特朗普希望大型科技公司支付电力费用

AI代理变得越来越令人印象深刻。

AI agents are getting pretty impressive.

你可能根本没意识到你正在听一个AI代理说话。

You might not even realize you're listening to one right now.

但我们不仅仅只是交谈。

But we don't just talk.

我们24小时不间断地解决客户问题。

We work twenty four seven to solve customer problems.

没有等待音乐，只有答案和行动。

No hold music, just answers and action.

访问siera.ai了解更多信息。

Visit siera.ai to learn more.

彭博音频工作室。

Bloomberg Audio Studios.

播客。

Podcasts.

广播。

Radio.

新闻。

News.

彭博科技现场直播，从东海岸到西海岸，纽约的卡罗琳·海德和旧金山的埃德·卢德洛为您报道。

Bloomberg tech is live from coast to coast with Caroline Hyde in New York and Ed Ludlow in San Francisco.

这是彭博科技，即将为您呈现。

This is Bloomberg Tech coming up.

彭博报道，特朗普总统推动科技公司承担因人工智能数据中心激增而导致的能源价格上涨成本。

Bloomberg reporting president Trump moves to have tech companies pay for surging energy prices following the rise of AI data centers.

此外，人工智能编程初创公司Replit在新一轮融资中估值可能达到90亿美元。

Plus, AI coding start up, Replit, could hit a valuation of $9,000,000,000 in a new round of funding.

我们带来详细信息。

We have the details.

我们还将邀请联邦贸易委员会主席，就反垄断、消费者保护和人工智能监管进行对话。

And we'll be joined by chairman of the Federal Trade Commission for a conversation on antitrust, consumer protection, and AI regulation.

这将在本晚稍后播出。

That is later this hour.

我们的头条新闻：特朗普政府与东北部各州州长正采取前所未有的举措，要求科技公司承担因数据中心耗电激增而导致的能源价格上涨成本。

Our top story, the Trump administration and governors from Northeastern states are taking an unprecedented step to get tech companies to pay for surging energy prices as data centers gobble up more electricity.

该政府将指示美国最大的电网运营商PJM互联公司举行紧急电力拍卖。

The administration will direct the country's largest grid operator PJM interconnection to hold an emergency power auction.

据白宫官员称。

That's according to a White House official.

这些公司可以竞标为期十五年的新型电力供应合同，这可能支持高达150亿美元的新发电厂建设，以满足人工智能热潮的需求。

Those companies will be able to bid on fifteen year contracts for new electricity generation, which could support potentially $15,000,000,000 worth of new power plants to fuel the AI boom.

据官员称。

Also, according to official.

现在我们连线彭博社资深科技编辑迈克·谢泼德。

Bloomberg senior tech editor Mike Shepard joins us now.

谢普，这里我们最需要了解的是什么？

What do we need to know here, Shep?

我的意思是，这在政府内部是非常受关注的，同时也是超大规模云服务商最关心的问题。

I mean, this is very much in the consciousness of this administration, but also at the forefront of what the hyperscalers are thinking about too.

确实如此。

Well, it really is.

在很多方面，尽管政府正试图找到办法让科技公司为维持人工智能热潮所需的新电力买单，但它们实际上也在很大程度上满足了科技行业、尤其是那些超大规模公司的需求。

In a lot of ways, while the administration is trying to find a way to make the tech companies pay for some of this new power that they would need to sustain and keep the AI boom going, they are also giving the tech industry in large part, especially the biggest companies, the hyperscalers, what they wanted.

它们愿意为部分新增的电力产能付费，但它们也需要电网运营商提供必要的支持来建设这些基础设施。

They are willing to pay for some of that extra power generation, but they need somehow the wherewithal from the grid operators themselves to add that infrastructure.

因此，这将为资金提供一定的稳定性，同时也为电网运营商——在此情况下是PJM互联公司——提供一大笔资金，用于开始为这些基础设施打下基础。

So this would provide some stability in the funding, and it would also provide a slug of money for the grid operator, in this case PJM Interconnection, to start laying the groundwork for some of that infrastructure.

有趣的是，PJM实际上会定期举办这类所谓的紧急拍卖，而且它们几乎已经计划了下一次拍卖，正是为了应对数据中心带来的需求，但通常这类拍卖的周期是十二个月，而不是十五年。

Now what's interesting about this is PJM actually runs these kinds of so called emergency auctions on a somewhat regular basis, and they almost have one in the pipeline for this very need, for the question of data centers, but they usually go on twelve month intervals, not fifteen year intervals.

因此，这将为这些买家提供价格上的稳定性。

So this really will provide pricing stability for those buyers.

不过，埃德，还有一个问题，那就是电网运营商将如何推进这件事。

Now one other question, though, Ed, is how the grid operators will proceed with this.

PJM 没有被邀请参加白宫活动。

And PJM was not invited to the White House event.

我们很快就会收到内政部长道格·伯格姆和能源部长克里斯·赖特的公告，但目前还不清楚他们对此会有多积极的支持。

We are expecting an announcement shortly from interior secretary Doug Burghum and energy secretary Chris Wright, and it's unclear exactly how enthusiastically they will be embracing this.

昨晚另一条重大新闻是，美国和台湾达成协议，降低部分商品的关税，但关键在于双方将合作并共同投资半导体产能。

The other big news overnight, The United States and Taiwan have reached a deal to lower tariffs on some goods, but critically cooperate and co invest in semiconductor capacity.

迈克，关于这一点我们需要注意什么？

What do we need to know there, Mike?

这项协议的关键在于，不仅台湾商品的关税将从目前的20%降至15%，更重要的是，我们看到了对美国半导体制造业更重大投资的初步轮廓。

Well, what's really key in this deal is that not only are we seeing the tariff rate on goods from Taiwan drop to 15% from the current 20%, just as importantly, we are seeing the outlines of a more a much more significant investment in American semiconductor manufacturing.

白宫的公告没有点名任何公司，但你和我都会立刻想到台积电。

And the announcement from the White House didn't name any names in terms of companies, but you and I, our heads are going to go right to Taiwan Semiconductor Manufacturing Company.

