本集简介
双语字幕
仅展示文本字幕,不包含中文音频;想边听边看,请使用 Bayt 播客 App。
AI代理变得越来越令人印象深刻。
AI agents are getting pretty impressive.
你可能甚至没意识到你此刻正在听一个AI代理。
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.
我们全天候工作,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
This
彭博科技。
is Bloomberg Tech.
接下来,英伟达计划在未来五年内投资10亿美元,与礼来公司合作建立新实验室,以加速人工智能在制药行业的应用。
Coming up, NVIDIA plans to invest $1,000,000,000 over five years in a new lab with Eli Lilly to speed up the use of AI in the pharmaceutical industry.
此外,派拉蒙起诉华纳兄弟探索公司,并计划提名董事进入董事会,因其日益激烈的收购行动旨在挑战奈飞。
Plus Paramount sues Warner Brothers Discovery and plans to nominate directors to the board as its increasingly hostile takeover bid takes aim to upset Netflix.
阿尔法特公司市值突破4万亿美元。
An Alphabet breaches $4,000,000,000,000 in market value.
这是CNBC的报道。
This is CNBC reports.
今年苹果选择了Gemini来驱动Siri。
Apple has picked Gemini to power Siri this year.
然后这些举措又回落了。
And then those moves fall again.
而且,事实上,我们看到整个市场都弥漫着焦虑情绪,埃德。
And, really, we've got anxiety across the markets more broadly, Ed.
深入分析市场对政府似乎正在加强对美联储施压这一态势的反应,这就是市场正在消化的内容,也是为何目前我们看到一种‘抛售美国资产’的态势。
Digging into the market implications of what seems to be a dialing up in the administration's fight with the Fed, that is what the market tries to digest and why we see a bit of a sell America theme at the moment.
是的。
Yes.
关于基准指标还有更多变动。
There's more moves in terms of benchmarks.
纳斯达克100指数的股票下跌了十分之一百分点,但当我们关注债券市场和美元等领域的动态时,会发现更大的波动。
The stocks were down by a tenth of a percent on the Nasdaq 100, but we're seeing bigger moves when you're looking what's happening in the bond market, in the dollar, for example.
我们看到三十年期收益率再次上升两个基点,而美元则走弱。
We're seeing thirty year yields once again pushing up two basis points, seeing the dollar on the downside.
所有这些都在我们试图消化宏观影响的同时发生,我们正等待明天的CPI数据,但你现在需要关注的是表象之下的情况。
All of this as we try and bake in the macro implications, we look to CPI tomorrow, but you're looking at what's underneath the hood right now.
是的。
Yeah.
过去一小时最大的突发新闻是,苹果今年已选择谷歌的Gemini模型来驱动Siri。
The big breaking news of the last hour is that Apple has selected Google's Gemini model to power Siri this year.
CNBC援引苹果公司的声明进行报道,你可以在新闻 headlines 发布的瞬间看到股价的剧烈反应。
Reporting from CNBC citing an Apple statement, and you can see a big reaction in the stock the moment the headlines hit.
但这种反应已经消退了。
It has faded.
让我带你回到十一月,当时彭博社的马克·古尔曼报道说,双方协议已非常接近,苹果每年将向谷歌支付10亿美元,但这只是临时安排,直到苹果开发出自己的基础模型为止,我们会持续关注这一进展。
I'll take you back to November when Bloomberg's Mark Gurman reported that a deal was very close, that Apple would pay Google $1,000,000,000 per year, and that this was just an interim until Apple could get its own foundation models together, but we'll continue to track that story.
今天早上市场上另一项有趣的交易是,英伟达与礼来公司共同投资10亿美元,用于支持一家位于硅谷、为期五年的AI制药实验室。
Another interesting transaction in the market this morning, NVIDIA co investing $1,000,000,000 with Eli Lilly into a Silicon Valley based lab split over five years, AI pharmaceuticals.
但这里的细节很有趣,卡拉。
But the details are interesting here, Kara.
我们想了解这些细节。
And we wanna get those details.
我们可以请凯蒂·格里菲尔来具体说明。
We can do exactly that with Katie Greifel.
她正在参加摩根大通医疗保健会议。
She's over at the JPMorgan Healthcare Conference.
确实,看看硅谷,正越来越接近医疗健康领域。
Indeed, looking at Silicon Valley, getting ever closer to the world of health.
凯蒂,你如何看待这笔交易?
Katie, what do you make of this deal?
是的。
Yeah.
尤其是当你记得礼来公司在10月份就与英伟达合作开发AI超级计算机时,这个头条新闻真是令人印象深刻。
This was a really interesting headline to wake up to, especially when you remember that Eli Lilly partnered with NVIDIA back in October to develop an AI supercomputer.
关于这笔具体交易的细节,你们在梳理时会发现,五年内将投入十亿美元在硅谷建造一个设施,基本上是一个新的实验室。
So the details on this specific deals as you guys run through them, 1,000,000,000 over five years to build a facility in Silicon Valley, basically a new lab.
除了这些之外,我们对条款的细节了解不多,但两家公司都将此描述为一项联合投资。
We don't have much details when it comes to the terms beyond that, but both companies have described this as a joint investment.
我刚刚实际上与礼来公司的首席财务官卢卡斯·蒙塔尔塞谈了谈,了解他们对人工智能与医疗保健以及这项具体合作的最终愿景。
And I just actually spoke to the CFO of Eli Lilly, Lucas Montarce, about what the ultimate vision is here when it comes to AI and healthcare and this specific partnership.
他说,这实际上关乎药物发现。
And he said this is really about drug discovery.
当然,现在还处于早期阶段。
It's still early days, of course.
他们还没有建成这个设施。
They haven't built the facility yet.
但他们的希望是,这将有助于发现药物,真正针对并治疗那些难以治疗的疾病。
But the hope is that this will help basically in discovering drugs to really target and treat those really hard to treat diseases.
比如基因疗法。
Think gene therapy, for example.
所以,这还处于早期阶段,但目标非常宏大。
So again, early days, but some big ambitions here.
彭博社的凯蒂·格劳菲尔德一整天都在旧金山的摩根大通医疗保健会议现场报道。
Bloomberg's Katie Graufeld just down the Road at the JPMorgan Healthcare Conference all day long here in San Francisco.
新年刚过去不到两周,英伟达就已经主导了新闻动态。
Less than two weeks into the new year, and NVIDIA is already dominating the news flow.
我想请来简·亨德森公司全球科技与创新团队的投资组合经理丹尼·菲什谈谈。
I wanna get to Denny Fish, portfolio manager on the global technology and innovation team at Jan Henderson.
坦率地说,英伟达是你管理的多个基金中的第一大持仓。
And the frank reality is that NVIDIA is is the top holding across a number of funds that that you manage.
你如何看待英伟达与礼来之间的这种关系?
What do you make of that relationship between NVIDIA and Eli?
是的。
Yeah.
这对我来说非常合理。
Well, it actually it makes a ton of sense to me.
我的意思是,如果我们回顾自ChatGPT问世以来的过去三年里吸引所有关注的东西,本质上都是人工智能的数字形态。
I mean, if if we think about what's grabbed all the attention over the last three years since the ChatGPT moment, it's effectively been the idea of the digital manifestation of AI.
但现实是,如果我们思考人工智能将如何长期影响经济和社会更广泛的层面,人工智能的物理形态所带来的机遇同样巨大,甚至更大。
But the reality is if we think about the profoundness of how this will impact the economy over time and society more broadly, The physical manifestation of AI is as big if not a bigger opportunity.
想想药物研发对社会意味着什么,再想想自动驾驶已经取得了多大进展,机器人技术未来将走向何方——这些不同领域都让英伟达大力投入变得非常合理。
If you think about drug discovery and what that means for society, and then just think about how far we've come with autonomous driving, where we're going with robotics, There are all these different verticals where it makes a lot of sense for NVIDIA to be leaning into.
坦白说,英伟达拥有世界上顶尖的人才。
And frankly, NVIDIA's got some of the best talent in the world.
看到这些与行业领军企业的合作,就完全说得通了。
And to see these partnerships with market leaders makes a ton of sense.
我有时会遇到一种关于循环性的论点,但现实是,如果你是黄仁勋,并且坚信这些市场机遇如此巨大,你就应该全力投入,像他们与礼来合作建立实验室那样进行联合投资,或像他们可能对OpenAI所做的那样进行直接投资。
And I bring it back sometimes you get this argument of circularity, but the reality is if you're Jensen and you have a firm belief at just how big these market opportunities are, you should be leaning in hard to this in invest co investments into labs like they're doing with Lilly or direct investments into companies like they're potentially doing with OpenAI.
关于自动驾驶,很多人都能理解。
With autonomous driving, a lot of people get it.
这个周末我对此进行了大量撰文。
I wrote about it extensively this weekend.
我们会在节目后面讨论这个话题。
We're gonna talk about it later in the program.
首先,市场推广将由梅赛德斯来推进。
The go to market in the first instance is from Mercedes.
我们是使用硬件和软件吗?
Do we use hardware and software?
嗯。
Mhmm.
