CES 展示机器人未来

它太重了，对吧？

It is too heavy, right?

那么，我们能不能推出一款更轻、更实惠的产品，同时还能提供大约80%的类似体验？

So how can we, can we come up with something, you know, lighter, more affordable, like, but we can deliver like 80% of that kind of experience.

这就是X射线玻璃应运而生的原因。

And that's where, you know, X-ray glass is coming about.

尤其是如果我们能用上最新一代的S芯片，对吧？

And especially if we had the latest generation one S, right?

我们还能进一步提升功能。

We further kind of bump up the features.

你知道，我们提到的一个非常独特的功能是2D到3D的实时转换。

You know, some of the, you know, very unique feature we say is two d to three d lifetime conversion.

我们使用自己的芯片来实现实时AI转换。

We're using our own chip to do this AI conversion in the real time.

这太棒了。

And it's amazing.

任何通过眼镜显示的内容，无论是手机、掌上游戏设备还是笔记本电脑，都能实时转换为三维效果。

Any kind of content coming with the glasses, whether it's a cell phone, game handheld gaming devices, your laptop, it can turn into three d in the real time.

我们来谈谈你们是如何降低成本的。

Let's talk about how you've got the price point down.

是的。

Yes.

是因为你们自己制造芯片吗？

Is it because you're you're making your own chip?

是供应链优化，还是你们在利润上做出了让步？

Is it a supply chain navigation, or are you just taking a hit from a profit margin perspective?

我们只是花了很多精力优化整个供应链。

We just, you know, we spend a lot of effort optimizing this whole supply chain.

你知道的，更好的良率、更高的产量，这些都有助于降低成本。

You know, better yield rate, you know, higher volume, all it's kinda gonna help to bring the cost down.

这是来自中国吗？

Is it coming from China?

那是其中一部分。

Part of that.

是的。

Yes.

考虑到国家安全问题，这是否构成风险或问题？

And is that a risk, an issue, given the national security concerns or any

之类的吗？

of that?

很好的问题。

A great question.

你知道，我们也有这种全球供应链策略。

You know, we do have this kind of global, supply chain, strategy as well.

它可能来自中国。

It can be China.

它也可能来自其他地方，你知道，甚至有一天可能来自美国。

It can be other part of the you know, it can be even you in The US in some day.

只是非常，非常快速地一下。

Just very, very quickly.

是的。

Yes.

这个CES，你们吸引了大量人流。

This CES, you've had a lot of foot traffic.

是的。

Yes.

简单反思一下这一点。

Just reflect on that really quick.

嗯，人们喜欢眼镜，也喜欢亲身体验。

Well, you know, people like glasses, and they like to experience.

实际上，尽管我们已经谈论眼镜很久了，但仍有很多人从未真正佩戴过眼镜。

Actually, even though we've been talking about glasses for so long, there's still a lot of people that haven't really tried on any glasses.

所以我认为这仍然是一个漫长的教育过程。

So I think this is still a long kind of education process.

外部首席执行官，徐驰。

External CEO, Chi Xu.

欢迎你再次做客彭博科技，这里是CES现场。

Great to have you back on Bloomberg Tech here at CES.

非常感谢。

Thank you so much.

接下来，我们将与美国国务院经济事务副国务卿雅各布·赫尔伯格对话，讨论芯片出口管制的现状。

Now coming up, we're gonna speak with Jacob Helberg, US state department undersecretary for economic affairs to discuss the state of chip export curves.

消息人士称，中国现已准备批准进口部分英伟达H200芯片。

The sources say China is now ready to approve imports of some NVIDIA h 200 chips.

这是彭博社本周四的重磅新闻。

The big story from Bloomberg this Thursday.

现在是中场休息。

It is halftime.

我们正在拉斯维加斯。

We are in Las Vegas.

这是CES，这是彭博科技。

This is CES, and this is Bloomberg Tech.

欢迎回到彭博科技。

Welcome back to Bloomberg Tech.

有许多新闻正在推动市场，无论拉斯维加斯这里正在发生什么。

There is a lot of news that's driving markets irrespective of what's going on here in Las Vegas.

其中之一是华纳兄弟探索公司的状况。

One of them is the Warner Brothers discovery situation.

派拉蒙天空之舞发布声明，重申其每股30美元的报价。

Paramount Skydance came out with a statement reaffirming their $30 a share offer.

但有趣的是，卡里，他们的计算是，他们希望纳入交易的有线网络价值为零，这很有趣，因为许多市场人士认为，你们对这一点的判断并不准确。

But interesting, Kari, like, the calculus, they're now saying that the cable networks that they want to include in the deal are worth $0, which is interesting because, like, a lot of the street and a lot of the market is saying, we don't think you're quite right on that.

请记住，奈飞只想收购工作室和流媒体业务，将传统有线网络剥离出去，而派拉蒙天空之舞，正如我们在彭博科技上所说，想要整个打包。

Remember, Netflix wants to buy just the studios and streaming, spin out the legacy networks and cable offering, whereas Paramount Skydance, as we say on Bloomberg Tech, want the whole enchilada.

无论如何，我们目前仍处于观望状态，等待进一步进展。

Either way, we're still kind of in a holding pattern until we move forward.