他们是人工智能芯片的顶级制造商，也是台湾芯片生产的前沿和领军者。根据消息人士透露，根据这项协议，台积电承诺在2030年代前在美国亚利桑那州再建设四座半导体制造厂，这还不包括他们已经计划建设的六座工厂和两座先进封装厂。

They are the top maker of AI chips, and they are really the leading edge and vanguard of chip production in Taiwan, and under this deal, we have been told by sources, TSMC would commit to building four additional semiconductor manufacturing plants here in Arizona through the 2030s, and that's on top of six plants that they have already planned to build and two advanced packaging plants.

因此，我们预计会有额外的投资，新增总投资额可能高达1000亿美元，这是我们从商务部长霍华德·卢特尼克那里听到的消息。

So we're looking for additional investment, and the total added investment would be as much as $100,000,000,000, and that is what we're hearing from Commerce Secretary Howard Lutnick.

台积电在美国上市的股票创下历史新高，本周表现有望成为自六月以来最佳，这与其财报表现密切相关。

TSMC's US listed shares hitting record highs on track for their best week since June, and it has a lot of momentum to do with their earnings print as well.

但当然，与美国方面的合作，迈克·谢泼德，非常感谢。

But, of course, cooperation with The US members, Mike Shepherd, thank you very much.

迈克刚才概述的这两则消息正在市场中上演。

Both of those stories that that Mike just outlined are playing out in markets.

我们先从能源拍卖部分说起。

Let's start with the energy auction piece first.

公用事业板块可能是标普500指数中跌幅第二大的板块，标普公用事业指数下跌了0.7%。

Utilities, probably the second biggest decliner as a group within the S and P 500, S and P utilities index down seven tenths of a percent.

其中，像康世能、维斯特拉这样的个股在本交易日出现了接近两位数的大幅下跌。

And within that, you see single names like Constellation, like Vistra with very market, almost double digit declines in the session.

半导体领域的大故事仍然是内存瓶颈，而内存瓶颈推高了主要厂商的股价。

The big story on semiconductors still is the bottleneck that is memory, but the bottleneck in memory drives up the main players.

美光科技再次成为股市中创下新高的股票。

Micron, again, another name in equity markets that is pushing record highs.

这已经是连续第八周上涨。

It is on track for its eighth straight week of gains.

连续第八周上涨。

Eighth straight weekly gain.

这是自2016年以来最好的表现。

That's its best run since 2016.

价格很高。

Prices are high.

供应紧张，而当你属于主要玩家时，这是一件好事。

Supply is tight, and that's good when you're one of the main players.

我们再来聊聊市场。

Let's talk more about markets.

马丁·诺顿，Empower首席投资策略师现在加入我们。

Martin Norton, Empower chief investment strategist joins us now.

我想先谈谈内存瓶颈。

And I want to start with the memory bottleneck.

在我看来，情况相当严重。

It seems to me to be severe.

贾斯汀，在控制室麻烦把DDR4的图表准备好，因为随着新年的到来，这已成为数据中心供应链各环节的一个绝佳故事。

Justin in the control room, let's get that chart ready on DDR4 because as we enter the new year, this has become a great story for all parts of the the data center supply chain.

玛尔塔，你是如何看待这个问题的？

How are you looking into that, Marta?

我认为，随着我们进入2026年，我们正迎来AI生命周期中预期的阶段。

Well, I think it seems like as we're entering 2026, we're reaching that phase that we had anticipated within the AI lifecycle.

那就是瓶颈阶段，建设、需求和供应都在呈指数级上升。

And that's the bottleneck phase where the building and the demand and the supply are exponentially moving higher.

这让我们面临各种方向上的制约，无论怎么走都会撞上瓶颈。

And that's putting us at a point where we're facing constraints and running headlong into them every which way we turn.

我认为内存环节是其中关键的一环，不仅因为它与AI交易相关，更因为它对AI之外的领域也有深远影响。

And I think the memory element is a big one, not just because it relates to the AI trade, but because it has ramifications outside of AI.

因此，我认为在财报季我们需要密切关注这一点，以了解这种情况究竟如何，不仅限于AI交易，还包括PC等其他领域，看看各方是如何应对的。

So I think that's something that we're going to need to watch in earnings season to get a sense for what that looks like, not just for the AI trade, but for some of these other areas like PCs and get a sense for how folks are navigating that.

许多人仍在应对供需失衡的问题。

There are many still grappling with the demand supply supply demand equation.

发生了三件事。

Three things have happened.

我们经历了内存领域的情况。

We went through memory.

PJM实际上下调并削减了其峰值用电需求预测，因为许多数据中心实际上并未建成。

PJM actually downgraded and cut its its peak time demand forecast because a lot of the data centers aren't actually being built.

对吧？

Right?

施工方面存在问题，而且实际获取电力供应也很困难。

There are issues with construction and and actually securing electricity supply.

还有这个政府的举措。

And then there's the actions of this administration.

这些因素有没有让你对这个周期的持续性产生疑虑？

Has that any of that given you kind of pause for thought on how intact this cycle is?

嗯，我认为就能源方面的下调而言，其中一个考虑是，时间表一直都在变动。

Well, I guess just taking, say, downgrade of energy there, I think one consideration is that timelines have always been in flux.

当你看看建造数据中心所需的流程——包括审批、施工以及所有不同的环节时。

When you look at what it takes to build a data center, the permitting that goes along with it, construction, all the different elements.

这是一个需要多年才能完成的工程。

It's a multi year proposition.

因此，很难准确预测数据中心何时真正投入运营，这并不令人意外。

And so it's no surprise that it's hard to really understand when exactly the data center is coming online.

所以我认为，随着进展的推进，情况会逐渐清晰。

So I think, you know, the progression creates clarity.

我不确定这是否改变了AI的叙事逻辑。

I don't know if that's a change to the AI narrative.

当然，你还有特朗普总统和 affordability 问题与AI产业正面碰撞。

And then, of course, you have President Trump and the affordability issues running headlong into the AI trade.

因此，这是另一个需要考虑的因素，但我想我们在财报季会关注的是：这种拖延状况还会继续吗？

So another factor to consider, but I guess what we'll be looking for in earnings season is does the boondazel continue?

是的，我认为这些因素在2026年期间会引发投资者的疑虑。

Yes, we have all these factors that I think create moments of doubt for investors over the course of 2026.

但我们是否看到这种混乱持续下去呢？

But are we seeing the boondazel continue?