但在制药、药物发现和研究领域,英伟达未来将如何实现市场推广?
But with pharmaceuticals, drug discovery, research, what is the go to market that NVIDIA is gonna pull off in the future?
到目前为止,如果我们看看人工智能在制药领域的早期应用,主要体现在临床流程中,比如让已经发现的药物更快进入流程、提升内部效率、市场营销等方面。
Well, so far, if we think about pharmaceuticals where we've seen the early benefits of AI have been in things like the clinical process, getting drugs that have already been discovered through that process, internal efficiencies, marketing, things like that.
真正的巨大机遇在于药物发现。
The real, you know, the big prize out there is drug discovery.
它能发现那些没有人工智能我们就无法解决、或者需要花费漫长得多时间才能解决的问题。
It's finding solutions to problems that we just couldn't get there without AI or would have taken much, much longer.
因此,市场策略实际上是与这些公司建立直接合作,以加速发现过程。
So the go to market is actually direct partnerships with these companies to fast forward the discovery process.
丹尼,这反映了我们正在看到的市场扩展趋势。
Denny, that speaks to the broadening out we're starting to see in the market.
我们还将讨论这一点,即标普500指数在回报方面实际上正面临一些挑战,相对于‘七巨头’中的部分公司。
We're also gonna talk about that, the the fact that we are seeing the S and P 500 actually doing a bit of a challenge in terms of returns versus some of the Mag seven names.
我们是否已经过度推高了英伟达的估值?还是说AI基础设施仍然是2026年最关键的关注点?
How much does it mean that we've already juiced the valuation side of the equation for NVIDIA or is AI infrastructure still going to be the key watchword for '26?
到目前为止,所有数据都表明,对AI基础设施的投资将继续保持强劲。
Well, so far all the data points would suggest that the investments into AI infrastructure are going to continue to be quite robust.
我们完全预期会如此。
And we would fully expect that.
回顾过去,刚才我们谈到阿尔法特和苹果的关系,再看看OpenAI、XAI和Gemini之间的跃迁效应,这非常健康。
So far, just looking back, mean, was talking about Alphabet earlier and the relationship with Apple, just watching the leapfrogging effects you know, with OpenAI and XAI and Gemini, that's very healthy.
只要我们继续看到这种趋势,再回到英伟达,他们上周已经宣布,Rubin已全面投入生产。
And as long as we continue to see that and then bringing back to NVIDIA, NVIDIA now is, you know, they announced last week, Rubin is fully in production.
这将对推理效率与训练产生实质性影响。
That is going to have material impacts as it relates to inference productivity versus training.
而推理是下一个关键环节。
And inference is the next big handoff.
因此,如果你相信我们现在正处于一个这些应用开始大规模部署的时期,推理需求将会飙升。
And so if you're a believer that we are now in this period where we're going to start seeing increasing deployment of these applications, inference is going to go through the roof.
因此,这将继续维持投资水平。
As a result, that's going to continue to sustain the investment profile.
所以我们预计它将保持相当强劲。
So we expect it to be pretty healthy.
现在,当谈到更广泛的标普500指数时,我要说,过去三年随着人工智能真正兴起,我们的一个观点是:这将对所有行业产生深远影响。
Now I will say when you're talking about the broader S and P 500, I mean one of the thesis that we've had for the last three years as AI has really started to take hold is this is going to be very impactful across all industries.
如果你认同这一点,并且那些能够因人工智能部署而实现收入增长、同时又能提升成本效率的公司,其估值倍数应该会表现良好。
And if you believe that and the companies that are able to get revenue lift because of deployment of AI, but then at the same time are able to drive cost efficiencies, that should be pretty good for multiples.
因此,我不惊讶于市场似乎正在提前预期人工智能可能带来的部分收益。
And so I'm not surprised to see the market, you know, sort of anticipating of sorts some of the benefits we might get from AI.
市场对谁必须承担AI基础设施成本以及其债务成本也变得更加审慎了,丹尼。
The market's also got more discerning on who has to pay for the AI infrastructure and what that costs in terms of debt, Denny.
我正在看另一个数字。
I'm looking at, yet another number.
这次穆迪称,到2030年数据中心基础设施建设将耗资3万亿美元。
3,000,000,000,000 this time coming from Moody's saying how much the data center infrastructure build out is gonna cost up until 2030.
我的意思是,麦肯锡已经将这一数字定为7万亿美元。
I mean, McKinsey had already put $7,000,000,000,000 on that number.
但谁来实际支付这笔费用,丹尼,你担心吗?
But is there a worry as to who has to actually pay for it, Denny?
当然,这是个问题。
Well, of of course, there is.
好消息是,大部分资金由超大规模云服务商承担,他们能够产生足够的自由现金流来履行已做出的承诺。
You know, the the good news is the majority of this is being funded by the hyperscalers, and they generate the free cash flows to actually be able to fulfill the commitments that they've made.
他们都在玩一场二十年的长期游戏。
And they're all playing the twenty year game.
我的意思是,我们刚刚看到Meta、谷歌、微软及其他公司之间达成了大量电力协议。
I mean, we've just seen that with the number of power agreements that are being struck between Meta and Google, Microsoft and others.
因此,我们对此感到相当乐观。
And so we feel pretty good there.
显然,债务市场必须保持健康。
Now clearly, the debt markets are going to have to remain healthy.
我们必须继续看到这些公司实现其雄心的进展。
We're going to have to continue to see progress as it relates to the ambitions of these companies.
但还有另一点需要注意。
But there's also something else too.
当你看这些数字时,其中存在大量重复计算。
When you look at those numbers, there's also a lot of double counting in it.
你必须仔细梳理,才能找到真正必要的部分。
You really have to parse it back to get to the real meat of what's necessary.
此外,这种部署速度本身也受到自然限制。
And then also there are natural governors at just how fast this can get deployed.
因此,资本的部署将非常谨慎,这与互联网泡沫时代大不相同,当时大量资本迅速涌入电信基础设施等领域,这些设施很容易就能部署到位。
So the deployment of the capital is going to be pretty measured and that's what makes this very different than say the .com era where so much capital came into areas like telecom infrastructure so fast and it was very easy to get that stuff deployed.
这很难。
This is hard.
看看德克萨斯州阿比林的Stargate项目,试图建设10吉瓦的设施,所需的劳动力、电力、专业知识,这都是巨大的挑战。
You look at what's going on at Stargate in Abilene, Texas and trying to put up 10 gigawatts, the labor that's needed, the power, just the expertise, and, you know, it's it's it's a heavy lift.
因此,这本身也构成了部署速度的自然限制。
And so that itself is a natural governor on just how fast it can get deployed.
这正是我们上周在CES试图推动的事情。
This is what we were trying to pass at CES last week.
嗯。
Mhmm.
丹尼,一个非常快速的基准假设是,需求仍然远远超过英伟达和AMD的供应能力。
The baseline assumption very quickly, Denny, is that demand still outstrips NVIDIA and AMD's ability to supply.
这对我们今年的前景意味着什么?
How does that set us up for this year?
这对基础设施来说是好事。
Well, that's good for infrastructure.
你知道,当你处于这种情况时,即使像台积电这样的公司,你的第二代架构。
You know, because when you're in a situation like that and even a TSMC, for example, where Your second topology.
是的。
Yeah.
是的。
Yeah.
当你只有有限的产能,只能如此缓慢地新建晶圆厂时。
Where where, you know, you only have so much capacity and you can only bring up incremental foundries so quickly.
因此,从定价角度来看,这非常好,尤其是当你处于垄断地位时。
And so as a result, that's really good from a pricing standpoint, particularly when you're the only game in town.
在某些方面,英伟达仍然是唯一的玩家。
And for certain aspects, NVIDIA is still the only game in town too.
丹尼、菲什、奥贾尼斯、亨德森,各位投资者,很高兴你们参与所有游戏。
Denny Fish O'Janis Henderson, investors, great to have you on all the games.
与此同时,派拉蒙正在加大力度支持华纳兄弟探索与Netflix的合并计划。
Meanwhile, coming up, Paramount in ramps up efforts to support a planned merger between Warner Brothers Discovery and Netflix.
媒体合并大戏的最新进展。
The latest in the media merger saga.
接下来是相关内容。
That's next.
这是彭博科技。
This is Bloomberg Tech.
派拉蒙与天空之舞。
Paramount Skydance.
该公司表示,计划提名董事进入华纳兄弟探索的董事会,以投票反对批准与Netflix的合并,同时该公司已提起诉讼,迫使华纳兄弟披露有关Netflix与华纳兄弟合并提案的信息。
Well, it says it plans to nominate directors to Warner Brothers Discovery's board to vote against the approval of a merger with Netflix, of course, and the company has filed a suit to force Warner Brothers to disclose information about the proposed Netflix Warner Brothers tie up.
新墨西哥州的娱乐记者菲利克斯·吉莱特加入我们,讨论这场日益激烈的潜在收购案。
New Mexico's entertainment reporter, Felix Gillette, joins us on what is an ever more hostile potential takeover.
是的。
Yeah.
所以起诉要求更多信息,他们究竟想向投资者证明什么?
