Meta从元宇宙转向人工智能设备

AI代理变得越来越令人印象深刻。

AI agents are getting pretty impressive.

你可能甚至没有意识到你正在听的正是一个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24/7全天候工作，以解决客户问题。

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.

分配。

Allocation.

对吧？

Right?

我们一直痴迷于GPU和AI加速卡，但在服务器设计中，你需要拥有与之相当甚至更多的CPU。

We've been obsessed about GPUs and AI accelerator cards, but you need to have an equivalent, if not greater number of CPU in the design of the server.

而英特尔和AMD在这方面做得很好，抢占了一定的市场份额，而英伟达则试图主导整个系统。

And that's an area that Intel and AMD have done well to get some market share where NVIDIA is trying to dominate the whole system.

显然，这是重大的举措。

And clearly, big moves.

卡罗，对于分析师上调评级来说，这几乎是超乎寻常的举动，尤其是英特尔目前表现尤为强劲。

Outsize moves almost, Caro, for an analyst upgrade, but Intel in particular soaring high at the moment.

是的。

Yeah.

想关注那些贡献大的关键点。

Wants to watch big points contributors.

让我们稍微换个话题，谈谈他们最近的一些投资。

Let's just flip gears a little bit and talk about where's one out on some of their investments of late.

对冲基金投资者。

Hedge fund investors.

他们自金融危机后从未有过这么好的状况。

Well, they haven't had it this good since the aftermath of the financial crisis.

行业追踪机构对冲基金研究公司收集的数据表明，2025年该行业迎来了自2009年以来最好的一年。

Data compiled by industry tracker hedge fund research showing that back in 2025, the industry posted its best year since 2009.

平均收益约为12.6%，但有些甚至高出许多。

Gains are about 12.6% on average, but some is growing much higher than that.

彭博社对冲基金记者赫玛·帕尔默现在加入我们。

Bloomberg hedge fund reporter, Hema Palmer, joins us now.

我们通常认为他们都拥抱了人工智能，量化基金表现不错，但谁在2025年真正脱颖而出呢？

And we like to think about the way in which they've all embraced AI and the quant funds are doing well, but who really shone in '25

为什么呢？

and why?

是的。

Yes.

这是一个非常有趣的年份，因为不同策略、不同规模的基金普遍表现非常好。

So, you know, it's such an interesting year because funds across strategies, across sizes generally did very, very well.

当我们看那些大赢家时，你会注意到一些专注于科技的公司。

And when we look at the big winners, you're looking at some of the tech focused firms.

如果你看看鲸鱼漫步，他们的表现相当不错。

If you look at Whale Walk, they did quite well.

如果你看看光街基金，他们已经从2022到2023年那段艰难时期中恢复并超越了历史高点。

If you look at Light Street, they got over their high watermark from the really difficult 2223 period.

你看到桥水基金的一些最佳回报了。

You're seeing some of the best returns over at Bridgewater.

如果我们看看宏观策略

If we look at macro

有一个表现相当不错。

One that's pretty good.

对吧？

Right?

没错。

Exactly.

出色的业绩。

Excellent numbers.

因此，几乎在每种策略中，你都能看到赢家。

So kind of across strategies, you're seeing winners across pretty much every strategy.

D。

D.

E。

E.

肖，多策略策略中超过20%，这对投资者和对冲基金来说非常有趣且是个好消息。

Shaw, over 20% in the multistrack strategy, so pretty pretty interesting and good news for investors and hedge funds.

是的。

Yeah.

海玛，十年前我在彭博基金学院读书、在电视台工作时，对冲基金就是这么做的——从混乱中创造现金。

Hema, when I was growing up in the in the the Bloomberg School of Hedge Funds ten years ago on the on the TV desk, That's what hedge funds do, make cash out of chaos.

这不正是它们的全部意义吗？

Isn't that the whole point?

我们手头的这批数据中，有没有什么独特之处能说明它们的做法有所不同？

Is there anything unique in in this bucket of data that we've got about what they're doing differently?

人工智能发挥了作用，人才崛起并带来了卓越的表现。

AI playing a role, talent stepping up and delivering big performance.

是的。

Yes.

所以今年的亮点在于我们看到的波动性——这种波动频繁进出，为参与各种交易机会提供了绝佳时机。

So what makes the year interesting is the volatility that we're seeing is the kind that goes in and out pretty frequently and provides great times to get in on different types of trading opportunities.