昨晚韩国传出一则重磅消息，那就是三星。

There's a really big story overnight out of Korea, and that is Samsung.

由于内存业务，季度利润翻了三倍。

Quarterly profit tripled because memory.

我们一直在谈论当前内存供应的限制和瓶颈，以及价格上涨的问题。

We've been talking so much about the supply constraint and bottleneck that is memory right now and pricing goes up.

但如果你是内存制造商，价格涨到这种程度，对你的利润通常是非常有利的。

But if you're memory maker and pricing goes up to that extent, it's usually pretty good for your bottom line.

股价下跌了约一个百分点，但这才是最大的头条新闻。

The stock down about a percentage point, but that was the big headline.

本周，内存芯片的短缺问题一直备受关注，不仅在数据中心领域，当然也在PC领域也是如此。

And it has been such a focus this week, the issue and bottleneck that is memory chips, not just in the data center context, but, of course, PC as well.

当然，因为黄仁勋在闪迪会议上强调了这一点。

And, of course, because Jensen Huang made that point at Sandisk.

股价因此剧烈波动。

The shares went wild.

也许这就是我们稍微减持三星的原因。

Maybe that's why we're pulling back a little bit on Samsung.

但让我们换个话题，因为昨天我们与安德鲁·埃勒创始人、帕尔默·拉基的对话中，另一个关键话题是我们讨论了美国与中国之间的竞争。

But let's switch gears a little bit because another key part of our conversation that we had yesterday with Andrew Ell founder, Palmer Luckey, was where we discussed US competition with China.

埃德，你首先问他关于特朗普总统提议限制国防承包商高管薪酬的问题。

And, Ed, you first asked him about, like, the compensation package after president Trump called for capping executive pay with defense contractors in particular.

听听这段话。

Just take a listen.

总统提议的一项规定是将薪资上限设为500万美元。

One of the provisions that the president proposes is a salary cap of $5,000,000.

没错。

Yep.

彭博社的文章中列出了这些首席执行官的一些薪酬情况。

In the Bloomberg story, outlined some of the comp of those CEOs.

你愿意说说你在安德里尔给自己发多少薪水吗？

Would you be willing to say what you pay yourself at Anderil?

如果你遵守的话，我拿

And if you would abide by I pay

我自己年薪10万美元。

my I pay myself $100,000 a year.

所以

So

远低于500万美元的上限。

Significantly than a $5,000,000 cap.

没错。

True.

但我还持有公司大量股份。

But I also own a lot of my company.

所以你知道，这项措施并不是为了针对新兴公司或初创企业设计的；举个例子，如果凯利·约翰逊今天还活着，他持有洛克希德·马丁、诺斯罗普·格鲁曼或雷神公司的大量股份，我想没人会对此有多大意见。

And so so the you know, this this measure is not really written in a way where it's intended to, you know, hit new companies, starting new things, you know, act like, if Kelly Johnson were alive today and he owned a bunch of, you know, Lockheed Martin or Northrop Grumman or Raytheon, I don't think that anyone would have nearly much of a problem with him.

是的。

Yeah.

我给自己每年支付10万美元。

I pay myself a $100,000 a year.

这就是我的薪酬方案。

That's my compensation package.

我拥有大量安多里尔的股份，因为我创办了这家公司。

I own a bunch of of Andoril because I started the company.

所以我的动力是尽可能打造最大的事业。

That's so my motivation is to try and build in the biggest thing possible.

我要说，人们对此提出了相当多的批评，说帕尔默，你真的在这里保持中立吗？

I will say, people have fairly critiqued and said, Palmer, you know, are are you really a neutral party here?

考虑到你正在与这些公司竞争，你真的有资格发表评论吗？

Are you really in a position to comment given you're competing with these companies?

我想说两点。

I'd say two things.

第一，这些规定同样适用于我。

One, these measures do apply in equal measure to me.

我现在不能支付股息。

I now cannot pay dividends.

如果我不投资新项目，不做这些新事情，我就不能进行股票回购。

I now cannot do stock buybacks if I'm not investing in new plans, if I'm not doing these new things.

另一点是，当你想要为自己辩护时，总是很棘手，因为我本意是想帮助解决这些问题。

The other thing is it's always tricky when you wanna you I'm in defense because I wanted to help solve these problems.

对吧？

Right?

这其实和游戏领域的Mod Retro情况一样。

It it it's actually the same thing with like Mod Retro in the gaming space.

哦，当然，Palmer会批评其他游戏公司。

Like, oh, of course, Palmer would criticize these other game companies.

毕竟，他也在游戏领域。

After all, he's in the gaming space.

但没错。

It's like, but yeah.

但我身处游戏行业，因为我想要解决这些问题。

But I'm in the gaming space because I wanna solve these problems.

这简直是个两难境地。

It's kinda this, like, catch 22.

如果你置身事外，他们就会说：那你为什么不做点什么呢？

Like like, if you're outside of it, they'll say, well, why don't you do something about it then?

而当你真的做了些什么，他们又会说：谁在乎你做了什么？

And you do something like, who cares what you're doing?

你只是问题的一部分。

You're just part of the problem.

这一直让我在情感上感到非常艰难。

It's always been emotionally difficult for me.