So suing for more information, what do they want to really demonstrate to investors here?
我的意思是,到目前为止,他们选择加剧敌意,而不是提高报价。
I mean, at this point, you know, they're choosing to raise the hostility versus raising their bid.
是的。
Yes.
我认为,他们长期以来一直在游说股东加入我们的要约,多次表示我们的报价优于Netflix的报价,指责华纳兄弟董事会在过程中缺乏透明度,现在他们采取这一行动,实质上是要求必须向股东提供信息,以便他们判断哪个报价更好。
And I think, you know, they've been campaigning shareholders, join our tender offer, they've said for weeks now, our offer was better than the Netflix offer, accusing the Warner Brothers board of not being transparent during the proceedings and now they're making this move to essentially say, you have to provide the information to shareholders so they can judge which is a better offer.
而要做到这一点,几个月来最重大、最具争议的问题一直是:如何评估华纳兄弟在将影视工作室和流媒体业务出售给Netflix之前计划分拆的有线电视资产?
And to do that, the big and most contentious issue for months now has been how do you value the cable assets that Warner Brothers is planning to spin off prior to selling the studios and the streaming business to Netflix?
Netflix的报价是每股27美元,仅针对工作室和流媒体分拆业务,而派拉蒙的报价是每股30美元,涵盖全部资产,但基本上将有线网络的价值定为零。
The Netflix offer $27 a share for the studios and streaming spin off the cable networks, Paramount $30 a share for all of it, but basically assigning a value to the networks of $0.
没错。
Right.
这真是一场大戏。
This is a saga.
这件事会随着时间推移逐渐明朗。
It will play out over time.
费利克斯,作为这里的编辑,这个问题对你来说真的很难,但接下来会发生什么?
It's a really difficult question for you as as as a as editor here, Felix, but, like, what happens next?
Paramount 有没有可能通过这种机制强行实施敌意收购?
Is there a real chance that Paramount can force a hostile takeover through this mechanism?
他们会尝试的。
They're gonna try.
你知道的。
You know?
我的意思是,目前很多股东似乎还在观望,不确定支持哪一方。
I mean, I think, like, you you know, the shareholders at this point, a lot of them seem to be sitting on the fence, in terms of who they're going to support.
我认为很多股东都希望 Paramount 能提出更高的报价,超过每股30美元。
I think a lot of them are hoping that Paramount would come in with a higher bid, go over the $30 a share.
但目前为止还没有发生。
That hasn't happened yet.
我认为,戴维·埃里森和派拉蒙一直在努力宣扬一种观点,即奈飞的交易根本不会发生。
I think, again, David Ellison and Paramount have been trying to advocate for this notion that the Netflix deal isn't even gonna happen.
他们面临太多监管问题。
They have too many regulatory issues.
撇开如果派拉蒙的报价被接受,他们在欧洲和美国也会遇到自身问题这一点不谈。
Putting aside the fact that if Paramount offer was accepted, they'd have their own issues in Europe, with The States.
所以,是的,这意味着局势正变得越来越敌对,下一步将是针对董事会成员提起诉讼和代理权争夺。
So yeah, I mean, this continues to just get more and more hostile and this is, the next step is this lawsuit and a proxy fight over the directors.
彭博社的费利克斯·吉莱特,非常感谢。
Bloomberg's Felix Gillette, thank you very much.
另一个重要新闻。
Another top story.
深势科技创始人梁文峰的对冲基金去年涨幅超过50%,成为中国量化基金中表现第二佳的基金。
DeepSeek founder Liang Wenfeng's hedge fund surged more than 50% last year, making it the second best performer among China quant funds.
强劲的表现帮助提升了深势科技自身的可用现金。
The strong performance helped boost the cash available for DeepSeek itself.
让我们连线彭博科技执行主编彼得·埃尔斯特罗姆。
Let's get out to Bloomberg Tech executive editor Peter Elstrom.
你知道,有两个故事,一个是关于量化基金的惊人表现。
You know, there's two there's the Bloomberg story, right, which is about a quant funds, you know, amazing performance.
但在报道中,这一点非常明确。
But in the reporting, it's super clear.
人们解读的结论是,这些基金的收益可以被DeepSeek的创始人获取,并用于为DeepSeek提供更大的研发预算来部署其模型。
The read through that people are making is that that the returns on that, the earnings from that funds performance can be taken by DCS founder and used to give DeepSeek a bigger budget for research and deploying its models.
没错。
Right.
没错。
Right.
确实如此。
That's exactly true.
是的。
Yeah.
这里有一个金融故事,但也有一个科技故事。
There's a so there's a finance story here, but there's also a tech story.
而这个金融故事非常令人印象深刻。
And the finance story is very impressive.
那家对冲基金,作为DeepSeek的诞生地,今年表现极为出色。
The hedge fund high flyer that was really the place where DeepSeek was born had a great year.
他们旗下各基金的平均回报率达到57%。
Returns were 57% on average across their various funds.
这对创始人梁文峰来说非常有利,正如你所提到的。
That is very good for the founder, Liang Wenfeng, as you mentioned.
与市场上其他量化基金相比,这一表现相当出色,同时也让他拥有更充裕的资金去进行新的投资。
It's good performance compared with the other quant funds out there, but it also gives him a much bigger checkbook to be able to go out there and make new investments.
这些投资很可能将用于DeepSeek。
Now that probably would be in DeepSeek.
自从一年前DeepSeek取得重大突破,宣布以远低于OpenAI的成本开发出性能相近的大型语言模型以来,他们已经获得了巨大势头。
DeepSeek has gained a lot of momentum since a year ago when they had their big breakthrough, dropped the bombshell that they had invented this LLM for a fraction of the cost of OpenAI's with very similar performance.
因此,你现在可以看到创始人将更多资金投入DeepSeek,或许会追逐一些在美国特别常见的资本密集型项目。
So now you could see the founder invest more money into DeepSeek, perhaps chase after a little bit of the capital intensive projects that we've seen in The US in particular.
到目前为止,我们还没有在中国看到过这类投资。
We haven't seen those kind of investments in China so far.
或者你可以投资其他类型的AI项目,这些项目可能并不直接与DeepSeek相同。
Or you could invest in other kinds of AI initiatives that may not be directly the same as DeepSea.
而关于DeepSeek的整个说法是,他们仅花费了600万美元,而这篇报道的理论认为,他们可能获得了高达7亿美元的收入。
And because the the whole claim about DeepSea because it only cost them $6,000,000, and the theory in this particular reporting is that maybe they're getting revenues of some $700,000,000.
如果你考虑在这里开展业务的成本,这笔收入将非常可观。
That could go a long way if you're thinking about the costs of doing the business here.
彼得,跟我们谈谈他们为何能如此出色地超越对手。
Peter, talk to us about why they're able to outperform so much.
他们押注的是什么?
What is it they're betting on?
这些量化基金是如何做到如此卓越的?
How are these quant funds managing to be so superior?
嗯,DeepSeek 早期能够做到这一点,部分原因在于当时对人工智能项目存在诸多限制。
Well, part partly with DeepSeek early on, they were able to do this because there were a lot of constraints around the AI initiatives in particular.
我们过去曾写过不少关于 High Flyer 的内容。
And we've written a fair bit about High Flyer in the past.
它们拥有这些量化技术,能够在中國市場上超越同行。
They have these quantitative techniques to be able to outstrip their peers within the China market.
正如你提到的,它们曾是第二大基金。
They were the number two fund, as you mentioned.
此外,中國市場一直非常火爆。
Also, the China market has been on fire.
已有大量投资投入到其他类型的人工智能模型中。
There have been a lot of investments in other kinds of AI models.
DeepSeek 尚未上市,但国内另外两个AI模型已经上市,为它们带来了巨大动力。
DeepSea hasn't gone public yet, but two of the other AI models within the country have gone public, giving them a lot of momentum here.
因此,市场表现非常强劲。
So the market's been very strong.
他们已经能够利用这一点。
They've been able to take advantage of that.
但正如你所说,他们现在拥有的资金是早期启动DeepSeek时的100倍。
But, again, as you say, they now have a 100 times as much money as they used to start up deep DeepSeek in the early days.
他们打算如何使用这笔钱?
What are they gonna do with that money?
他们打算如何投资?
How are they gonna invest it?
当然,目前他们还没有公开谈论此事,但我们预计High Flyer和DeepSeek未来会有更多动作。
They're not talking about this publicly, of course, at this point, but we would expect more things to come from High Flyer and DeepSeek in the future.
我的意思是,想想MiniMax上周五上市了。
I mean, just think MiniMax went public on Friday.
股价上涨了100%。
It was up a 100%.
今天又上涨了15%。
It's up 15% again today.
彼得·埃尔斯特姆为您带来所有亚洲相关资讯。
Peter Elstrom on all things Asia.
我们非常感谢。
We so appreciate it.
谢谢。
Thank you.
接下来,再聊聊亚洲。
Coming up, a bit more on Asia.
中国电动汽车能否规避欧盟的关税?
Could Chinese EVs actually dodge tariffs in The EU?