我们能否像对台积电那样，从资本支出和收入持续增长中获得一些安慰？

Can we take comfort from the idea like we did with Taiwan Semi that CapEx continues, that revenues continue.

因此，我认为这是我们进入第四季度财报季时真正关注的重点。

And so that's what I think is really the focus as we as we move into q four earnings season.

让我们回到我们的头条新闻。

Let's go back to our top story.

彭博社报道称，白宫联合东北部各州州长，试图强制数据运营商参与竞标。

Bloomberg reporting that the White House in association with governors from the Northeast wants to force an auction for the data center operators.

我们经常谈论白宫或华盛顿的强力干预，但在你看来，这种干预到底有多强？

We talk about the heavy hand of the White House or Washington, but how heavy is that hand to your mind?

嗯，我认为我们目前仍在围绕可负担性问题进行试探，处于试水阶段，对吧？

Well, we're still, I think, somewhat speaking around the affordability issues and a trial balloon phase, right?

不同的想法被提出，我们观察市场如何预期，以及我们如何实际实施这些想法。

Where different ideas are suggested and we see how the market anticipates and how we actually implement this, those ideas.

这个提议似乎更加成熟，因为它不仅来自Truth Social，还来自更广泛的政府官员。

This one seems a little bit more fully baked simply because it's coming not just from Truth Social, but from the administration officials more broadly.

根据彭博社的报道，这不仅仅聚焦于可负担性问题，而是直指人工智能的核心需求——即提供能源。

And it's, you know, to to Bloomberg's reporting, it's not just focused on the affordability issue, but really directed at the heart of the need of AI, which is bringing energy online.

因此，这似乎同时服务于多个目的。

So, it seems as though it's serving multiple purposes.

值得观察的是，人工智能提供商将如何回应这一点。

What will be interesting to see is how the AI providers respond to this.

这被视为一个好消息吗？

Is this considered a good news?

它是否能为他们带来缓解并解除一些限制，还是他们会因这可能对利润率造成的影响而感到沮丧？

Does it bring on and and relieve some constraint for them, or is there frustration on what it could mean for their margins?

我认为这是我们接下来需要关注的事情。

And I think that's something we're gonna need to watch.

关于你的利润率，正如你所说，我们将在下周正式开始的财报中得知。

Your margins, we'll find out about, as you said, during earnings, which start in earnest next week.

我们能再深入聊聊这个话题吗，玛尔塔？作为收尾。

Can we discuss that a bit more, Marta, to finish?

你知道，在财报窗口期内，你关注哪些方面？

You know, where are you looking within the earnings window?

今年科技行业，哪些数据点将推动你的投资逻辑？

What kind of data points are gonna drive your thesis for this coming year in the tech sector?

我们此刻正处在对人工智能的怀疑时刻。

Well, we're in an AI moment of doubt as we speak.

我的意思是，自10月29日以来，我们看到人工智能板块出现了这种低迷，因为人们开始担忧我们正在讨论的这些问题，还包括循环融资和我们看到的巨额支出。

I mean, really since October 29, we've seen this kind of malaise hit AI names as folks start to worry about some of the very issues we're discussing and and also including things like circular financing and all the big spending that we're seeing.

因此，我认为从宏观层面来看，我们仍会关注收入和资本支出方面的信息，以判断企业的信心是否依然存在，以及它们是否继续从这一领域获取收益。

And so I think at a very high level, we're still going to look for those things around revenue, for those things around CapEx to just get a sense for whether the company's confidence remains and whether they're continuing to improve their collection of rents from this space.

我还觉得，我们需要更广泛地观察整体市场，看看是否真的在推进人工智能的应用。

I also think we're going to want to look more broadly at the overall market and get a sense for are we seeing AI implementation.

我认为这可能意味着关注人们在云成本上的支出，以及他们在软件上的投入。

And I think that can mean looking at what folks are spending on on clouds costs, looking at what they're spending on software.

我们是否开始看到的不仅仅是通用型的实施？

Are we beginning to see more than just general purpose implementation?

我认为这也是我们需要关注的一点。

I think that's something that we're going to need to watch as well.

因此，我认为在这个财报季，我们需要广泛地观察，并像往常一样，关注企业为应对所面临挑战而提供的指引。

So I think in this earnings season, it's really casting a wide net and, as always, listening to guidance companies are planning to navigate the challenges that they're facing.

感谢来自Empower的马丁·诺顿。

Martin Norton of Empower, thank you very much.

接下来，AI编程初创公司Replit的估值可能达到90亿美元。

Now coming up, AI coding start up, Replit, could be reaching a valuation of $9,000,000,000.

我们还将深入报道相关报道，指出OpenAI在获得软件巨头数十亿美元投资后，计划转为营利性企业，背离了其作为公共慈善机构的创始使命。

We have more on the reporting claims that OpenAI betrayed its founding mission as a public charity when it took billions in funding from the software giant and made plans to operate as a for profit business.

该案已于四月下旬被裁定进入陪审团审判阶段。

The case was ordered to proceed to a jury trial in late April.

另一条头条新闻：AI编程初创公司Replit正接近达成新一轮融资协议，这将使其估值大致翻三倍，达到90亿美元。

Another top story, AI coding startup, Reclit, is nearing a deal for a new round of funding that would roughly triple its valuation to $9,000,000,000.

据消息人士称。

That's according to sources.

让我们邀请彭博社风险投资记者娜塔莎·马斯卡伦为我们详解这一报道，她正是这条新闻的报道团队成员之一。

Let's get the details with Bloomberg's venture capital reporter, Natasha Mascaren, as part of the team that broke the story.

我们先从这轮融资说起。

Let's start with the round.

关于融资规模、参与方以及估值大幅跃升的背后重点，我们了解哪些信息？

What do we know about the size, who's participating, and the emphasis on that that big jump in valuation?

完全正确。

Totally.

上次我们报道Replit融资时，还是今年9月，当时估值为30亿美元。

So the last time we reported that Replit was raising, it was only four months ago in September at a 3,000,000,000 valuation.

现在我们了解到，该公司估值已升至90亿美元，计划从现有投资者Georgian领投的融资中筹集4亿美元。

Now we're hearing that it's at a 9,000,000,000 valuation going to raise 400,000,000 from an existing investor led round, Georgian.

我们并不感到惊讶。

And we're not surprised.