当我们思考你所处的位置，美国与中国相比的情况时

When we're thinking about where you stand, stand, where US stands versus China

没错。

Yep.

在无人机技术方面，情况如何？

Where does it in terms of drone technology?

我们正处于这个背景下，没错。

We're we're in this context Yeah.

这种地缘政治紧张局势今年初始于委内瑞拉。

Of a geopolitical strain that started this year off with Venezuela.

我们现在把目光转向俄罗斯。

We look now to Russia.

你如何看待中国和台湾问题？

How do you see China and and the Taiwan question?

中国拥有一个极其强大的无人机产业。

China is an incredibly strong drone industry.

这在很大程度上是国家干预的结果。

It is largely a result of state intervention.

这主要是政府在产业政策、供应链投资方面采取的措施，以及中国在全球达成的贸易协议，使这些无人机能够自由进入其他市场。

It's largely a result of industrial policy, of supply chain investments that the government has made, and even trade deals that China has made around the world to allow for for these drones to freely fly out in other marketplaces.

是的。

Yeah.

中国拥有世界上最好的无人机产业。

China has the best drone industry in the world.

根本没法比。

It's not even close.

这绝对是美国的一个弱点。

It's definitely a weakness that The United States has.

最近你看到了对大疆无人机的禁令。

Now you saw recently this ban on DJI drones.

是的。

Yes.

这样做的目的是试图解决这个问题。

The intention there is to try and fix that.

这些问题是可以解决的。

You can fix these things.

美国有能力制造无人机。

The US is capable of building drones.

只是我们无法与中国的劳动法规、排放法规、燃煤电厂的能源价格等竞争。

It's just we aren't capable of competing with the labor laws, emissions laws, energy prices from coal plants, etcetera that you see in China.

因此，我希望对大疆无人机的禁令能在未来一到三年内催生一个强大的美国无人机市场来取而代之。

So I'm hoping that this ban on DJI drones results in a strong US drone market popping up to take its place over the next one, two, three years.

Androle公司最近在日本展示了一款完全由日本零部件制造的无人机。

Androle actually has a drone that we were showing off in Japan recently that's made with 100% Japanese parts.

日本是另一个能够自主制造无人机的国家。

Japan is another nation that can build their own drones.

他们不需要大疆，但当中国允许其产品大量涌入市场时，这很难在经济上站得住脚。

They don't need DJI, but it's hard for that to make economic sense when China's allowed to flood the market.

刚才那位是Androle创始人帕尔默·拉基。

That was Anderrill founder, Palmer Lucky.

关于美国与中国的关系以及半导体背后的国家安全问题，我们邀请到美国国务院经济事务副国务卿雅各布·赫尔伯格，他同时也是希尔与山谷论坛的创始人，我们去年曾参加过该论坛。

For more on The US relationship with China and the national security concerns behind semiconductors, we're joined by Jacob Helberg, US State Department under secretary for economic economic affairs, but also the founder of the Hill and Valley Forum, which we were able to attend last year.

在与中国相关的背景下，我们之前没来得及和帕尔默讨论的一件事是，习近平主席最近重申了实现与台湾统一的长期目标。

There was something in the in the in the China context which we didn't get to with Palmer, which was a piece of recent news with president Xi reiterating, I would say, a long stated goal of reunification with Taiwan.

我之所以想从这里开始，雅各布，是因为在你的Paxilica工作中，每个人都试图规划一种情景：如果这一情景成真，台湾的半导体产能将被切断。

Now the reason I wanna start there, Jacob, is is in the your efforts with Paxilica, the scenario that that everyone is is trying to plan for Taiwan and access to capacity in the the semiconductor side being shut off if that scenario were to unfold.

从2026年开始，能否请您分享一下您对此的最新看法？

Starting 2026, could you just bring us your latest thinking on that, please?

当然可以。

Absolutely.

很高兴能在这里参加拉斯维加斯的CS大会，这里堪称美国创新以及Paxilica国家如韩国和日本创新的绝佳窗口。

And it's great to join you guys here at CS in Las Vegas, which is such a great window into American innovation and innovation from a lot of Paxilica countries like South Korea and Japan.

就在这一具有里程碑意义的宣言签署仅仅三周后，我们就能展开这样的讨论，令人无比振奋。

It's incredibly exciting that we're talking barely three weeks after the signing of this landmark declaration.

最终，让我来说明为什么这一点至关重要，以及为什么如此多的人在谈论技术竞争。

Ultimately, let me put into context why this is important and why so many people are talking about technology competition.

如今，很明显，如果说二十世纪依赖钢铁，那么二十一世纪正日益依赖硅和计算能力。

Today, it's clear that if the twentieth century ran on steel, the twenty first century is increasingly running on silicon and compute.

我们已经看到，在美国，三分之一的GDP增长来自人工智能，随着生产力开始加速，整个经济领域的增长正在提速。

And we're already seeing in The United States a third of our GDP growth coming from AI and growth picking up and accelerating as productivity is starting to accelerate growth in sectors across the economy.