我们将与Coxswain Automotive的斯蒂芬妮·瓦尔德兹·斯特里蒂讨论全球电动汽车战争。
We're gonna discuss the global electric vehicle wars with Stephanie Valdez Stritti from Coxswain Automotive.
接下来是这个话题。
That's next.
这是彭博科技。
It's a Bloomberg Tech.
每天,数以百万计的客户与我这样的AI助手互动。
Every day, millions of customers engage with AI agents like me.
我们昼夜不停地工作,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,只有即时解答和行动。
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.
欧盟正在考虑对中国出口到该地区的电动汽车设定最低价格。
The European Union is weighing minimum prices for EVs exported to the block from China.
这实际上将取代高额关税。
That actually would replace steep tariffs.
这是否表明,尽管美国敦促欧洲对中国采取强硬立场,但贸易紧张局势却有所缓和?
Is it a sign that we've got some easing trade tensions even as The US presses Europe to take a tough line on Beijing?
今天来讨论这个问题的是考克斯汽车公司行业洞察部门主管斯特凡妮·瓦尔德兹·斯特里蒂。
Here to discuss this is Stephanie Valdez Stritti, director of industry insights for Cox Automotive.
这是我们最近收到的新闻头条之一,可能比亚迪将能够进入欧洲市场,而无需承受吞噬其利润的高额关税。
That was sort of the latest news headline that we get that maybe BYD will be able to sell into Europe and not have to have huge tariffs that eats its margin.
我们能看到它们成功进军欧洲市场并持续扩张吗?
Are we gonna see them able to tackle the European market and continue to scale?
是的。
Yeah.
你知道,如果我们看一下电动汽车市场,中国显然占据主导地位。
You know, if we look at the electric vehicle market, right, China definitely is dominating.
你想想他们的策略就是出口。
You think about their strategy is to export.
对吧?
Right?
他们已经出口了数百万辆汽车,并在提升市场份额方面取得了很大成功。
They've exported millions of vehicles, and they've had a lot of luck in increasing share.
所以英国的情况基本上是一种妥协。
And so The UK, it's basically, you know, it's a compromise.
对吧?
Right?
欧盟、欧洲需要实现这些气候目标。
The the EU, Europe needs the to meet those climate goals.
对吧?
Right?
他们需要提高电动汽车的普及率。
They need to increase their EV adoption.
他们依赖中国提供矿物、电池。
They rely on China for minerals, right, batteries.
所以我认为这是一个不错的妥协,而且我认为中国将继续扩大其在这些市场的份额,不仅限于英国,还包括一些新兴市场。
So I think it's a good compromise, and I think China will continue to, increase their share in those markets, not only UK, but some of the emerging markets as well.
与此同时,我经常在社交媒体上收到的一个最大反馈是,美国市场缺乏比亚迪或中国品牌的车型。
Meanwhile, one of the biggest feedback I often get on my social is about the lack of BYD or Chinese offerings here in The United States.
这种情况会改变吗?还是说他们将永远被排除在外,因为美国市场主要押注特斯拉,同时也在更大力度地抵制电动汽车?
Is that ever gonna come to bear, are they just forever gonna be shut out because it's really bets on Tesla and also pushing away against EVs more
broadly?
broadly?
是的。
Yeah.
你知道,如果你看看美国,我认为这是两党都能达成共识的一点:把中国挡在外面。
You know, I think if you look at The US, that's one thing I think both political parties can agree on is to keep China out.
所以我认为我们现在就处在这个状态。
And so I think that's kind of where we're at now.
未来,也许我们会看到类似的情况发生在日本和韩国身上。
In the future, you know, maybe we could see something that similar happen to the Japanese and the Koreans.
他们前来,建立合作伙伴关系,并建设制造工厂。
They come, they build a partner, and they build manufacturing plants.
但就目前和近期而言,除了我们已有的吉利旗下的沃尔沃和极星之外,我们不会拥有其他品牌。
But for now, the near term, we're not gonna have those other than the ones that we already have under Geely, which is the Volvo and Polestar.
但毫无疑问,消费者在海外旅行时已经接触到电动汽车,体验了中国整车制造商的产品,并发现了其高品质和高科技的特点。
But definitely, I think consumers are seeing EVs when they travel abroad and getting the experience of riding in a Chinese OEM and discovering how good quality, high-tech.
因此,我认为在合适的价格下,这会引发美国消费者的共鸣。
And so I think that it would resonate with US consumers at the right price.
斯蒂芬妮,上周在CES上,英伟达推出了完整的自动驾驶软硬件解决方案,这迅速引发了关于特斯拉的疑问。
Stephanie, last week at CES, NVIDIA outlined a full stack hardware software solution for autonomous driving, and it very quickly raised questions about Tesla.
所以我向黄仁勋问了这个问题。
So I asked Jensen Huang about it.
听好了。
Listen to this.
我认为特斯拉的自动驾驶系统是全球最先进的。
I think the Tesla stack is the most advanced AV stack in the world.
而且我认为特斯拉的自动驾驶系统是全球最先进的。
And and, I think the Tesla AV operations is the most advanced in the world.
而且我相当确定他们早就已经在使用端到端的AI了。
And and I I'm I'm fairly certain that that they were already using end to end AI.
问题是这样的。
So here's the thing.
梅赛德斯将在本季度推出一款搭载英伟达技术、支持点对点无手驾驶的车辆。
Mercedes will ship a vehicle this quarter capable of point to point hands free using NVIDIA technology.
特斯拉终于在市场上遇到了真正的威胁吗?
Does Tesla finally have something real in the market to worry about?
我认为是的。
I think they do.
你想想,就连Rivian这样的其他玩家,在他们的AI日上也公布了他们的计划。
You think about some of the even Rivian's another player, right, that they at their AI day, they announced their plan.
我确实认为竞争非常激烈。
I think definitely there's a lot of competition.
我仍然认为,我们在取得巨大进展,但挑战之一仍然是监管和安全问题。
And I still think, you know, we're making some tremendous progress, but I think one of the challenges continues to be regulation, safety.
但我认为英伟达宣布的技术能够真正识别那些边缘案例,这将有助于提升监管方面的可信度。
But I think the NVIDIA announcement with that technology to really identify those edge cases, which will help with regulations, more credibility.
所以毫无疑问,特斯拉确实面临一些挑战。
So definitely, Tesla has some challenges out there for sure.
我觉得这很有趣的原因是,如果你是特斯拉车主,你就会对全自动驾驶(FST)了如指掌。
The reason I find it fascinating is that if you are a Tesla owner, you know, you you know all about full self driving FST.
这是特斯拉宣传的一部分。
It's a part of the pitch from Tesla.
但汽车市场的绝大多数,至少在美国,可能他们不会考虑这个,好吧。
But the vast majority of the the auto market, at least in America, maybe they don't think about, okay.
我会根据车辆能否做到这一点来购买。
I'm gonna buy a vehicle based on its ability to do this.
你认为从现在开始这会改变吗?
Do you think that changes from here on in?
我认为,从消费者角度来看,更关注的是车内的体验。
I think I think more from consumer is more about the in car experience.
所以我们不断听到所谓的软件定义汽车。
So we keep hearing this software defined vehicles.
我不认为消费者会想着:我想要自动驾驶或完全自动驾驶。
I don't think they're going into, like, I want this autonomous or full self driving.
更重要的是,我在车内能享受到什么样的科技体验,也就是车内的体验?
It's more about what kind of consumer experience can I have with technology in the vehicle, so in vehicle experience?
也许是指具备一些ADAS技术,但我不认为这是主要驱动力。
So maybe it's the ability to have some of that ADAS technology, but I don't think that's the main driver.
更重要的是,这种体验会是什么样的,以及这辆车还能为我提供哪些其他功能。
It's more about what's that experience gonna be and what other options that vehicle provide for me.
在很多方面,这关乎营销,让乘客感受到它的体验,从而持续激发他们对更多功能的需求。
In many ways, it's about marketing, getting people in the cars to understand what it feels like and then sort of perpetuate that demand for wanting more and more and more.
从你的角度看,消费者在安全方面的心理准备到位了吗?
From your perspective, is the consumer mindset there from a safety perspective?
从这个角度来看,法规也到位了吗?
Is the regulation there from that perspective as well?
是的。
Yeah.
我觉得是的。
Think so.
我认为这是关键的一点。
I think that's the one thing.
如果你想想,你知道,当我们谈到机器人出租车时,现在我们已经开始看到Waymo、Zoox以及其他一些公司,包括特斯拉,让消费者体验这些自动驾驶的机器人出租车。
If you think about you know, I think when you think about robotaxis, right, we're starting to see a lot of that with Waymo and Zooke, some other companies, Tesla, getting consumers into those robotaxis, autonomous.
所以他们终于能亲身体验了,但我仍然认为,让消费者接受电动汽车已经很难了,而要让他们接受自动驾驶汽车则更难。
So they're getting to finally experience that, but I still think, you know, it's been hard to get consumers to trans, you know, kinda consider electric vehicles to get into an autonomous vehicle isn't more so.