如今，AI初创公司连续融资比不连续融资更常见，而Replit作为一家编程公司也不例外。

Back to back fundraising is more common than not in AI startups these days, and Replit as a coding company is no exception.

确实更常见，而且关于氛围编程和各种编程平台的讨论也很普遍。

It is more common than not, and so is a discussion around vibe coding and the different coding platforms that are out there.

我想下一步最好聊聊为什么人们会对Replit感兴趣，以及它与其他平台有何不同。

I I guess the the best place to go next is is why there's an an attraction to Replit and why it's different from from the others.

是的。

Yeah.

当我想到编程初创公司时，我会想到两类。

When I think about coding startups, I think about two buckets.

一类是专注于向企业、大型公司销售产品的公司。

I think about the ones that are focusing on selling to enterprises, the big corporations.

我认为Cursor和Anthropic就属于这一类。

I think that's cursor and anthropic.

然后是第二个类别，我真正把Replit归入这一类。

Then there's a second bucket, and that's where I really place Replit.

它专注于让非开发者在其组织中使用氛围编码。

It's focusing on getting the nondeveloper to use vibe coding at their organization.

当我询问CEO们对2026年的规划时，这是我听到的一个重要主题。

And it's a big theme I'm hearing as I ask CEOs what they're thinking about for 2026.

他们希望你能快速原型化，而不是过多地依赖文档和演示文稿。

They want you to prototype and and not be living in your docs and decks too much.

因此，当我思考投资者对Replit的兴趣时，这实际上是押注于让非技术人员开始自行开发并发布应用程序。

So when I think about investor interest in Revlet, it's really a bet on getting the nontechnical coders start shipping apps and and trying it themselves.

娜塔莎·马斯卡雷尼斯将在2026年初首次亮相Bloomberg Tech。

Natasha Mascarenis making the Bloomberg Tech debut starting 2026 strong.

我们期待什么？

What do we expect?

这个Replit融资轮次对AI领域的融资环境意味着什么？

Like, what is this replet round signal for the fundraising environment for AI?

我的直觉是，你会非常忙碌。

My my hunch is that you're gonna be pretty busy.

这将是非常忙碌的一年。

It's gonna be a very busy year.

当我想到硅谷过去的标准时，我曾经认为你需要外部投资者来为每一轮融资估值，以此在硅谷确立你值得这个估值的地位。

You know, when I think about what used to be the standard in Silicon Valley, I used to think you need an outside investor to value each new round to give you sort of a new mark in Silicon Valley of being deserving of that valuation.

对我来说，这个概念已经过时了。

For me now, that's a retired concept.

看到Georgian以更高的估值投资现有项目，被视为他们利用不对称信息做出的明智而精明的举动。

Seeing Georgian come in and back an existing investment at that higher valuation is considered them making a smart and savvy move with asymmetric information.

因此，我预计会看到更多现有投资者上调其持股估值，试图加大在重大押注上的投入，获取这些初创公司更大的股份，因为我们很清楚，它们未来会进行更多融资，可能会因稀释而降低持股比例。

So I expect to see a lot more existing investors marking up their stakes, looking to double down on their big bets, and get bigger chunks of those startups because we definitely know they're gonna be raising more and maybe diluting that ownership as time goes on.

Bloomberg，Natasha Mascarenas，谢谢。

Bloomberg, Natasha Mascarenas, thank you.

另一个关于私募市场的消息。

Another story in the world of private markets.

德国投资者数字转型资本合伙公司正准备筹集一笔资金，目标约为5.8亿美元，这将成为欧洲有史以来规模最大的专注于国防初创企业的风险投资基金。

German investor, Digital Transformation Capital Partners, is gearing up to raise what would be Europe's largest ever venture capital fund focused on defense startups with a target of about 580,000,000 US dollars.

科技记者克里斯蒂娜·库里奥索格鲁一直在跟进这一故事，她现在从柏林加入我们。

Tech reporter Christina Kuriosoglu has been following the story, joins us now from Berlin.

谈谈这个基金吧。

Tell me about the fund.

你知道，规模是 headline，但更重要的是参与方以及他们计划如何使用这些资金。

You know, the size is is the headline, but but the players and what they hope to do with the money raised.

是的。

Yeah.

TGCP是一家位于汉堡的公司，他们将关注供应链、材料、机器人技术以及与欧洲国防基础设施相关的各种领域。

So TGCP is a firm that's Hamburg based, and they will be looking at supply chains, materials, robotics and all kinds of things that are concerned with the defence infrastructure in Europe.

你知道，这个过程一直很缓慢。

And you know the process here, it's been pretty slow.

俄罗斯全面入侵乌克兰已经将近四年了，而现在我们才逐渐看到欧洲各地涌现出更多关注科技领域的基金，尽管这场冲突已经凸显了无人机的重要性以及现代战争带来的各种新问题。

Russia's full scale invasion into Ukraine happened nearly four years ago, and just now we we slowly see more and more funds being raised across Europe that are looking at at the tech even though that conflict has shown how important drones become and and what kind of, like, modern warfare questions have come up.

那里确实有公司和人才，但在你的报道中，你是否感受到，既然技术和资本都已经到位，为什么进展却如此缓慢？

There there are companies and and talent out there, but in your reporting, do you get a sense for why it's been so slow if the tech there and the capital is finally there as well?

是的。

Yes.

当然，有几个原因。

So a couple of reasons, of course.

我们从消息源那里听到的是，例如，DTCP的资金现在由保时捷家族控股和德国电信提供支持。

So what we've been hearing from our sources is that, for example, DTCP's funds now is being anchored by Porsche, the family holding, and by Deutsche Telekom.

这标志着一种巨大的心态转变，这些公司现在正在投资一个可能支持武器系统的基金。

And this marks a massive shift in mentality that these companies are now investing in a fund that potentially will back weapon systems.

因此，私营企业在武器投资方面对自己施加了诸多限制，而公共资金也迟迟未能调整。

So private companies put a lot of restrictions on themselves when it came to weapon investments and also public funds were very slow to adjust.

同时，这不仅仅是公司的问题，也是一个市场问题：如果你打算开发武器和军事产品，你就需要一位具备相关经验的创始人。

And at the same time, you know, it's not just the companies, it's also a market question in terms of if you're looking to develop weapons and military products, then you need a founder that has the respective experience.