今年七月，特朗普总统在一场由我与白宫人工智能事务主管大卫·萨克斯共同主办的‘赢得人工智能竞赛’活动中发表了具有里程碑意义的演讲，他在演讲中明确宣布，美国必将赢得人工智能竞赛，并将此确立为美国政府的官方政策，不惜一切代价实现这一目标。

So back in July, president Trump rolled out this landmark speech at an event I co hosted with White House AI czar David Sacks called winning the AI race, where it was a landmark moment where he declared that The US, is set to win the AI race and that it is the official policy of the US government to do whatever it takes.

几个月后，我很高兴地报告，国务院已制定了一项战略，将这一宏大目标分解为三个部分。

A few months later, I'm happy to report that at the state department, we have adopted a strategy that breaks down this broad goal into three parts.

我们希望通过推动创新、抢占市场份额和保障供应链，帮助美国政府赢得人工智能竞赛。

We wanna help the US government win the AI race by leading an innovation, by gaining market share, and by securing our supply chains.

我们与七个技术最先进的国家共同启动了Paxilica计划，这些国家包括日本、韩国——三星和SK海力士的所在地，以及三菱等企业所在国。

So we launched Paxilica with a group of seven countries, and, they are the most technologically advanced countries, including Japan, South Korea, which is home to Samsung, x SK Hynix, Mitsubishi.

目前，我们正积极开展密集磋商，计划于2026年进入Paxilica计划的实施阶段。

And we are engaged in, a lot of intensive talks to now transition to the implementation phase of Paxilica in 2026.

你一直非常焦虑，多年来一直在警告美国和那些掌握权力的人，我们不该以如今这种方式对华开放。

A lot of your anxiety and your you know, you've been at this for a very long time trying to warn The United States and those in positions of power now in the power that you have yourself that we shouldn't have been exposed to China in the way that we have been.

从国家安全的角度来看，你如何看待最近有关H200芯片进入中国的情况？

Now from a national security perspective, what do you think of the latest that maybe h two hundreds go into China?

到本季度，他们可能会批准阿里巴巴、比亚迪使用英伟达的芯片？

That by this quarter, they might sign off for Alibaba, for BYD to be able to access NVIDIA's chips?

因此，赢得供应链的一部分目标意味着我们需要扩大市场份额。

So part of the goal of winning the supply chain means we need to expand market share.

有时创新与技术扩散之间存在一定的张力，因为当你扩散技术时，有时会略微牺牲创新，因为更多人获得了这项技术，从而缩小了你的技术优势。

And sometimes there's a bit of a tension between innovation Because when you diffuse technology, sometimes you're compromising a little bit on innovation because more people have access to to that technology which narrows your technological edge.

我们通过出口H200，确保全球开发者基于美国技术栈进行开发，我们希望确保通过与海湾国家如阿联酋、沙特阿拉伯以及其他尚在洽谈中的国家的战略双边协议，使美国模型始终领先，以确保我们的公司拥有远超他人的计算能力，从而使我们的大语言模型保持领先。

Part of what we're doing by exporting our h two hundreds is making sure that the world's developers are building on top of the American stack, and we wanna ensure that American models actually stay ahead through these strategic bilateral deals in countries in The Gulf like The UAE, the Saudi Arabia, and, a number of others that are on the way, but still too early to disclose here today in order to make sure that our companies have by far the most compute capacity so that our LLM stay ahead as well.

雅各布，关于市场份额这一块，DC许多人士向我们提出的担忧是，如果我们不向海湾地区出口，中国就会这么做。

Jacob, on on that market market share bucket, you know, the concern that that many in DC outlined to to us is if we don't export to The Gulf, China will.

从美国的角度来看，这样想对吗？

Is that the right way to think about it for America?

美国？

America?

我认为这是一个机遇，因为自然界厌恶真空。

I think it's an opportunity where there is nature abhors a vacuum.

我们必须趁现在填补这个真空。

We have got to fill the vacuum while we can.

我们拥有世界上最好的技术，全世界都知道这一点。

We have the world's best technology, and the worlds know that.

最终，当美国允许创新者真正主导市场份额时，美国才得以繁荣。

And, ultimately, America has thrived when we allow our innovators to actually dominate market share.

他们确实主导了市场。

They did dominate.

英伟达曾占据95%的市场份额，而现在詹森·黄一无所获。

NVIDIA had 95% of the market, and now Jensen Posey has zero.

也许他正被允许逐步夺回一些份额。

And maybe he's being allowed to claw some back.

马儿难道没有已经跑掉了吗？

Has not the horse bolted?

我们难道不是已经让寒武纪和华为成长起来了吗？

Have we not already allowed Cambra Con to grow and Huawei to grow?

我们真的能够重新掌控并建立属于自己的全球技术体系吗？

Are we really going to be able to clean back and have the stack of our own worldwide?

这正是我们推出Paxilica的原因，以确保在保护最敏感技术的同时，对出口我们的先进技术采取明智的策略。

And that's exactly why we launched Paxilica to make sure that we have a sensible approach to exporting our best technologies while protecting our most sensitive technologies.

我们正在与合作伙伴进行双轨对话。

We're having dual track conversations with our partners.

我们希望保护敏感技术，但同时也必须找到获取市场份额的途径。

We wanna protect our sensitive technologies, but we have gotta have a path to gain market share.

我认为大家都达成了共识。

And I think everyone's on the same page.