因此,我认为从消费者的角度来看,自动驾驶汽车在可信度和安全认知方面还有大量工作要做。
So I think there's still a lot of work to be done in terms of credibility and safety perceptions for autonomous vehicles from a consumer perspective.
斯蒂芬妮,我们期待2026年。
Stephanie, look forward to 2026 for us.
你预计在美国的电动汽车领域会有哪些情况?有没有想到任何可能的增长?
What is it you expect to have in in United States, in particular, in in electric vehicles, and any growth that may or may not come to your mind?
是的。
Yeah.
你知道,2025年最终的结果完全符合我们的预期。
You know, 2025, it ended exactly exactly where we thought it would.
我们的报告明天发布,但全年销量预计将同比下降2%,约为130万辆。
Our report comes out tomorrow, but we're gonna be about 2% down year over year, round up to two 1,300,000 vehicles.
但进入2026年,我们已经取消了胡萝卜和大棒。
But going into 2026, right, we got rid of the carrot and the stick.
我们不再有《通胀削减法案》的税收激励。
We no no longer have the IRA tax incentive.
企业平均燃油经济性标准也放宽了,加州的豁免也取消了。
The corporate average fuel economy standards have been less stringent, the Calvert waiver.
但尽管如此,2026年仍将有20多款新的电动汽车上市。
But given all that, we still have 20 plus new EVs launching in 2026.
所以我认为我们仍然会看到增长势头。
So I think we're still gonna see momentum.
这会很难,但我认为消费者将拥有更多产品。
It's gonna be hard, but I think consumers are gonna have more product.
但我认为我们过去经常谈论的一点是可负担性。
But I think the one thing which we've talked about before in the past is the affordability.
如果你看一下当前的产品阵容,提供的电动汽车中超过60%的价格高于65,000美元,这对许多人群来说都太高了。
If you look at the current product lineup, EVs that are offered, over 60% or over $65,000 that would a lot of demographic.
斯蒂芬妮·瓦尔德兹·斯特里蒂,考克斯汽车公司行业洞察总监。
Stephanie Valdez Streety, director of industry insights, Cox Automotive.
非常感谢。
Thank you so much.
接下来,我们将与卢克斯资本联合创始人彼得·赫伯特交谈,讨论该公司最新基金、最大基金,以及硅谷与华盛顿的碰撞。
Coming up, we'll speak with Lux Capital cofounder Peter Herbert about the firm's latest fund, its biggest fund, Silicon Valley meets DC.
这是彭博科技。
This is Bloomberg Tech.
欢迎回到彭博科技。
Welcome back to Bloomberg Tech.
我们将在节目后期深入探讨一个故事。
There's a story we're gonna go deeper on later in the program.
总统表示,如果信用卡公司不遵守10%的利率上限,它们将面临非法活动的指控。
The president saying that if credit card companies don't respect an interest cap of 10% in the year, they will be facing illegal activity claims, basically.
这两家更具科技属性的公司——Klarna和Affirm——在消息公布后股价均下跌。
These are two of the more tech focused names, Klarna and Affirm, both down on the news.
我还注意到,美国运通的股价也显著下跌,这是一个明显的例子。
I would note American Express is also down as an example significantly.
我们稍后会在节目中详细讨论这一点。
We're gonna get to that later in the program.
我们已从谷歌获得确认,该公司已与苹果达成一项多年协议,让Gemini成为下一代AI语音助手Siri的技术基础。
Back to one of the top stories, we've had confirmation from Google that it has entered into a multiyear agreement with Apple for Gemini to be the underpinnings of the next gen AI Siri.
你可能注意到,就在节目开始前约东部时间10:30,这一消息传出后,Alphabet股价大幅上涨,使其市值达到3.4万亿美元。
You see when the news broke just before the program about 10:30 eastern time, a big jump in Alphabet shares that sent market cap $34,000,000,000,000.
随后股价回落,我还想提醒一下,彭博社在去年11月曾报道,苹果公司即将与谷歌达成一项协议,每年支付约10亿美元,以使用这一规模高达1.2万亿参数的模型。
They then faded, and I'd remind you as well that Bloomberg did report in November that a deal was close for Apple to pay Google about $1,000,000,000 a year for use of that very large 1,200,000,000,000 parameter model.
卡罗,还有别的吗?
Caro, what else?
让我们深入探讨一下Alphabet最近的动向,以及它相对于整个市场的表现。
Well, let's dig in on Alphabet on its recent moves, but also where it stands versus the rest of the market.
彭博社股票记者卡门·莱因克一直在深入研究一个观点,即我们或许正在进入更广泛的投资格局。
Bloomberg equity reporter Carmen Reinke has been really digging in to the idea that maybe we're broadening out.
也许‘七巨头’不再是一个稳赚不赔的选择。
Maybe Mag seven isn't just a surefire bet.
去年,Alphabet还是一个稳赚不赔的选择。
Alphabet was a surefire bet last year.
请告诉我们,我们开始在这些关键的巨型科技股中看到哪些新的区分迹象?
Tell us about the discernment we're starting to see among these key mega caps.
是的。
Yeah.
我的意思是,我们去年看到的情况至少在今年头几周仍在继续,赢家和输家变得非常重要。
I mean, what we saw last year is continuing at least in the first few weeks of this year where the winners and losers are really important.
因此, Alphabet 去年是赢家。
So Alphabet was a winner last year.
它是表现最好的股票之一,也是 Mag7 组中仅有的两只跑赢大盘的股票,另一只是英伟达。
It was the biggest outperformer and one of only two stocks in the Mag7 group that outperformed the broader market, the other one being NVIDIA.
因此,它已经牢牢确立了在人工智能趋势中的主导地位,而保持这一势头至关重要,尤其是考虑到我们已经开始看到其他一些公司略显滞后。
So it's really established itself as dominant in the AI trend, and, you know, maintaining that momentum going forward is really important, especially considering that we're starting to see some of these other names lag a little bit.
你知道,它们的整体增长正在放缓。
You know, they're just slowing down in terms of overall growth.
进入财报季,这些趋势预计将继续。
And heading into earnings season, some of those things are expected to continue.
对吧?
Right?
我们只是正面临规模效应的制约。
We're just we're hitting the the law of large numbers.
预计增长将放缓,这可能也会反映在股价上涨上。
Growth is expected to slow, and so that might translate, you know, into stock gains as well.
今天早上对谷歌和苹果以及关于改进Siri的交易消息来说,真是有趣的一天。
Been an interesting morning for Alphabet and Apple and the news flow around the deal to to improve Siri.
我刚刚在思考那个股票代码本身。
And I was reflecting on on literally just the the ticker.
对吧?
Right?
你知道,谷歌股价飙升,市值突破4万亿美元,但随即又回落了。
You know, Alphabet soars and breaches $4,000,000,000,000 market cap, then kind of fade straight away.
今天早上你在市场上看到人们有什么反应?
What did you see in the markets this morning in terms of how people reacted?
是的。
Yeah.
我们看到了谷歌和苹果股价的大幅上涨。
So we saw that big spike for both Google and alpha or sorry, Google and Apple shares.
展开剩余字幕(还有 211 条)
这使谷歌的市值突破了4万亿美元。
It did push Google over 4,000,000,000,000 in market cap.
它最近也超过了苹果,实际上是标普500指数中仅次于英伟达的第二大公司。
It's also surpassed Apple recently, so it's the second largest company in the S and P 500, really, in the market other than NVIDIA.
有趣的是,我们确实看到那些涨幅有所回落。
And it's interesting because, you know, we did see those those gains fade a little bit.
但现在它们又上涨了。
They're up now.
我认为这两只股票目前都处于上涨区间。
I think both stocks are are trading in positive territory.
但整体市场,我认为宏观新闻仍在施加压力。
But the overall market, I think we are still seeing a lot of macro news weigh on this.
今天市场走势一直起伏不定。
It's been a little bit all over the place today.
因此,大型科技股显然仍然至关重要。
So big tech is obviously still super important.
这些股票非常重要。
These stocks are really important.
它们规模太大了。
They're so large.
它们对整个指数的影响很大,但我们也看到市场正在广泛分化。
They, you know, add a lot and can weigh a lot on the broader index, but we also are seeing that there is a broadening out.
人们正在关注其他行业,而不仅仅是科技巨头在驱动市场的涨跌。
People are looking at other sectors, and, you know, it's not just big tech that's always driving the gains and losses in the market.
彭博社的卡门·莱因克,非常感谢。
Bloomberg's Carmen Reinke, thank you very much.
让我们从公开市场转向私募市场,前沿科技领域的兴趣非常高。
Let's go from the public markets to private markets where interest in frontier tech is incredibly high.
Lux Capital为其第九支基金筹集了15亿美元,这是这家风投公司二十多年历史上规模最大的一笔。
Lux Capital has raised $1,500,000,000 for its ninth fund, the largest in the venture firm's more than twenty year history.
Lux Capital的合伙人兼联合创始人彼得·阿巴里在这里与我们一同位于旧金山。
Peter Abare, partner and cofounder of Lux Capital is with us here in San Francisco.
谢谢
Thank you
感谢邀请我。
for having me.
这是令人兴奋的消息。
It's exciting news.
对吧?
Right?