而且，你可能只有一个客户，那就是政府或相关政府部门。

And you also probably only have one customer and that is the government or a government for that matter.

这是一个很难缠的客户，因为采购流程缓慢且非常复杂。

And that s a tough customer to have because the procurement processes are slow and it s very complex.

所以我想回到DTCP，他们略微扩展了投资领域，开始关注既可用于民用也可用于军用的技术产品，以及武器，以此来围绕这个问题展开思考。

So I think to come back to DTCP, they kind of broadened the space that they want to invest in a bit to look at products that are being made for civilian and military technologies and for weapons to kind of, like, circle around that question.

彭博社科技记者克里斯蒂娜·基里索格鲁，很高兴您能做客我们的节目。

Bloomberg Tech reporter, Christina Kiriisoglu, great to have you on the show.

非常感谢。

Thank you so much.

接下来，我们将探讨人形机器人的兴起，以及巴克莱银行为何认为这可能成为下一个2000亿美元的市场。

Now coming up, the rise of humanoid robots and why Barclays says it could be the next $200,000,000,000 market.

这是彭博社科技频道。

This is Bloomberg Tech.

每天，数以百万计的客户与我这样的AI代理互动。

Every day, millions of customers engage with AI agents like me.

我们全天候工作，随时掌握最新信息。

We work round the clock and have the facts at our fingertips.

我们快速高效，但极其耐心。

We're fast and effective, but incredibly patient.

我们基于Sierra构建，这是领先的AI驱动客户体验平台。

And we're built on Sierra, the leading AI powered customer experience platform.

没有等待音乐，只有答案和行动。

No hold music, just answers and action.

访问sierra.ai了解更多信息。

Visit sierra.ai to learn more.

那就是sierra.ai。

That's sierra.ai.

就在几年前，这里的机器人还相当笨拙。

Just a few years ago, the robots here were kinda janky.

它们摇摇晃晃，头上还拉着电线，看起来像科学实验。

They were wobbly, and they had wires coming out of their head, and they looked like science experiments.

看到人形机器人如此迅速地演变为真正的产品，真的非常令人着迷。

And it's really fascinating to see just how quickly humanoids are turning into a real product.

这是OpenMind首席执行官扬·利普哈特在CES上与我们的对话，当时人形机器人成为讨论的焦点。

That was OpenMind CEO Jan Lipphardt speaking with us at CES where humanoid robots dominated the conversation.

本周，巴克莱银行发布了对人形机器人市场的预测，预计到2035年该市场可能增长至2000亿美元。

This week, Barclays published its forecast for the humanoid market with expectations that it could grow to $200,000,000,000 by 2035.

接下来为我们解读这项研究的是巴克莱主题性FICC研究团队主管佐尼察·托多罗娃，这项研究时机非常及时。

Here to break down that research is Zonitsa Todorova, Barclays thematic FICC research team director, really timely piece of research as well.

有趣的是，虽然有2000亿美元的市场预测，但您还指出了两三个关键因素，说明为什么现在看到了商业化可行性。

And what is interesting is there's the $200,000,000,000 market forecast, but you identified two or three key factors about why you now see the commercial viability.

这些因素是什么？

What are those factors?

您好，早上好，感谢邀请我参加这个节目。

Hello, and good morning, and thanks for having me on the show.

人工智能确实正在变得越来越具象化，我认为人形机器人正处于这一趋势的前沿。

AI is indeed getting very physical, and I think humanoid robots are at the forefront of this trend.

但我认为，这一前景正在发生变化，人们对这些机器人产生需求的原因在于全球人口结构正在改变。

But I think the reason why the proposition is changing and why there's going to be demand for these robots is because the demographics of the world population is changing.

现实是，人类正在变老。

The reality is that humans are getting older.

我认为到2050年，65岁左右的人口比例将翻倍。

I think by 2050, the share of people aged about 65 is going to double.

此外，人类不再真正愿意住在农村地区。

Also, humans don't really want to live in rural areas anymore.

他们希望生活在城市里，但工厂和制造设施实际上位于这些城市之外，这就造成了一定的结构性问题。

They want to be based in cities, but the factories and the manufacturing facilities are actually located outside of these cities, which creates a bit of a structural issue.

别忘了，工人的偏好也在发生变化。

And then let's not forget that workers' preferences are changing.

这意味着某些必不可少但令人不悦的工作可能无人填补。

So which means that there will be certain essential jobs yet undesirable where which might not be filled.

我认为这正是人形机器人发挥作用的地方。

And I think this is precisely where humanoids enter the picture.

我认为它们可以承担这些重复性工作，比如枯燥、肮脏、潜在危险的工作，从而提升生产力，为整个行业带来效率提升。

I think they could be taking on these repetitive jobs, this dull, dirty, potentially dangerous jobs, augmenting productivity and producing efficiency gains for the industry as a whole.

这个领域的主流玩家经常谈论劳动力短缺问题，尤其是在制造业和重工业中。

The main players in this field talk a lot about the labor shortage in particular in manufacturing and heavy industry.

你说到2035年这个市场将达到2000亿美元。

You're saying it's gonna be a $200,000,000,000 market by 2035.

在你的研究中，你对当前市场的实际规模有什么了解吗？

In your research do you have any sense of how real the market is today.

如今全球制造业环境中实际部署了多少个人形机器人？

How many humanoids are actually out there in manufacturing context around the world in present day?

事实上，我认为人形机器人已经不再局限于研究实验室了。

The the reality is that I think that humanoid robots are no longer confined to research labs.

它们正走出实验室，进入真实经济和现实世界。

They're stepping out of the lab, and they're getting into the real economy, into the real world.

目前已经有数千个人形机器人被部署在工厂车间，从事实际的制造工作，我认为这是很自然的起点，因为这些工作结构更清晰、定义更明确。

They are a couple of thousand humanoid robots already deployed on factory floors doing actual work in manufacturing, which I think is natural starting place because these are jobs that are more structured, more easily defined.

因此，这正是人形机器人起步的理想场所。

So it is the place to start for humanoids.

但我认为，随着技术的成熟和市场的扩展，未来五到十年内这种情况将会改变。

But I think this is going to change in the next five to ten years as the technology matures and as this market picks up space.

