时间还会适得其反。

Companies are spending way more on marketing and lobbying that backfires half the time.

直接给人们东西就好了。

Just give people stuff.

虽然这听起来很犬儒，但在实际操作中，它为什么就一定得是犬儒的呢？

And as cynical as that sounds, why does that actually have to be cynical in practice?

这整件事其实归结为一个简单的等式。

The whole thing here comes down to a simple equation.

一个社区会不会觉得，data center 落在自己社区所带来的价值，超过了它造成的成本？

Does a community feel that the value that they get from a data center being in their community is more than the cost that it incurs?

坦白说，即使是那些非常支持 data center 的人，也没有，或者至少不应该，把所有担忧都说成是编出来的。

And frankly, even people who are extremely pro-data center aren't, or at least shouldn't be, dismissing all of the concerns as actually made up.

Pirate Wires 的 Mike Solana 写道，AI 的 messaging 简直烂得一塌糊涂，但我在现实生活中从来没见过任何一个知道 Dario 这个名字的人。

Pirate Wires' Mike Solana writes, AI messaging has been absolute dog poo, but I've never met a single person in the real world who knows the name Dario.

我听到过关于外观的担忧、噪音的担忧、房产价值的担忧。

I've heard aesthetic concerns, noise concerns, property value concerns.

这些担忧是合理的，也需要被解决。

These are legit and need to be addressed.

其他所有事情都是干扰。

Everything else is a distraction.

再说回建公园这类事情，如果那个核心问题——这些成本换来的好处到底值不值——光靠它们自己支付电费、缴纳地方税还算不过来，那这其实就跟任何其他谈判一样。

And coming back to things like building parks, if that core equation of are those costs worth it for the benefit doesn't come up just on the basis of covering the cost of their own electricity and paying local taxes, well, it's just like any other negotiation.

开发商还能再加些什么，才能让这笔账对社区来说变成正的？

What more can the developers add to make the equation come up positive for the community?

很明显，数据中心建设方已经收到这个信号了。

Clearly, the data center builders are getting the memo.

Meta 最近设立了一个十亿美元的基金，用来投资他们建设地点周边的地方项目。

Meta has recently launched a billion-dollar fund to invest in local initiatives around where they're building.

在他们位于 Ohio 的 Pike County 设施上，OpenAI 宣布将由 OpenAI 直接提供四千万美元，用于他们说会由当地社区来确定的重点事项；而且这还不包括他们声称在整个运营周期内会带来的数亿美元州和地方税收。

In their Pike County, Ohio facility, OpenAI has announced $40 million from OpenAI directly for priorities that they say will be defined by the local community, and that is in addition to what they claim will be the hundreds of millions of dollars in state and local tax revenue over its lifetime.

他们还会向大约八十四万四千名 Ohio 的大学、社区学院和技术学校学生赠送价值八千四百万美元的 Codex 额度。

They're also giving away $84 million in Codex credits to approximately 844,000 Ohio college, community college, and technical students.

还有一个迹象也说明他们终于收到信号了：Microsoft 今年早些时候宣布，他们已经决定停止和地方政府使用 NDA，以提高透明度。

And in further signs that they're finally getting the memo, Microsoft earlier this year announced that they had made the decision to end the use of NDAs with local governments in order to increase transparency.

那那些政客呢？

But what about those politicians?

如果他们做得好，他们就有机会成为中间协调者，把负面影响降到最低，同时带来一些非常惊人的正面成果。

Well, if they do it well, they have an opportunity to be the intermediaries that minimize the negatives and deliver some incredible positives.

有报道称，West Virginia 正在研究如何利用数据中心，彻底取消他们的所得税。Americans for Tax Reform 的主席 Grover Norquist 似乎对这个计划非常兴奋，他建议所有州都应该要求数据中心缴纳的税款用于在地方层面降低房产税、在州层面降低所得税。

Reports are coming out that West Virginia is exploring how they could use data centers to eliminate their income tax altogether, which is a plan that Americans for Tax Reform President Grover Norquist seems very excited about, suggesting that all states should require that tax payments by data centers will go to reduce property taxes at the local level and income taxes at the state level.