第九支基金,有史以来规模最大的。
The ninth fund, the largest ever.
你们拒绝了大约十亿美元的资金。
You turned away about a billion dollars of capital.
但非常有趣的是它的持续性。
What's very interesting about it, though, is the consistency.
投资主题保持一致。
The themes are the same.
重点是一样的。
The focus is the same.
你打算如何使用这笔基金?
What do you wanna do with this fund?
所以我们将继续做同样的事情,只是加快步伐。
So we will be doing more of the same, but just accelerating that pace.
我们已经在这二十多年里投资于物理、计算和生命科学领域,但始终聚焦于这些不同科学学科的前沿和尖端。
So And we've been investing now for more than twenty five years in physical, computational, life sciences, but really at the cutting edge and the frontier of all these different scientific disciplines.
正如我们所说,科学本身不会自我扩展,因此我们将把投资的重点放在这一方面。
And so as we say, science doesn't scale itself, and so that's where we're going to be putting the thrust of our investment.
彼得,我们很高兴你能来参加这个节目。
Peter, we enjoy having you on the program.
这家公司,你、乔什、迪娜,当你们所有人一起来的时候。
The firm, you, Josh, Dina, when they cut when when all of you come on.
但看起来非常有纪律,非常沉稳。
But it seems very disciplined, very calm.
检查规模,十万以上。
Check sizes, a 100,000 up.
但这很令人兴奋。
This is exciting, though.
这些是发展最快的领域之一。
These are some of the domains that are moving the fastest.
实际上,您是否认为这一届政府允许了您这样做?
And, actually, would you reflect that that this administration has allowed you to do that?
国防领域以及其他与人工智能相关的领域,确实是特朗普政府迅速支持的方面。
It it the areas of defense, other areas adjacent to AI is something that the the Trump administration has moved quickly to support.
因此,在国防、广泛的再工业化以及我们所从事的许多领域,现在确实正处于一个关键时期。
So certainly in areas like defense, kind of broadly, reindustrialization, a lot of the things that we do, it's certainly having a moment.
这实际上反映了正在发生的科技浪潮。
And that's really reflected, one, from the technological waves that are happening.
目前每个人最关心的一件事就是物理人工智能。
One of the things that is top of everyone's mind right now is physical AI.
目前,摩根大通的医疗健康大会正在旧金山举行。
Right now there's also the JP Morgan health care event happening in San Francisco.
你正在见证医疗机器人和计算药物发现领域的巨大突破。
You're seeing huge breakthroughs in medical robotics and in computational drug discovery.
但毫无疑问,纵观世界,这是一个充满不确定性的时代,国防技术因大国重新武装和聚焦威慑而迅速发展。
But absolutely, you look at the world, it's an uncertain place and defense technology has been really soaring as large nation states rearm and focus on deterrence.
彼得,安德鲁,当然,你们做出的关键投资之一,我们上周还邀请了CES的帕尔默·拉基做客节目。
Peter, Andrew, of course, one of the key bets you made, and we had Palmer Lucky on the show last week from CES.
这个基金中,有多少资金是用于加码之前的投资,又有多少是用于完全全新的、来自你们收件箱的初创公司?
Of this fund, how much will be to double down on prior bets, or how much will be to take out totally new bets on new companies coming to your inbox?
因此,这个基金的核心理念是将自己视为一张白纸,从大约40到50个核心持仓的组合开始。
So the spirit of this fund is really thinking about as a a blank slate, starting with the portfolio of probably 40 to 50 core positions.
历史上,我们的一些公司曾从一个基金跨越到另一个基金进行跨基金投资,但这通常并不是我们每支新核心基金所聚焦和秉持的理念。
Historically, we've had companies that have gone from one fund to another, cross fund investing, but it's generally not the the focus and philosophy of each new core fund that we raise.
从十万到一亿。
A 100,000 to a 100,000,000.
你们给自己设定了一个非常广阔的范围。
That's a huge breadth that you're giving yourselves.
你们觉得会在哪些领域感到兴奋并进行投资?
Where do you think you will be excited to deploy?
基础模型有进展吗?
Have the foundational models moved forward?
我们是否看到了更多关于LLM构建中更具体的细分领域的有趣发展,还是更侧重于应用?
Are we seeing more interesting areas of more specific niche parts of the building, the LLMs, or is it more about application?
你们觉得在哪些方面会感到兴奋?
Where are you thinking about getting excited for?
我们的思维模型和框架实际上是将AI从二维推进到三维。
So our mental model and framework really is what we describe as moving from two d, two dimensional AI to three d.
这主要涉及机器人技术、自动化和生物学领域。
And that's really areas of robotics and automation and biology.
但这是第一次看到AI从数字世界走向物理世界。
But it's seeing AI for the first time coming out of the digital world into the physical.
其中一个例子就是这里巨大的机遇。
And just one example of that is really just the momentous opportunity here.
大约90%的美国GDP并非原生数字化的。
Approximately 90% of US GDP is not digitally native.
它存在于物理世界中。
It is in the physical world.
它是建筑业。
It's construction.
它是重工业,这占美国GDP总额3万亿美元。
It's heavy industry, and that's $30,000,000,000,000 of total US GDP.
因此,对于专注于软件、数据及其他特定领域的技术投资者来说,真正存在机会将总可服务市场扩大十倍。
So the opportunity really exist for technology investors focused on software, data, other very specific areas that now have the opportunity to basically 10 x total addressable market.
彼得,你是这家公司的联合创始人
Peter, you're you're a cofounder
了这家公司。
of this firm.
从第一支基金到第九支基金,有限合伙人的构成发生了变化吗?你们的投资者主要来自哪里?
Has the composition of the LPs changed from fund one through to fund nine and where the the interest in you is coming from?
我们最早的信念支持者是一位名叫比尔·康威的先生,他是卡莱尔公司的创始人之一。
Earliest believer in our in in Lux was a gentleman Bill Conway, one of the founders of the Carlisle.
你好。
Hello.
因此,他就是Lux Ventures一号投资人。
And so he was he was Lux Ventures one.
如今,我们代表大学捐赠基金、基金会和州养老金管理着70亿美元的资产。
Today, we manage $7,000,000,000 on behalf of endowments and foundations and state pensions.
所以,从最初的一位个人投资者——尽管他近乎一个机构——到现在服务世界上一些最顶尖的机构,这一转变显而易见。
So, obviously, going from an initial individual, albeit a quasi institution, to now serving some of the world's most elite institutions.
我非常感兴趣你们所看重的核心能力是什么。
I'm really interested in the core competencies you're looking for.
稍后在节目中,我们将邀请人形机器人公司One X,他们已经推出了自己的世界模型。
Later in the program, we we have One X, the humanoid robotics company on right, and they're they're out with their own world model.
但目前的问题是,正如NVIDIA上周明确表示的那样,如果一家公司是人形机器人公司或不同的物理AI产品提供商,他们几乎会包揽硬件和软件的所有工作,那么在这种情况下,还有什么值得寻找的呢?
But the problem right now is that NVIDIA, as was made very plain last week, will they're basically coming out and saying, we'll do it all for you on the hardware and software side if you're a humanoid robot company or or a different physical AI offering, what's left to look for in that regard?
信不信由你,不会只有一家公司统治所有其他公司。
Believe it or not, there will not just be one company that rules all others.
因此,存在许多不同的机会。
And so there are a number of different opportunities.
即使像NVIDIA这样举足轻重的公司,硅芯片领域仍有其他新兴力量,软件和训练方面更是如此。
And even with NVIDIA, as prominent as they are, there are other up and comers on the silicon side, certainly on the software side, training.
有许多公司正在从不同角度切入物理AI的实现,从这里旧金山的物理智能,到Luxe投资组合公司Applied Intuition——它正在打造移动领域的操作系统。
There are a number of different companies that are attacking, again, physical, implementations of AI from physical intelligence, that's here in San Francisco, to Applied Intuition, which is a luxe portfolio company, which is really creating an operating system in mobility.
因此,机遇非常丰富。
So there is abundant opportunity.
这些机会是全球性的吗,彼得?不仅限于消费市场,还包括这些公司的诞生地?
And is that global, Peter, in terms of not just the consumer base, but where these companies are getting birthed?
你们的布局范围有多广?
How broad do you go?
One X并不是一家美国公司。
Well, One X, not an American company.
这个市场上有许多全球竞争对手。
There are many global competitors in this marketplace.
即使在AI领域,我们也是总部位于日本的Sakana AI公司的投资者。
Even in AI right now, we're an investor in a company called Sakana AI based in Japan.
在上一期基金中,我们在欧洲进行了大约十几项投资。
We've made probably about a dozen or so investments across Europe in the last fund.
因此,技术是全球性的。
So technology is global.
它分布广泛。
It's broadly distributed.
旧金山仍然是中心,但你会看到它从全球众多大型研究机构中涌现出来。
San Francisco remains the epicenter, but you will see it popping up coming out of a lot of large research institutions across the globe.
令人着迷。
Fascinating.
感谢彼得·阿伯特,感谢您作为合伙人、联合创始人以及Lux Capital最新基金的贡献。
Peter Abert, we thank you, partner, cofounder, Lux Capital on the latest fund announcement.