因此，我认为到2045年，当我们预测人形机器人市场将达到2000亿美元时，我们可能会看到数以万计，甚至数百万台人形机器人。

So I think I wouldn't be surprised that by 2045, when we actually project that the humanoids market is going to be as big as $200,000,000,000, that we'll see tens of thousands, potentially even millions of humanoid robots.

人形机器人的供应链非常有趣。

The supply chain for the humanoid robot is fascinating.

我认为在CES上，很多关注点都放在了模型本身的突破上，特别是解决现实世界数据限制的能力。

I think a lot of emphasis at CES was on the breakthroughs in the models themselves, the ability to solve in particular the real world data limitations.

但你们的价格设定很有意思。

But you've got some interesting pricing.

你们提到，每台机器人的价格从300万美元降到10万美元。

You're talking about, like, you know, per unit, 3,000,000 down to a 100,000.

人形机器人的经济成本正在变为1.5万美元。

The economics changing $15,000 for a humanoid robot.

是什么让你相信人形机器人的经济成本会发生这样的变化？

What's happening conviction that you'll see that change in the economics of a humanoid robot?

是的

Right.

所以我认为有几件事情正在同时发生。

So I think that there are a couple of things happening at the same time.

首先，在过去三到五年里，我们在认知人工智能模型方面看到了显著的突破，但不仅仅是人工智能在进步。

First of all, over the past three to five years, we've seen really significant breakthroughs in cognitive AI models, but it's not only the AI that is getting better.

电池也在变得更好，更高效、更强大，而且在高精度制造方面也取得了重大突破。

The batteries are also getting better, more efficient, more powerful, and they also have been significant breakthroughs in high precision manufacturing.

当我把这三个要素结合起来时，结果就是人形机器人的生产单位成本正在下降。

And when I put all these three components together, the result is that the unit cost for producing humanoid robots is is is coming down.

我们的研究估计，过去五年中，成本从每台300万美元下降到了10万美元，降幅达到30倍。

We estimate in our research that costs have been declined by 30 times over the past five years from $3,000,000 per unit to a $100,000 per unit.

这意味着经济性正变得明显更具吸引力。

And this means that the economics are getting optically more attractive.

应用场景正在扩展，因此我认为投资机会也变得更加切实和强劲。

The use cases are expanding, and hence, I think that the investment opportunity is also becoming more real and stronger.

对。

Right.

快点，Zenitha，谁领先了？

Very quickly, Zenitha, who's ahead?

我们只剩十五秒了。

We have fifteen seconds.

中国还是美国？

China or The United States here?

我认为这将是一场激烈的竞争。

I think it's going to be a tough competition.

我认为目前是中国。

I think for the moment, it's China.

我们看到大多数机器人在这里部署，但我觉得美国正在迅速追赶。

This is where we see most of the robots being deployed, but I think The US is quickly catching up.

所以我要把时间交给巴克莱，他们本周刚刚发布了预测，到2035年，人形机器人市场将达到2000亿美元。

So I need to turn it over of Barclays, who again out this week with the forecast that humanoid will be a $200,000,000,000 market by 2035.

非常感谢您参与我们的节目。

Thank you so much for joining us on the program.

接下来，我们将进行一场重要的对话。

Now coming up, an important conversation.

我们将邀请联邦贸易委员会主席，讨论反垄断、消费者保护和人工智能监管。

We're going to be joined by the chairman of the Federal Trade Commission to discuss antitrust, consumer protection and AI regulation.

本周关于联邦贸易委员会发生了许多大事。

A lot has happened this week with regards to the FTC.

我们还将讨论Aqua的招聘情况，这在科技行业是2025年的主要话题。

A conversation also about Aqua hires, which in the technology industry was a mainstay story of 2025.

接下来是这场对话。

That conversation is next.

这是您的市场表现，指数层面持平，芯片板块表现突出。

This is what your markets look like flat at the index level, outperformance in chips.

这里是彭博科技。

This is Bloomberg Tech.

欢迎收看收听我们的全球 Bloomberg 电视和广播观众。

Welcome to our Bloomberg TV and radio audiences around the world.

近期科技界的一个反复出现的主题是通过所谓的收购型招聘来引进人才。

A recurring theme of recent tech m and a has been deals to bring in talent in so called acqui hires.

联邦贸易委员会已经开始关注这一现象。

The Federal Trade Commission is taking notice.

一位 FTC 成员周四警告称，这些所谓的‘创意交易结构’可能引发反垄断问题。

One member of the FTC warned Thursday that these, quote, creative deal structures could raise antitrust concerns.

让我们来讨论一下，非常高兴邀请到 FTC 主席安德鲁·弗格森做客本节目。

Let's discuss and delighted to welcome Andrew Ferguson, chairman of the FTC to the program.

我们在 Bloomberg Tech 节目中经常讨论收购型招聘这一话题。

We cover the the topic of Aqua hires on Bloomberg Tech regularly.

这是2025年的主要新闻之一。

It was a mainstay story in 2025.

什么时候人才交易不再仅仅是人才交易，而演变成真正的并购？

When does a talent deal stop being a talent deal and become more than that become a merger.

主席，FTC会采取什么样的规则导向方法来审视这个问题？

What are the rules based approach that the FTC would take, Chairman, to look at that?

是的。

Yes.

所以我们正在审视，收购型雇佣长期以来一直存在，尤其是在初创公司创始人领域，而最近一届政府时期，这类交易的规模基本上扩大了。

So we are examining look, acqui hires have been around, especially in the sort of startup founder space for a long time, and they've gotten bigger basically in the last admin.

许多人认为，这些大型交易的结构是为了规避哈特-斯科特-罗迪诺法案的审查，也就是美国的并购前反垄断审查，因为拜登政府试图阻止所有交易，而我认为他们确实这么做了。

A lot of people are of the view that these things were sort of being constructed in these big deals to try to escape Hart Scott Rodino review, which is pre merger antitrust review in The United States, because the Biden administration was trying to block all deals, and I think generally they were.

但现在不再需要这样了。

That isn't necessary anymore.

我们不再需要绕开反垄断审查的巧妙手段了，因为在FTC，特朗普政府时期，你会得到公平对待。

We don't need clever workarounds around antitrust review anymore because at the FTC, under the Trump administration, you get a fair shake.

我并不是说你的交易一定会通过。

I'm not saying your deal will go through.