所有加入Paxilica的人都明白，无论你对这个国家持何种看法，全球供应链正集中于一个国家，这风险太高、过于脆弱，最终会成为全球经济的负担。

Everyone who joined Paxilica understands that the world's supply chain is being concentrated in one country, no matter how you feel about that country, isn't too risky, it's too brittle, and is ultimately a liability to the global economy.

因此，我们发起了这个经济安全联盟，联合全球最先进的科技公司和国家，以确保我们能多元化这些供应链。

And that's why we started this, economic security coalition with the world's most advanced technology companies and and countries in order to make sure that we diversify those supply chains.

雅各布，这些公司都出席了在拉斯维加斯举办的CES展会。

Jacob, those companies are here at CES Las Vegas.

你在CES期间希望达成什么目标？

What what do you hope to achieve in the time that that you have at CES?

因此，我们为2026年的路线图部分目标是加强政策协调。

So part of what we're looking for our road map in 2026 is we want to focus on policy coordination.

我们希望扩大Paxilica的成员规模，并专注于基础设施项目。

We want to expand membership of Paxilica, and we want to focus on infrastructure projects.

我们对基础设施的看法是，希望通过先进物流来聚焦供应链的动脉。

The way that we view infrastructure is we want to focus on the arteries of supply chains through advanced logistics.

我们希望聚焦工业产能的核心力量，特别是晶圆厂、工厂和炼油厂。

We want to focus on the muscle might of industrial capacity, particularly with fabs, factories, and refineries.

我们还想关注燃料，尤其是资本和能源。

And we also wanna focus on the fuel, particularly capital and energy.

与在这里的公司交流时，一个特别有趣的地方是，我们可以实际检验这些想法，与那些每天最接近这些问题的建设者们一起验证。

One of the things that's so interesting about talking with companies here is we can get to actually stress test a lot of these ideas with builders and with people who are the closest to these problems day to day.

雅各布，从地缘政治的角度来看，这一周非同寻常。

Jacob, this week has been extraordinary from a geopolitical perspective.

我想听听你对这个问题的总体看法。

And I wanna just get your thoughts high level.

你对PAC silica的看法是，它是一种通过权力实现的手段。

Your view on PAC silica is that it's sort of piece through power.

我们是否应该这样理解委内瑞拉发生的事情——俄罗斯船只的行动，以及目前全球持续发生的情况？

Is that how we should be interpreting what's occurred Venezuela with Russian ships with what continues to occur worldwide right now?

我认为，归根结底，过去几周事件的一个明确启示是：世界必须明白，特朗普政府言出必行，行必有果。

So I think at the end of the day, one of the takeaways of the events the last few week is very clear that, the world needs to know that the Trump administration says what it means and means what it says.

我认为卢比奥国务卿非常清晰地阐述了美国和特朗普政府在委内瑞拉问题以及格林兰岛等热点议题上的立场。

I think secretary Rubio spoke very eloquently about where The US stands and the Trump administration stands on Venezuela and other issues that have been in the news like Greenland.

最终，对国际规范最大的威胁，是西方 hemisphere 和美国在北极从北面、拉丁美洲从南面步步紧逼时却昏昏欲睡。

Ultimately, the biggest threat to international norms is, the Western Hemisphere and The United States being asleep at the switch while we are encroached from the North through the Arctic and from the South through Latin America.

最终，Paxilica 的目标是与我们最亲近的技术伙伴合作，采取一种非常务实的方法，推动战略性投资协议，真正推动局势向前发展，使我们所有人受益。

Ultimately, Paxilica is an effort to to pursue a very pragmatic approach by working with our closest technological partners to actually force strategic investment deals and allow moving the needle in a way that actually benefits all of us.

这就是为什么这个联盟如此多元化。

That is why this coalition is so diverse.

这非常非常新。

It's very, very new.

这是一个全新的国家组合，包括新加坡、以色列、阿联酋等玩家，这些国家通常不被纳入二十国集团和七国集团等传统框架中。

It's a completely new grouping of countries that includes players like Singapore, like Israel, The UAE, people who aren't used to being included in in traditional legacy forms like the g twenty and the g seven.

然而，这些国家的影响力远超其体量，我们非常自豪能与它们合作。

And yet these countries punch far above their weight, and we're we're very proud to partner with them.

雅各布，很高兴能和你有这段交流时间。

Jacob, it's great having some time with you.

我也很期待稍后能和你同台亮相。

I'm excited to be on stage with you a little bit later as well.

雅各布·赫尔伯格是美国国务院主管经济事务的副国务卿，同时也是海伦·巴里论坛的创始人，该论坛汇聚了风险投资界与华盛顿方面围绕中国议题的对话。

Jacob Helberg, he's US Department of State Undersecretary for Economics Affairs, and, of course, founder of the Helen Barry Forum, which is bringing together a VC and Washington surrounding the China issues.

与此同时，接下来我们将继续探讨人工智能的未来，随着企业不断在数字世界与实体世界之间来回切换。

Meanwhile, coming up, more on the future of AI as companies shift in and out of the digital world and back into physical.

接下来登场的是凯ndred风投的管理合伙人张史蒂夫。

Steve Zhang, his Kindred Ventures managing partner joins us next.

这是彭博科技。

This is Bloomberg Tech.