接下来,我们将讨论这家公司——OneX,Neo机器人的制造商。
Coming up, we talk about that company, OneX, the maker of Neo.
该公司表示,其新的AI模型将使家用机器人具备从零开始学习新任务的能力。
It says its new AI model will give its home robot the ability to learn new tasks from scratch.
我们将采访这位首席执行官。
We'll talk to the CEO.
这里是彭博科技。
This is Bloomberg Tech.
我们为您带来机器人领域的最新消息。
We've got news for you from the world of robotics.
humanoid机器人Neo的制造商One X Technologies对其One X世界模型(一个基于物理的AI视频模型)发布了更新。
One X Technologies, the maker of the humanoid robot Neo, has an update for its One X world model, an AI video model grounded in physics.
据该公司称,此次更新将使Neo能够通过简单的文本或语音指令执行新的AI功能。
According to the company, the update will make Neo capable of executing on new AI abilities with a simple text or voice prompt.
即使机器人之前从未执行过该任务,也是如此。
That's even if the robot has zero experience of doing the task before.
加入我们的是One X Technology的首席执行官兼首席技术官伯恩德·伯恩迪希。
Joining us is 1x Technology CEO and CTO, Bernd Berndich.
欢迎回到Bloomberg Tech。
Welcome back to Bloomberg Tech.
谢谢。
Thank you.
我想先从简单的话题开始。
I I wanna keep this really simple to start.
这个模型更新后,Neo能够执行哪些之前无法首次完成的任务?请举个例子。
What is an example of a task that the update to the model allows Neo to do that it couldn't have done for the first time without the update?
对我来说,关键在于那些不在数据集中的事情,但机器人依然能给出合理的应对方式。
So, I mean, to me, it's all about, like, things just being anything that you don't have in your dataset, but still being able to have a sensible approach.
我第一次真正看到这一点时,是一个简单的问题,只是问机器人:‘你能从白板上把便签纸拿下来并读一下吗?’
So the first time I really saw this, and it was a simple question, just asking the robot, Hey, can you pick the Post it note down from the board and read it?
但它可能从未
But it may never
见过便签
have seen a Post
纸。当然,我们没有任何关于使用机器人从板子上拿起便签并查看和阅读其内容的训练数据。
it And it's note of course, we don't have any training data on us using robots to pick Post it notes off a board and look at it and read what's on it.
但它实际上能做得很好。
But it can actually do it really well.
但这还不是通用人工智能。
And it's not AGI yet.
也有一些它失败的例子。
There are examples where it fails.
但它所具备的是对任何事物都能采取非常合理的应对方式,而这正是学习的核心。
But what it does is the ability to have this very sensible approach to anything, which is the cornerstone of learning.
因为现在,你只需要让机器人通过在现实世界中实际实验来自主学习所有这些任务。
Because now all you need is the robots teaching themselves how to do all these tasks by actually experimenting and doing this in the real world.
这个模型有多好?
This model, how good is it?
你知道吗,上周在CES上,英伟达大力宣传了使用其开源模型的优势。
You know, last week at at CES, NVIDIA extolled the virtues of using their open models.
在Neo的大脑中,是一个英伟达的推理芯片。
And inside Neo, the brain of Neo is is an NVIDIA inference chip.
对吧?
Right?
那为什么不直接使用英伟达的模型呢?
So why not just use NVIDIA's model too?
我们与英伟达有着非常深入的合作,并在我们的技术栈中大量使用了他们的技术。
We have a very deep collaboration with NVIDIA, and we do use a lot of their technology in the stack.
对我们来说,关键在于具身智能。
I think to us, it's all it's much about the embodiment.
Neo在过去十年中被设计成尽可能接近人类。
Neo has been designed, really, over the last decade to be as close to a human as possible.
因为如果你取用世界上所有的知识,比如你在YouTube或任何视频内容上能看到的一切,用这些数据训练一个模型,然后你希望机器人去拿起杯子或打开门,但如果这个机器人与人类的相似度不够,这种方法就失效了,因为你不会以同样的方式完成这些任务。
Because if you take all of the knowledge that we have in the world, like everything you can see on YouTube or any kind of video content, and you train a model on this, and now you want to pick up the cup or open the door, if your robot is actually not similar enough to a human, this doesn't work anymore because you wouldn't do the task in the same manner.
这一点在One X上非常独特。
And this is quite unique to One X.
我认为有几件事让我们与众不同:我们的机器人在与世界互动时非常接近人类,而且它们是安全的,因此可以真正尝试去做一些事情。
Think there's a couple of things that really puts us apart, and it's that we have robots that are so close to humans in how they interact with the world, but also that they're safe, so they can actually try to do something.
即使它们失败了,世界依然安然无恙。
And if they fail, then the world is still okay.
你并不希望因为机器人试图开门而把门划伤。
You don't want your door to be scratched because there was a robot trying to open it.
这让我们处于一个独特的类别中,我们的AI方法可能与当前的标准略有不同,也正因如此,我们对这些世界模型感到无比兴奋——终于,机器人技术的扩展开始遵循与Sora或竞争对手的Vio等大型视频模型相同的视频预训练扩展规律。
And this puts us in a category where our approach to AI is probably a bit different, like what the standard is now, and also why we're so very excited about these world models where finally we see robotic scaling following the same scaling laws as, for example, video pre training from the big video models like Sora or Vio from our competitors.
伯恩德,再多谈谈安全性吧,因为你提到机器人如何避免划伤门。
Bernd, go in on the safety a little bit more because you're talking about how it stops your door getting scratched.
但对许多人来说,真正的安全问题在于:当机器人开始自主学习完成任务时,风险也随之而来。
But for many, it's about, well, when robots start learning to do things themselves, therein lies the safety issue.
你们设置了哪些安全防护措施?
What guardrails do you have in place?
我们在这方面有一些令人兴奋的工作,很快就会公布。
So we have some exciting work on that that will come out soon.
我们在安全方面正采取正确的措施,遵循相关标准,并协助制定部分标准,同时确保我们的安全工作经过外部审计。
We are doing the proper things on the safety side with respect to following the standards, helping develop some of the standards, but also ensuring that we, for example, have externally audited work of our safe safety work.
由独立的第三方机构进行审计,当然。
By an independent By by independent third party, of course.
但对我来说,安全包含多个层面。
But to me, safety has many layers.
我们称之为被动固有安全,即人类实际上非常安全,因为我们必须主动努力才能伤害彼此。
You have what we call the passive intrinsic safety, where humans are actually very safe in the sense that we have to actively work to hurt each other.
我们通常不会意外地伤害彼此。
We don't hurt each other by accident, usually.
这一点我们已经融入了机器设计中。
This is something we build into the machine.
同样,你必须做到柔软、合规、轻量化、低能耗,这样才能在人群中和环境中保持安全。
The same kind of just you have to be soft, compliant, lightweight, low energy, so you can be safe among people and in the environment.
当然,还有AI对齐,这同样重要,你需要确保无论执行什么任务,都能始终选择最安全、风险最低的路径。
Then of course you have the AI alignment, which is equally important, and how do you ensure that you can always take the safest, least risky path for whatever task you want to achieve.
这也是这些世界模型令人无比兴奋的地方。
And this is also something that's incredibly exciting about these world models.
是的。
Yeah.
它们对世界的理解如此深刻,对物理世界的运行机制也如此清晰,因此你不仅能问‘我该如何完成这个任务’,还能问‘我该如何以最安全的方式完成这个任务?’
That they understand the world so well, and also how the physical world works, that you can not only ask for how am I going to do this task, but how am I going to do this task in a manner that is as safe as possible?
可能会出现哪些问题?
What are the things that could possibly go wrong?
然后模型会主动推理,比如:
And then the model actively reasons about, like, hey.
这是我能预见到的可能出错的地方,所以我选择这条路径,这是最安全的路径。
Here are the things that I can visualize going wrong here, so I'm gonna take this path here, which is the safest path.
对。
Right.
除了最终加速NEO在现实世界中的部署外,所要解决的问题在于现实世界数据收集的负担。
The the the problem being sold for other than ultimately accelerating deployment of NEO in the real world is the burden of real world world data gathering.
目前,很多注意力都集中在仿真和合成数据上。
A lot of focus right now on on simulation and synthetic data.
稍微谈一谈这个话题。
Just talk a little bit about that.
我们的部分观众是行业人士。
Some of our audience is is our industry.
他们更偏技术,想知道你们是如何用这个特定模型实现的。
They're more technical, and they want to know how you did it with this particular model.
对。
Yeah.
所以,如果简化到最根本的部分,就是如果你的实体机器人足够接近人类,那么我们从视频中学到的这些知识实际上能很好地迁移过来。
So I think, really, if boil it down to the simplest parts, it is if your embodiment, if your robot is close enough to a human, then all of these learnings that we have from video, they actually transfer pretty well.
一旦你能做到这一点,你的机器人就能在你能提出要求的任何任务上接近胜任,你的智能不再依赖于你能从人类那里收集的数据量。
And once you can do that, and your robot can now approach almost any task as long as you can ask for it, your intelligence doesn't scale with the amount of data you can collect with humans anymore.