这是个挺有意思的练习。

This is a fun exercise.

你的城镇需要多大的数据中心，才能替代你们的房产税？

How large a data center would your town need to replace your property taxes?

他们用一个两万人口的小镇做了点粗略估算，按美国地方政府平均征收房产税的水平来算，结果发现，引用一下，简而言之，往宽了算，一个甚至还没达到二百五十兆瓦 hyperscale 门槛的数据中心，就足以让你们完全取消房产税。

Doing some back-of-the-napkin math for a 20,000-person town using the average rate that U.S. local governments collect in property taxes, they found that, quote, in short, being generous, you could drop your property taxes entirely with a data center that doesn't even hit the hyperscale threshold of 250 megawatts.

但是，而且这是个很大的但是，如果选择的政策是暂停令，那这一切都不可能发生。

But, and this is a big but, none of this can happen if the policy chosen is moratoriums.

说清楚，暂停令什么也解决不了。

To be clear, moratoriums do nothing.

解决不了问题。

To solve the problem.

它们只是把问题往后拖。

They simply kick the can down the road.

对于那些内心真正关心的其实只是想阻止变化的人来说，暂停令会让他们感觉不错，但它们并不会真正解决任何问题，也不会创造任何机会去争取可能得到的好处。

They feel good to people whose core emotional consideration is actually just trying to stop change, but they do not actually address any problems or create any opportunities to find the good that might be won.

接下来是一个真正打破叙事的点，你不会在那些标题党新闻里看到。

And here's a real narrative violation that you're not going to see in clickbait headlines.

令人震惊的是，人们并不愚蠢。

People, shockingly, aren't stupid.

他们对数据中心确实有真实的担忧，是的，这些担忧来自各种不同的问题。

They have real concerns about data centers that are, yes, based on a variety of different issues.

你随便问两个人，为什么不喜欢数据中心，可能会得到三个不同答案。

Which, ask any two people and you might get three different answers for why they don't like data centers.

但事实证明，如果你去问他们，他们更希望看到的是解决那些具体问题的方案，而不是把一切都停掉。

But it turns out, if you ask them, they'd rather have plans that addressed those specific problems than just shut everything down.

Morning Consult 最近的一项民调发现，相比暂停令，选民更倾向于给数据中心制定规则。

A recent Morning Consult poll found that voters prefer data center rules to a moratorium.

当被问到他们更希望采取什么方式来处理数据中心建设时，百分之二十三的人说应该暂停所有新项目，但百分之五十七的人说，制定清晰规则，让负责任的项目可以建设。

When asked what their preferred approach to data center construction would be, 23% said a moratorium on all new projects, but 57% said clear rules, let responsible projects be built.

当被问到一个符合披露、成本和公众审查要求的项目时，百分之二十四——是的，几乎四分之一——的人说应该阻止它建设，但百分之五十八的人说应该允许它继续推进。

When asked about a project that meets disclosure, cost, and public review requirements, 24%—yes, almost a quarter—said stop it from being built, but 58% said allow it to move forward.

当被问到这些公司不只是付清自己该付的费用，而是支持更广泛的振兴，比如要求公司为其设施所需的电网升级买单时，只有百分之十三的人反对，百分之七十八的人支持。

When asked about the companies not just paying their way, but supporting a broader revitalization, i.e., requiring companies to pay for the grid upgrades their facilities need, just 13% opposed that and 78% supported it.

这个议题的讽刺之处在于，它也许是我见过的最有可能达成政治妥协、最容易赢下来的议题。

The irony of this issue is that it is maybe the most winnable political compromise issue that I have ever seen.

这些数据中心对于建设它们的人，以及依赖它们的更大基础设施来说，绝对是任务关键级别的。

These data centers are absolutely mission critical to the people who are building them and the larger infrastructure that relies on them.

归根到底，这一点并不重要，而且我相信从长期来看也不会重要：也许在以前，大家都没怎么关注的时候，事情会更容易推进，可以像过去一直那样，通过 NDA 和极少数把关人来做，而不是必须去获得真正当地社区的批准。

It does not matter ultimately, and I believe will not matter in the long run, that it might have been easier before when people weren't paying attention to do things as they had always been done through NDAs and a very small number of gatekeepers rather than having to go get the approval of the actual communities.