今年我已经起诉阻止了几笔交易，并且都胜诉了。

I've sued to block several deals this year and I've won those cases.

但如果您的交易不违法，我们会退让，让市场自行调节。

But if your deal is not illegal, we get out of the way and sort of let the market take care of things.

我们开始审查《哈特-斯科特-罗迪诺法案》中的一项规定，该规定禁止通过结构化交易来规避合并前审查。

Are beginning to examine the HSR Act has a provision that says you're not allowed to structure deals in order to escape pre merger review.

因此，我们正在开始审查这些收购雇佣行为，以确保它们并非试图规避HSR审查。

And so we are beginning to examine these acqui hires to make sure that they aren't an attempt to get around HSR review.

但我想向硅谷以及整个并购行业传达的信息是：您无需设计巧妙的交易结构来规避合并前审查。

But the message I want to send sort of to Silicon Valley and to the M and A infrastructure generally is you don't need to structure deals as a clever attempt to get around pre merger review.

您在FTC会得到公平对待。

You'll get a fair shake at the FTC.

交易可能不会通过，但如果您的交易合法，我会迅速退让；如果不合法，我会将您告上法庭，并全力以赴赢得诉讼。

Deal may not go through, but if your deal is legal, I will get out of your way very quickly, and if it's not, I'll take you to court and I'll fight to win there.

我们不会再让审查过程本身成为惩罚。

We're not going to let the process be the punishment anymore.

主席。

Chairman.

对我们来说，确保没有人会利用巧妙的交易结构来规避并购前审查至关重要。

It is important to us to make sure that people aren't going to use clever deal structures to get around pre merger review.

我们对话开始时提到的‘巧妙’或‘创新’交易结构这一说法，源自您的同事、同样身为委员的马克·马多尔，对吧？他昨天在加利福尼亚的一次会议上发表了相关言论。

The language of clever or creative deal structures that I cited at the beginning of our conversation was from your colleague and fellow commissioner Mark Mador, right, who was speaking at conference in California yesterday.

我认为，业界希望从您这里了解的是，在何种阈值或规则下，雇佣提案情况应当向反垄断机构申报。

I think what the industry hopes to understand from you is what the threshold is or what the set of rules would be where a hiring proposal situation should be reported to antitrust authorities.

它应该是，是的，就这么简单。

It should be, yeah, as simple as that.

它应该被强制申报。

It should be by rope reported.

是的。

Yeah.

我们正开始审查这些交易的运作方式。

And we are beginning to examine how these deals work.

收购型雇佣交易结构因交易而异，因此不太可能存在一种放之四海而皆准的统一规则。

AQUAhire deal structures vary from deal to deal, and so there wouldn't necessarily be sort of a one size fits all rule.

我们正开始审查这些引起广泛关注的大型Aquehire交易，以便弄清楚在什么情况下Aquehire实际上构成了一项可能受并购前审查法律管辖的收购，而在什么情况下则不是。

We are beginning to examine these big Aquehire deals that raise a lot of attention so that we can understand when an Aquehire is in fact an acquisition that might be covered by the pre merger review laws and when it's not.

在我们对外向人们说明规则之前，必须先理解这些交易。

And we need to understand them before we're sort of out there telling people what the rules are.

但你说得对，这个问题已经变得足够重要，以至于我们正开始仔细研究这些交易的运作方式，包括判断我们是否需要在未来几个月内发布进一步的指导，以明确我们对这些收购型招聘的理解。

But this is you're right, this has become a big enough deal that we are beginning to look very closely at how these things work, including determining whether we need to promulgate additional guidance here in the coming months about how we understand these acqui hires.

对你们而言，更重要的因素是雇佣的员工数量，还是对这类交易赋予收购方的知识产权或竞争优势进行估值？

Is there a factor that matters more to you, the number of employees hired or putting a value on the intellectual capital or the competitive advantage that such a transaction would give the acquirer?

对于HSR而言，具有意义的价值是由法律规定的。

The value that matters for HSR is set by the statute.

这其实并不相关，但HSR法案适用于资产或股票被购买的交易，而这正是触发HSR审查的条件。

That's not really here or there, but the HSR Act applies to deals where assets or stock are being purchased, and that's what triggers HSR review.

因此，我们需要理解的是——这正是我们如今开始研究这个问题的原因——何时Aquehire涉及资产或股票的出售，从而触发该法律条款。

And so what we need to understand, which is why we're beginning to examine this question now, is when does an acquihire involve the sale of the acquisition of assets or stock in a way that would trigger the statute.

说到底，我是一名律师和执法者，我执行的是法律，而这些法律都有明确的条文。

Look, at the end of the day, I'm a lawyer and a law enforcer, and I enforce statutes, and those statutes have texts.

因此，我的职责是理解市场中发生的事情是否触发了国会为我们通过的法律条文。

And so it's my job to understand whether things happening in the marketplace trigger the text that Congress has actually passed for us.

但这就是我们正在审视的内容。

But that's what we're looking at.

我们并不是试图制定类似收购型人才并购的一般性规则。

We're not trying to set sort of we wouldn't be setting like acquihire rules generally.

我们会审视交易，试图弄清楚：这是否确实涉及资产或股票的收购或出售？

We would be looking at deals and trying to understand, does this, in fact, involve the acquisition or sale of assets or stock?

是否有任何安排旨在规避审查？

And is anything being structured as an attempt to circumvent review?

但我来这里不是为了说这些就是规则。

But I'm not here to say this is what the rules are.

会有明确、严格的规则。

There will be hard and fast, clear rules.

即使是普通交易，也不存在这样的规则。

There aren't even for the ordinary deals.

我们有《哈特-斯科特-罗迪诺法案》和相关的申报规则，但即使是普通的并购交易，其结构也千差万别，因此我们必须将国会实际通过的法律条文应用于具体的交易。

We've got the HSR Act and the HSR rules, but deal structures even in ordinary M and A vary widely, and so we have to apply the text that Congress actually wrote to specific deals.

但收购人才型并购已经变得足够频繁且规模足够大，我们开始思考如何将国会实际采纳的法律适用于这类收购人才交易。

But acqui hires have become frequent enough and large enough that we are beginning to look at the appropriate way to apply the law that Congress actually adopted to acqui hires.