人工智能在现实世界中的影响成为CES展会热议的焦点，投资者密切关注哪些技术能从炒作走向现实，或许还能推动这一现实的实现。

AI's real world impact is dominating the conversation here at CES with investors watching closely to see which technologies move from hype to reality and maybe they foster that reality.

我们邀请到了史蒂夫·张，他是Kindred Ventures的管理合伙人兼创始人，这家早期风险投资公司以支持定义类别的初创企业而闻名，在我们想到你与Uber的合作，以及众多公司取得重大退出和并购成就时，你都做出了重大投资。

Steve Zhang's with us, managing partner and founder of Kindred Ventures, early stage venture firm known for backing category defining startups, making significant bets on when we think of the work that you did with Uber, what we think of so many of the companies that have had significant exits and M and A that's happening.

但对你来说，我认为本次展会的明星之一是Neuro。

But for you here, I think one of the stars of the show has been neuro.

对吧？

Right?

我们看到了一个重磅消息：我们曾邀请Lucid上节目，但现在Uber与Neuro联手，共同打造未来的机器人出租车。

We saw this big announcement coming from we had Lucid on the show, but Uber and indeed neuro teaming up to provide the future of robotaxis.

但这是一个竞争激烈的领域。

But it's a crowded field.

你们是如何决定Neuro将在自动驾驶竞赛中胜出的？

How did you decide neuro was gonna be helping win in that autonomous fight?

当然。

Sure.

如果你回溯时间，回顾优步的历史，你会记得十年前，优步曾宣布将投资自动驾驶技术。

So if you rewind time and you look at the history of Uber, you'll remember, ten years ago, Uber had announced that they were gonna, invest in self driving car technology.

自那以来，发生了许多事情。

Now a lot of things happened since then.

最初，他们计划在特拉维斯·卡兰尼克的领导下推进这项业务，但在达拉上市前不得不放弃整个业务。

So they originally, had planned it with Travis Kalanick at the helm of the company and then had to jettison that whole business before going public under Dara.

我认为达拉重新拾起了这项计划，并表示我们需要加速我们的战略。

And I think what Dara did is pick that back up and say, we need to accelerate our strategy.

大约两年前，这些计划开始付诸实施。

So about two years ago, a lot of those, plans were put into motion.

有趣的是，Neuro 是少数几家像 Waymo 一样，曾在旧金山湾区街道上使用激光雷达和摄像头进行训练的公司之一。

And what was interesting was, Neuro was a company that had been one of the few companies like Waymo that had been training, around Bay Area streets with their LiDAR and their cameras.

他们最近推出了一项自动驾驶智能平台，本质上是一个操作系统，允许任何第三方汽车制造商具备自动驾驶能力。

And, they recently launched a, autonomy intelligence platform, which is basically an operating system to allow any third party automaker to have self driving capabilities.

所以你在这里看到的是Neuro、Lucid和Uber之间的一次精彩合作，不仅为Lucid汽车改装了完整的四级自动驾驶系统（不是二级加加辅助），而是实现了类似Waymo和特斯拉FSD的全自动驾驶能力，并在Uber平台上与Lucid共同推出。

And so what you saw here was a great launch between Neuro, Lucid, and Uber to not only fit retrofit a a Lucid car with, autonomy, full level four, not eight s, right, not level two plus plus assistance, but full self driving capability like Waymo and Tesla FSD, and launch that with Lucid on the Uber network.

今年年底，你就可以在湾区打开Uber应用，乘坐一辆自动驾驶汽车。

So you'll be able to open up your Uber app at the end of this year in the Bay Area and be able to take a self driving car.

这就像一个轻资产模式。

It's like capital light model.

没错。

Yes.

让别人承担硬件和系统的负担。

Let someone else take on the burden of the hardware and the stack.

你只需要做车队管理。

Just do the fleet management.

但对我来说，那个世界更大的新闻是阿尔法·罗密欧。

But for me, the bigger piece of news in that world was Alfa Romeo.

嗯。

Mhmm.

对吧？

Right?

因为英伟达也在推出同样的东西，说：嘿。

Because NVIDIA is coming with the same thing saying, hey.

我们有完整的解决方案。

We have a full stack solution.

我知道你注意到了这一点。

I know that you you caught notice of that.

所以，AGX Thor，作为Neural的一部分，是

So, AGX Thor, which is inside of Neural, is part of

那里面的推理部分。

that Inside the inference.

对。

Right.

对。

Right.

所以英伟达在做的事情非常聪明，他们正在销售芯片。

And so what what NVIDIA is doing, and it's very smart of NVIDIA, they are selling chips.

对吧？

Right?

所以他们想卖给你一个入门套件。

And so what they wanna do is sell you a starter kit.

他们想给你一个开源模型。

They wanna give you an open source model.

他们想给你开放的数据集，并说：尽情发挥吧。

They wanna give you open datasets, and they say, go to town.

我们希望你构建一个自动驾驶平台，以适配这些汽车，适配你的网约车或配送网络、物流网络。

We want you to build a self driving car platform, to fit these cars, to fit your ride sharing or delivery network, your logistics network.

他们想说的是，我们希望市场上有众多竞争对手，这样我们才能卖给你更多的GPU。

And what they're saying is, we want to have many competitors out there so we can sell you more GPUs.