但你完全可以相信，如果给出一条可执行的前进路径，尤其是那种最大的考量基本上只是要在建设这些 data center 的社区里多花一些钱，那么所有参与 data center 的人都会抓住这个机会。

But you better believe, if given an actionable path forward, especially one where the biggest consideration pretty much just comes down to spending more money in the community that those data centers are built in, everyone involved with the data centers will jump at the chance.

所以，如果你想找一些积极的方向，就去关注那些真正提出前进路径、并且借此参与这个议题的政治人物。

So if you're looking for positive places to go, look for the politicians who are jumping on this issue by actually offering a path forward.

Wisconsin 州长候选人 David Crowley 最近发布了一份关于 data center 的新闻稿。

Wisconsin gubernatorial candidate David Crowley recently put out a press release on data centers.

这份新闻稿写道：如果 Crowley 当选州长，他会让各个社区自己决定，他们到底要不要 data center。

The press release reads: As governor, Crowley would let communities decide for themselves whether they want a data center at all.

如果他们说不要，这个决定就会得到尊重。

And if they say no, that decision would be honored.

如果一个社区说要，那么开发商必须承诺完全透明，包括禁止 NDA。

If a community says yes, the developer must commit to full transparency by banning NDAs.

提供百分之百可再生能源，这将有助于减少我们燃烧的煤炭数量。

Provide 100% renewable energy, which will combat the amount of coal we burn.

承担百分之百的基础设施和电网升级成本。

Cover 100% of the cost of infrastructure and grid upgrades.

要求向该州的 Focus on Energy Fund 捐款，这笔钱可以用来降低全州 Wisconsin 居民的公用事业费率。

Require contributions to the state's Focus on Energy Fund, which could be used to bring down utility rates for Wisconsinites statewide.

用工会劳动力、项目劳工协议，以及社区福利协议来建设 data center。

Build the data center with union labor, project labor agreements, and community benefit agreements.

创造薪酬良好、足以养家糊口的工作岗位。

Create good-paying, family-supporting jobs.

并保护当地水资源，以及我们整体的环境。

And protect local water and our environment as a whole.

我认为这就是我们接下来会看到的方向：聪明的政治人物会制定一些开发商实际上能够满足的清单，但这一切要建立在社区拥有自主权的基础上。

I think this is the direction we're going to see, where the smart politicians create checklists that the developers can actually meet, but do so on the basis of community agency.

这对 Democrats 和 Republicans 都适用。

And that goes for Democrats and Republicans.

Ohio 州长候选人 Vivek Ramaswamy 在本月初向 Ohio 居民承诺，如果他们的社区里建了 data center，那么他们将不再需要支付自家的电费，他们会缴纳更低的房产税，而且 data center 必须遵守所有空气和水质标准。

Ohio candidate for governor Vivek Ramaswamy, at the beginning of the month, pledged to Ohioans that if a data center was built in their community, that they would no longer have to pay for their home's electricity, that they would pay lower property taxes, and that the data center would be required to abide by all air and water quality standards.

是的，我们需要建设更多 data center。

Yes, we need to build more data centers.

但如果我们真的能够走过这段充满争议的时期，并且在这样的基础上建设它们，让人们觉得自己的声音被听见了，也真正理解自己如何从中直接受益，那对未来来说，将是一个好到难以想象的基础，远远好过我们原本可能会拥有的那种基础。

But if we can actually come through this contentious period and build them on these sorts of foundations where people feel heard and they actually understand how they're benefiting directly from them, that is an unfathomably better foundation for the future than what we might've had.

我不会惊讶于这个问题在好转之前先变得更糟，至少在大众观感和媒体标题方面会是这样。

I wouldn't be surprised to see this issue get worse before it gets better, at least when it comes to popular perception and media headlines.

但是，说真的，当政治人物开始意识到，他们只要借鉴那些已经在其他社区有效、并且正在发挥作用的模板，就能带来所有这些好处，并且成为把这些点串起来的英雄，那时我认为我们就会看到情况向好的方向转变。

But boy, when politicians start realizing that they can deliver all these benefits and be the heroes that connect the dots by simply picking up on templates that have and are working for other communities, that's when I think we'll see a change for the better.