它实际上取决于你部署的机器人数量,因为你现在只需要足够多的机器人去实际执行这些任务。
It actually scales with the number of robots you've deployed, because now you just need enough robots actually trying to do all these things.
而执行这些任务是有用的,对吧?
And trying to do these things are useful, right?
因此,你不仅产生了有用的工作成果,而且在这个过程中,机器人不断学习并迅速提升。
So you also produce useful work, but in the process, the robot learns and it quickly gets better.
现在,你只需要让足够多的机器人遍布社会,执行足够多样的任务,从而获得广泛的数据覆盖,这样你就能在通往通用智能的扩展道路上取得非常显著的进展。
And now you just want enough robots across society doing enough different tasks so that you get a very large data coverage, and then you're progressing very, very well on your scaling loss towards general intelligence.
但这现在已经不再依赖于必须通过远程操控来让人类收集数据。
But it's now independent of actually having to use humans to gather the data through a teleoperation.
远程操控。
Teleoperation.
没错。
Exactly.
伯恩特·贝内克,One X的首席执行官兼首席技术官。
Burnt Bernek, CEO and CTO of One X.
很高兴你再次回到彭博科技节目。
It's great to have you back here on Bloomberg Tech.
非常感谢。
Thank you very much.
卡罗,还有许多其他新闻头条。
Caro, plenty of other news headlines.
是的。
Yeah.
埃德,现在是谈论科技的时间了。
Ed, it's time now for talking tech.
首先,Monmeta任命了一位前特朗普总统高级顾问担任新设立的高级管理职位。
First up, Monmeta has appointed a former top adviser to president Trump to a newly created senior management role.
迪娜·鲍威尔·麦科密克将指导公司的AI基础设施工作,包括为大型数据中心项目争取政府和投资者的未来资金。
Dina Powell McCormick will guide the company's AI infrastructure efforts, including securing future funding from governments and investors for massive data center projects.
此外,Melmetta表示,为遵守澳大利亚新出台的儿童社交媒体禁令,已关闭了近55万个账户。
Plus, Melmetta also says it has shut down almost 550,000 accounts in Australia to comply with the country's new social media ban for kids.
这项法律于去年生效,规定Facebook和Instagram等平台必须禁止16岁用户注册账户,否则将面临最高3300万美元的罚款。
Now the law, came into effect last year, mandates that services like Facebook and Instagram block 16 year olds from having accounts or face fines of up to $33,000,000.
而Anthropic则正积极进军医疗领域,推出允许患者和医生通过其AI聊天机器人访问医疗信息的工具。
And Anthropic, well, it's making a push into health care with tools that allow patients and doctors to access medical information on its AI chatbot.
这项新服务符合美国医疗隐私法规,允许医疗机构和用户处理受保护的健康数据。
Now the new offering is compliant with US medical privacy regulations, allowing providers and consumers to field protected health data.
埃德。
Ed.
好的。
Okay.
接下来,随着总统特朗普呼吁将信用卡利率上限设为10%并持续一年,信用卡发行机构正高度紧张。
Coming up, credit card issuers are on edge as president Trump calls for a 10% interest rate cap on credit cards for one year.
我们将在下一条为您详细介绍。
We have the details next.
这里是彭博科技。
This is Bloomberg Tech.
沃尔玛正与谷歌母公司 Alphabet 合作,在 Google Gemini 平台上提供人工智能增强的购物服务,这是该零售商将其技术应用扩展至服装、日用品乃至食品娱乐等各业务领域的一部分。
Walmart is partnering with Alphabet to offer AI enhanced shopping on Google Gemini's platform, part of the retailer's race to apply the technology across its operations from apparel and consumables to entertainment of food products.
未来几个月,客户将能够在 Gemini 的浏览器或移动应用中购买商品。
Customers will be able to purchase items on Gemini's browser or mobile app in the coming months.
卡丽?
Carrie?
让我们再多聊聊消费者、科技与支付领域,比如特朗普总统正推动将信用卡利率一年内上限设为10%。
Let's talk a bit more about the consumer and and tech and payments, for example, because president Trump is pushing for a one year cap on credit card interest rates at 10%.
这一举措可能颠覆信用卡行业,并对主要的“先买后付”服务商产生连锁影响。
It's a move that could upend the credit card industry and also ripple through major buy now, pay later providers.
负责消费金融报道的纽伯格的佩吉·史密斯就在现场。
Newberg's Paige Smith, who covers consumer finance, is right here.
看。
Look.
如果信用卡因10%的利率上限而无法再服务广大消费者,这是否会引发人们转向其他融资方式?
Is this going to send shock waves and people towards alternative forms of financing if, for example, these credit cards just can't serve the broad consumer that they used to if 10%
如果上限是10%呢?
is the cap?
这是个非常好的问题。
That's a really good question.
我认为现在还难以确切判断消费者对此的反应,因为这还处于早期阶段。
I think we are a bit early to exactly see how consumers are responding to this because it's still early days.
需要明确的是,特朗普总统在周五晚上以及昨晚的表态,实际上只是称信用卡公司——也就是发卡机构和大型银行——应当采取这一措施。
Just to be clear, president Trump's statement on Friday evening and then again last night was really just saying that that that credit card companies, essentially issuers and big banks, should be doing this.
但他实际能用来实施这一提案的法律手段仍非常不明确。
The legal levers that he can actually pull to implement such a proposal would are still very unclear.
我们尚未看到任何正式的提案。
We haven't seen any sort of formal proposal.
因此,说金融科技公司——比如SoFi、Affirm或Klarna之类的机构——会填补这一空白,还为时过早;尽管它们的股价似乎已经对此作出反应,并展现出在这方面的潜在机会。
So to sort of say that there would be fintech, maybe fintech alternatives, sort of folks along the lines of SoFi or Affirm or Klarna actually stepping into the void is a bit premature, but although the shares have sort of seemed to react and seem to be identifying some potential opportunities along those lines.
佩吉,总统声称一些信用卡提供商的利率高达28%到30%,正如他所说,而且消费者可能甚至没有意识到自己正在被收取30%的利息。
Paige, the the president's claim is that that some credit card providers are charging 28 up to 30% as he put it, but also that the consumer might not even realize that they're being charged interest at a rate of 30%.
在我们的报道和证据中,我们是否知道某些信用卡提供商确实存在这种情况?
In our reporting and and in evidence, do we know that to be true among some credit card providers?
我的意思是,我仅代表我个人发言。
I mean, I'll speak for myself.
我认为今天早上和周末这确实是一个很好的提醒,每个消费者都应当检查自己的信用卡利率。
I think it would be fair to say that it was a good reminder this morning and over the weekend to be checking interest rates on credit cards as every consumer should be.
所以我认为,消费者在刷卡、挥卡或使用信用卡时所承担的费用究竟有多高,这一点确实不是广为人知的事实。
So I I think it is fair that it's not a widely known fact of exactly how much it costs consumers to be swiping, tapping, using their credit cards across the board.
你很好地提醒我们,这仍处于早期阶段,任何拟议的利率上限都需要国会通过法案才能实施。
You do a great job of reminding us that this is still early stages, enacting any sort of proposed cap would take, I believe, an act of congress.
但请让我们向前看,一些关键高管对此潜在举措发表了哪些回应。
But push us forward to what some of the key executives have said in response to this potential move.
当然。
Most definitely.
我上节目前看到一句让我印象深刻的话,是SoFi的首席执行官安东尼·诺托,我记得他说过,SoFi的发展前景令人兴奋。
A quote that stood out to me just before I came on air was I saw that the SoFi CEO Anthony Noto, I believe the quote was giddy up in terms of the opportunities for SoFi.
SoFi确实向消费者提供信用卡,但这家公司最出名的还是学生贷款再融资和消费者的个人贷款。
SoFi actually does offer credit card to its consumers, but it's really a firm that's kind of best known for student loan refinancing and also personal loans for consumers.
这大概就是整体的基调。
So that's kind of a flavor of the tone.
一些高管积极行动,而另一些则持观望态度,想看看这一举措会对他们的业务产生怎样的实质性影响或带来哪些机遇。
Some executives are striking while others are maybe kind of hanging back a bit and seeing substantively what impact this could have or what opportunities it could present for their businesses.
彭博社的佩吉·史密斯,非常感谢。
Bloomberg's Paige Smith, thank you very much.
本期《彭博科技》节目到这里就结束了,卡莉。
That does it for this edition of Bloomberg Tech, Carrie.
本周刚开始就有这么多内容。
So much already as we kick off this week.
别忘了收听我们的播客。
Don't forget to check out our podcast.
你可以在终端上找到它。
You find it on the terminal.
它在苹果、Spotify 和 iHeart 在线销售。
It sells online on Apple, Spotify, and iHeart.
回到旧金山和纽约,不再在拉斯维加斯,这里是 Bloomberg Tech。
Back in San Francisco and New York, no longer in Vegas, this is Bloomberg Tech.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕,帮助你打破语言障碍,轻松听懂全球优质播客。