因此，Neural实际上是最早使用这一模式多年的合作伙伴之一。

And so, Neural is actually one of the very first partners that's been using that for years now.

英伟达是Neural的投资方。

NVIDIA is an investor, in Neural.

因此，Neural和Uber可以说是他们的典范，也是他们业务的旗舰案例。

And so, you know, Neural and Uber are sort of the prime example and hopefully a flagship for them and their business.

但总的来说，在这次展会上，英伟达正在扮演造王者的角色。

But, you know, NVIDIA on a larger scale at this show is what they're doing is they're king making.

他们正在为整个行业造王者。

They're king making an entire sector.

他们说，一年前我们还在谈论机器人。

They're saying, you know, a a year ago, were talking about robotics.

今年，他们将这一概念转向了物理AI。

This year, they've transitioned that into physical AI.

现在他们谈论的不再只是机器人。

Now they're not just talking about robotics.

他们谈论的是自主性和移动性，因为这一趋势正达到高潮。

They're talking about autonomy and mobility because it's coming to a crescendo.

所以，他们非常聪明地说，我们就像美联储。

And so, they're very, very smart in saying, we're sort of like the Fed.

对吧？

Right?

我们是AI领域的美联储。

We're gonna we're the Fed for AI.

我们会告诉你们，这就是头条。

We're gonna tell you That's the headline.

头条。

Headline.

对吧？

Right?

对吧？

Right?

是的。

Yeah.

是的。

Yeah.

是的。

Yeah.

我们会告诉你每个人将去往哪里。

And we're gonna we're gonna tell you where everyone's gonna go.

你可能正处于这股浪潮之中。

You may be on that wave.

你可能不在那股浪潮中，但你应该在。

You may not be on that wave, but you should be.

顺便说一下，我们拥有模型、数据、芯片以及整体计算能力，助你构建这一切。

And by the way, we have the models, we have the data, and we have the chips and overall compute for you to to go build that.

所以，你知道，很多人在争论，NVIDIA的王国现在是否面临威胁？

And so, you know, think I there's been a lot of, argument about, you know, is NVIDIA's kingdom, right, at risk right now?

我认为，他们持续带来惊喜，并领先竞争对手数年。

And I would say that they continue to surprise and be several years ahead of their competition.

所以我们非常兴奋，但也在密切关注

So, we're we're very excited, but we're watching

我的意思是，你看。

I mean, look.

你以投资优步而闻名，最终，优步和Lyft成为两大赢家。

You were known for backing Uber, and then ultimately, there were two key winners out of that, Uber and Lyft.

当我想到你也在关注的领域时，我想问，自动驾驶领域能有多少玩家？

When I think about the areas that you're also thinking of, when I mean, how many players can there be in autonomous?

基础模型领域又能有多少玩家？

How many players can there be in foundational models?

你投资了Perplexity，而大家都认为，可能只有Anthropic、OpenAI和Gemini能占据一席之地。

You're in perplexity, and everyone's thinking maybe there's only an anthropic and an open AI and a and a Gemini space.

当然。

Sure.

正如你正确指出的，任何创新曲线最初都有许多竞争者，最终会缩减到两到三家。

As as you're rightfully noting, any innovation curve starts with many competitors, and it whittles down to two or three.

这仅仅是硅谷的一个经验法则。

It's just a it's a it's a rule of thumb in in Silicon Valley.

所以现在，Neuro进入了这个领域。

So right now, neuro is entering the space.

有Waymo，有特斯拉，还有Neuro。

There's Waymo, there's Tesla, and there's neuro.

在中国，还有其他几家竞争对手。

In China, there's several other competitors.

对吧？

Right?

在欧洲，一个都没有。

In in Europe, there are none.

对。

Right.

所以在四级自动驾驶能力方面。

And so on level four capability.

所以我认为正在发生的是，由于资源、训练数据以及在所有街道上进行训练所需的时间，竞争正在逐渐减少。

So I what I think is happening is, you you're seeing a whittling down because of the resources, the training data, the amount of years that you need to do to to train across all these streets.

这并不是一家你可以从零开始创办的软件公司。

It's not a software company that you can just start up out of scratch.

所以我认为，在机器人领域，你也会看到这种竞争的逐渐缩减。

So I think I think one of the things that you'll see also in robotics is you'll see this this whittling down of competition.

你看到波士顿动力公司了，它被现代汽车收购，这非常有趣。

And you see Boston Dynamics, which is owned by Hyundai, which is very interesting.

它们表现得相当不错，并且发布了一款人形机器人。

They're doing quite well, and they they revealed a humanoid.

我一直对人形机器人的发展保持谨慎和怀疑态度。

You know, I've been very careful and skeptical about the development of humanoids.

但我想说，这是我第一次看到真正实用的人形机器人在实际运作。

And I would say that this is the first time I've seen humanoids in action, that look practical.

因此，我认为机器人领域也会出现类似的情况。

And so I think you'll see that in robotics as well.

我认为在生成式媒体中也会看到这一点。

I think you'll see that in generative media.

我认为在智能代理中也会看到这一点。

I think you'll see that in agents.

因此，许多这些平台将逐渐缩减至两到三家，最多四家玩家。

So a lot of these platforms will start to whittle down to two, three, maybe four players.