今天这期加长版的 AI Daily Brief 到这里就结束了。

For now, that's going to do it for this very extended episode of AI Daily Brief.

一如既往，感谢你的收听或收看，我们下次再见，peace。

Appreciate you listening or watching as always, and until next time, peace.

已剔除 2 处广告（点击展开查看）

M11:01广告 · 已剔除

All right, friends, quick announcements before we dive in.

First of all, thank you to today's sponsors, KPMG, Blitzy, Harbor, and HyperAgent.

To get an ad-free version of the show, go to patreon.com/AIDailyBrief, or you can subscribe on Apple Podcasts.

And to learn more about sponsoring the show, you can go to aidailybrief.ai.

There is a link for sponsor information.

There's also a link to other things going on in the ecosystem, such as our free upcoming webinar about how to put agentic loops to work for you as a knowledge worker that's coming up next Wednesday.

It is free.

You can register on the website.

Again, check it all out at aidailybrief.ai.

M118:56广告 · 已剔除

A new study from KPMG and the University of Texas at Austin found that when people work with AI, similar skills don't guarantee similar outcomes.

Researchers studied more than 500 early-career professionals and found that the best performers consistently amplified the value of AI by guiding, evaluating, and refining its outputs.

These top performers, called AI amplifiers, weren't defined by what they knew alone, but by how they worked with AI.

Learn more about what separates AI amplifiers from everyone else at kpmg.com/us/aiamplifiers.

Blitzy's understanding of massive codebases unlocks autonomous security fixes, modernization, and new features.

So what happens when there's no legacy code at all?

Greenfield is supposed to be the easy part.

Clean slate, no technical debt.

But even greenfield moves at human speed, one sprint at a time.

Blitzy changes the unit of work from the developer to the project, autonomously planning, building, testing, and validating entire applications from scratch.

Hundreds of thousands of lines of production-ready code.

One Blitzy customer stood up a brand new application, 534,000 lines of code, compressing a 65-week roadmap into 2 weeks.

Another shipped an entire application with no frontend engineer.

Legacy or greenfield, the answer is the same: software at the speed of compute.

Build what's next at blitzy.com.

That's B-L-I-T-Z-Y dot com.

Every episode, I talk about the competition between OpenAI, Anthropic, xAI, Google, and Meta.

And if you've been listening for a while, you might have a favorite.

Maybe you think OpenAI and Anthropic can stay ahead, or perhaps Meta's open-source strategy can win out.

Whatever your view, every AI lab creates a different investment opportunity.

Harbor Capital Advisors' AI Lab Ecosystem ETF suite lets you invest in the ecosystem behind the AI lab you believe in.

Search Harbor AI Lab Ecosystem ETFs wherever you invest or follow @HarborCapital on X to learn more.

Visit harborcapital.com for a prospectus containing investment objectives, risks, fees, expenses, and other important information.

Read and consider it carefully before investing.

Risks include principal loss and artificial intelligence-related risks.

Harbor ETFs are distributed by Foreside Fund Services, LLC.

Harbor is not affiliated with AI Daily Brief, and the funds are not affiliated with, sponsored by, or endorsed by any AI lab.

This is a paid advertisement and not personalized investment advice.

Investing involves risk, including possible loss of principal.

This episode of the AI Daily Brief is brought to you by HyperAgent, where you run fleets of agents your team can manage together.

New users get $1,000 in inference.

Forget local agents and chat workflows waiting on your laptop to be prompted.

HyperAgent deploys always-on agents in the cloud, doing real work across the tools your team already uses.

Marketing's agent turns competitor moves into landing pages.

Sales' agent enriches leads, drafts emails, and updates the CRM.

Ops' agent chases the paperwork and tracks the budget.

Every agent has access to shared context and follows your rules about scope and approvals.

It's time you add agents that feel like teammates.

Hire yours at HyperAgent, built by the team at Airtable.

Claim your $1,000 in inference at hyperagent.com/AIDailyBrief.