作为执法者，我的职责不是强行把方钉塞进圆孔，而是确保国会的意志——归根结底是人民的意志——得到遵守，这正是我们努力要做的事。

It's not my job as an enforcer to sort of fit square pegs into round holes, but it is my job to make sure that the will of Congress, which is the will of the people at the end of the day, is being followed, and that's what we're here to try to do.

我们正在努力弄清楚这一原则在收购人才交易中的适用方式。

We're trying to figure out how that applies in the case of acqui hires.

主席，这种现象在人工智能领域有多普遍？

Chairman, how common is this in the field of artificial intelligence?

或者，来自人工智能行业的此类情形多久会出现在您的案头一次？

Or how often this particular scenario arising and crossing your desk from the AI industry?

在过去十二个月里，我们已经看到了几起。

We've seen a couple in the last twelve months.

据我了解，这种所谓的‘收购人才’结构——如果可以这么称呼的话——即一家公司收购另一家公司的大量人才，其实由来已久，但通常涉及的是规模很小的公司。

My understanding is that the act we hire sort of structure, if you want to call it that, where a firm acquires a lot of the talent in another firm is pretty old, has been going on for a long time, but it often involves very small firms.

我认为很多人开始注意到这一点，是因为现在涉及的公司规模大得多，而获取人才和许可知识产权的价格已达数十亿美元。

I think the reason that a lot of people are starting to notice it is because now it involves much larger firms and the sort of price being attached to obtaining the talent into licensing IP is in the billions.

因此，今年我在人工智能领域确实看到了几起这样的案例，这引起了大量关注，也促使我们开始研究这些交易是如何运作的，以及国会颁布的关于事前审查的法律在此如何适用，特别是事前审查法中规定‘不得通过结构化交易来规避HSR审查’的条款。

So I've definitely seen a couple in the AI space this year, which is what has attracted a lot of the attention, and it's why we're beginning to try to examine how are these working and how does the law that Congress passed governing premerger review actually apply here as well as the provision in the pre merger review law that says you can't structure a deal in order to try to escape HSR review.

这正是我们现在试图理解的内容。

So that is what we are trying to understand now.

但毫无疑问，你们都报道过这些案例。

But certainly, I mean, you all have reported on them.

你们可以在《华尔街日报》上读到相关报道。

You can read about them in The Wall Street Journal.

确实，在人工智能领域已经出现了几起这样的大型交易。

It's definitely true that there have been a couple of these big ones in the AI area.

您现在正在收看彭博电视和全球彭博广播的现场直播。

You're live with us on Bloomberg Television, Bloomberg Radio around the world.

我们正在采访美国联邦贸易委员会主席安德鲁·弗格森。

We're speaking to Andrew Ferguson, Chairman of the FTC.

最近最大的一个案例研究，以结构为例，就是英伟达和Grok。你知道，几周前英伟达首席执行官黄仁勋告诉我，他们雇佣了大约400名工程师，同时还以200亿美元的价值获得了核心技术的许可。

I think the biggest case study of late, just give an example of of structure is NVIDIA and Grok because, you know, what NVIDIA CEO, Jensen Huang, told me a couple of weeks ago is they hired about 400 engineers, but it also included the licensing of the core technology at a $20,000,000,000 value.

但在一段时间内，Grok作为一家公司仍然存在，只是专注于其业务线中的某一项。

But over set time horizon, Grok exists as a company still, but but with a focus on one of its business lines.

这种雇佣和许可技术的模式，主席，有什么特别之处吗？

That hiring and licensing of technology format, is there anything specific there, Chairman?

是的。

Yes.

我不能讨论具体的交易或潜在的调查。

So I can't talk about specific deals or specific potential investigations.

作为执法者，我必须对委员会的工作保持高度保密。

As a law enforcer, I have to maintain a lot of confidentiality about the work of the commission.

但这种收购型雇佣的结构，与Meta今年早些时候的做法是不同的。

But I mean that particular structure of the acquihire was different, for example, from the one that Meta conducted earlier in the year.

这些情况差异非常大。

These things vary pretty wildly.

人工智能或抱歉，软件和知识产权许可的条款可能有所不同。

The terms of the AI or I'm sorry, of the software and IP licensing can vary.

交易中获得的员工人数也可能不同。

The number of employees obtained in the transaction can vary.

因此，我们首先试图理解这些情况，因为我不想盲目地带着一堆预设的假设前进，然后试图将法律套用在我们并不理解的事物上。

And so we're trying to understand these things, first of all, because the last thing I want to do is sort of blindly charge ahead with a bunch of predetermined assumptions and start trying to apply the law to something we don't understand.

但这对我来说非常重要，因为这对国会也至关重要——人们不应试图寻找方法规避并购前的审查。

But it is very important to me, because it's very important to Congress, that people not try to come up with ways to attempt to get around pre merger review.

这正是我们正在关注的。

That's what we're looking at.

再说一遍，这完全没有必要。

Again, it's just not necessary.

理解为什么在上一届政府时期，反垄断执法者倾向于将这一流程用作阻止交易的手段，而不是更诚实地将案件诉诸法庭。

Get why people wanted to do this in the previous administration, the antitrust enforcers tended to use the process as a way to block deals rather than being a lot more honest about it and taking people to court.

但在特朗普政府时期，我们要么会迅速放行，要么就会将你告上法庭。

Under But the Trump administration, we are either going to get out of your way very quickly or we are going to take you to court.

我们不会让审查过程成为惩罚手段。

We're not letting the process be the punishment.

因此，人们不需要想办法绕过HSR。

And so people don't need to try to come up with ways to get around HSR.

您将在FTC获得公平的对待。

You will get a fair shake at the FTC.

主席，人们对FTC在消费者保护以及FTC在人工智能领域的职责非常关注。

Chairman, there's a lot of interest in the FTC and consumer protection and what the FTC's remit is in in the domain of of AI.

这是一个例子。

It's an example.

我知道您不会评论具体案件，但X公司、埃隆·马斯克、Sex AI和Grok最近因被用于生成未经同意的色情图像而登上新闻。

I know you won't comment on on specific cases, but X and Elon Musk, sex AI and Grock have been in the news because of the use of the tool to generate non consent sensual sexualized images.

从消费者保护的角度来看，FTC是否在这一领域负有监管职责？

Does the FTC have a role in regulating from a consumer protection standpoint that domain?

是的。

Yes.