今年将是这一趋势之年，这就是为什么我们在这里看到的更多证据表明，推理所需的计算量将超过训练。

And this year will be the year of that, which is why inference you know, one of the other things that we're seeing here is that more further evidence that inference is going to exceed training in terms of compute.

这种转变，如果可以借用一个比特币术语的话，意味着很多。

And that flip that flippening, if you will, right, if I can use a Bitcoin term, that flippening is, means a lot.

因此，对计算资源的需求远远超过了供应。

So demand for compute is far exceeding supply.

即使存在HBM短缺，这种情况也会加剧。

Even with the HBM shortage, it's gonna increase that.

因此，计算领域正上演一场生死大战。

So there's a battle royale happening here in compute.

史蒂夫·张，Kindred Ventures带来本周最重大的新闻。

Steve Zhang, Kindred Ventures hitting literally the biggest stories of the week.

我们从CES带来了独家消息，还有更多内容即将推出。

We've had hair out of CES, and we have much more to come.

敬请继续关注。

Stay with us.

这是彭博科技。

This is Bloomberg Tech.

目前市场上的主要故事是科技股下跌。

The story in markets right now is tech is down.

英伟达下跌了百分之零点八，表现远低于预期。

Nvidia, off by eight tenths of 1%, have been much lower.

整个星期以来，我们并没有看到预期中的狂热情绪。

Over the course of the week, we've kind of not had any of the of the euphoria we expected.

我们所有人来到拉斯维加斯，都以为英伟达——在史蒂夫·张看来，它是人工智能的造王者、人工智能的美联储——会带来一些深刻的洞见，推动股价继续上涨，但实际情况并未如预期那样发展

We all came to Las Vegas thinking that NVIDIA, who in Steve Zhang's words was the kingmaker of AI, the fed of AI, would tell us something profound that would keep the stock going higher, and it's not really panned out that

方式。

way.

尽管我们确实说过，到2026年将有5000亿美元的收入流入，这还会更高。

Even though we kinda did said, look, that $500,000,000,000 of of inbound I see in revenue in 2026, that's gonna be higher.

也许我们甚至能在中国实现两千亿规模，但预期依然如此之高。

Maybe we even get h 2 hundreds into China, and yet so high expectations.

看吧。

Look.

AMD，莉萨·苏还能做些什么呢？

AMD, what more could Lisa Su do?

她创造了一个新术语。

She made up a term.

是Yoto Flop吗？

Was it yoto flop?

她谈到了仍需的巨大计算规模，以及她看到的成功模式，但股价却下跌了。

She talked about the sheer scale of compute that's still necessary, the winning formula that she's seeing, and stock is down.

但高通真正给市场留下了深刻印象。

Qualcomm, though, really managed to impress the market.

它涉足机器人领域，

It's from robotics and

进入了大家早已在讨论的市场。

a into the market that everyone was already talking about.

你知道吗？

You know?

所以你看到芯片制造商说，好吧。

So you see the chipmakers say, okay.

下一阶段是物理人工智能。

The next phase is physical AI.

我们确实学到了一些深刻的教训，因为外面的环境很艰难。

We did learn some hard lessons that it's difficult out there.

内存是一个大故事。

Memory is a big story.

还有机器人。

And then there's robots.

对吧？

Right?

这是消费电子展，现场真的有机器人展出。

It is the consumer electronics show, and there were real robots to see.

那我们听到了什么？

And what did we hear?

有14种不同的人形机器人，更别提那些随处可见的毛茸茸的伙伴了，它们将成为你孩子或老人的陪伴。

14 different humanoids, not to mention the furry friends that we're seeing around here that are gonna be a companion to your child or or to the elderly.

在这里亲临CES现场，真是非同寻常。

It's been extraordinary to be here on the floor of CES.

你有没有看到这些创新的水平？哪些东西会真正留下来，哪些肯定不会？

And do you see the level of innovation, stuff that will stick, stuff that definitely won't?

我还没试过那种边吃边播放音乐的棒棒糖。

I still haven't tried that lollipop that plays music as you eat it

在那里。

there.

是的。

Yeah.

我们谈成了几笔交易。

And we got some deals.

我们有一些并购交易。

We got some m and a.

我们还筹集了一些资金，但这是开启新一年的最佳方式。

We got some fundraising, but it is the best way to start the year.

确实如此。

It is.

我的意思是，我们还在谈论那些估值和崩盘的情况。

I mean, boy, still talking about those value valuations and flopping as well.

本期《彭博科技》就到这里。

That's it for this edition of Bloomberg Tech.

但别忘了，今天晚些时候将播出对MiniMax联合创始人兼首席运营官的独家访谈。

But don't forget, later today, an exclusive interview is coming with the cofounder and COO of MiniMax.

就在CES现场，我们将讨论其在香港上市的事宜。

That's right here from CES to discuss the Hong Kong listing.

中国AI领域的IPO市场正热火朝天。

China IPO market for all things AI is on fire.

2026年的IPO前景看好。

IPO twenty twenty six looking good.

回顾一下这期播客。

Recap on the pod.

你知道在哪里能找到它。

You know where to find it.

这是来自拉斯维加斯的Bloomberg Tech，最后一次播报。

For the final time from Las Vegas, this is Bloomberg Tech.