因此，人工智能为许多专注于科技的基金带来了巨大助力。

So AI has been a great lifter for a lot of the tech focused funds.

但如果你看看Viking Global，这家公司像其他老虎基金如Kotu那样在科技和人工智能领域的投入并不多，表现也没那么好。

But if you look, for example, at Viking Global, which Viking doesn't do as much tech and AI as, say, other funds like other Tiger Cubs, Kotu, for example, they didn't do as well.

它们的涨幅仅有约8.6%，在科技基金中排名靠后。

They were only up about 8.6%, putting them lower in the pack of tech funds.

Tiger Global通常被视为科技投资者，它们在多头头寸上表现不错，但空头头寸侵蚀了这些收益，导致整体涨幅也低于7%，排名靠后。

Tiger Global, which we typically think of as a tech investor, they did well in their long book, but their short positions ate into those gains, putting them also lower in the pack, more than 7% gain.

所以，人人都赚了钱，但真正的问题是谁实现了两位数的增长。

So, you know, everyone made money, but the real question is who made the double digit gains.

你看到Lightstreet从许多热门股票中获得了不错的回报。

And you see with Lightstreet good returns from a lot of those sort of popular names.

彭博社的艾玛·帕尔默带来对对冲基金的分析。

Bloomberg's Emma Palmer with the hedge fund breakdown.

非常感谢。

Thank you very much.

今天，摩根大通拉开了大型银行财报周的序幕。

Now JPMorgan kicked off a week of big bank earnings today.

其中一个关注点是人工智能对这些银行支出的影响。

One area of focus, the impact AI could have on spending at those banks.

首席执行官杰米·戴蒙坚称，该银行将投资人工智能以提升效率，并跟上竞争对手的步伐。

CEO Jamie Dimon insisting the bank will be spending on AI to drive efficiencies, but also keep up with competitors.

摩根大通的年度支出数据出人意料，比市场预期高出90亿美元，人们想知道这是否归咎于人工智能。

There was a surprise in the expenses number for JPMorgan of the year, 9,000,000,000 above what the street saw, and they wanted to know if AI was to blame.

让我们邀请亚历山德拉·穆萨·维西达。

Let's bring in Alexandra Musa Visidae.

她是Evident公司的联合首席执行官兼联合创始人，Evident是一个用于衡量和追踪金融服务行业人工智能采用情况的平台。

She's co CEO and cofounder of Evident, a platform that benchmarks and tracks AI adoption across the financial services sector.

这是一场非常有趣的财报电话会议。

It was really interesting earnings call.

通常与杰米·戴蒙对话，但这次是富国银行的迈克·梅耶直接说：嘿。

Often it's with Jamie Dimon, but this was Wells Fargo's Mike Mayo basically saying like, hey.

这些支出看起来非常高，而你们一直在大力宣传摩根大通的AI。

Those expenses seem very high, and you've been talking up AI a lot at JPMorgan.

这是否直接相关？

Is that directly correlated?

杰米·戴蒙的观点是，我们确实会投入。

And Jamie Dimon's point was kind of like, we will be spending.

但并没有你们想象的那么多。

It's just not as much as you think.

我们必须投入，因为所有人都在投入。

We have to because everyone is.

摩根大通在银行的AI实施竞赛中领先吗？

Is JPMorgan ahead in this AI implementation race in the banks?

是的。

Yeah.

好吧，谢谢。

Well, thank you.

谢谢你，埃德。

Thank you, Ed.

是的。

Yeah.

今天看到你们的讨论很有意思，但总的来说，正如你们所知，我们追踪了北美和欧洲最大的银行在AI部署方面的进展，而摩根大通明显处于领先地位。

It was it was interesting to see the the your discussions today, but sort of stepping back and overall, as as you know, we map the biggest banks in North America and Europe on their AI deployment, and JPMorgan is very much leading.

他们已经连续三年半领先。

They're outnumber one three and a half years in a row.

因此，他们在AI部署方面处于领先地位。

So they are leading on AI deployment.

他们在将AI部署并融入整个银行方面也处于领先地位。

They are leading on, you know, deploying and embedding AI throughout the bank.

他们确实在科技上投入了大量资金，而且其中增长最快的部分正是用于AI，以便在未来逐步削减这些成本。

They do also spend a lot on tech, and a growing proportion of that is on AI precisely to be able to take over time some of these costs out.

我们看到的许多部署都集中在内部流程上，以提升效率，同时也带来了一些收入增长。

A lot of the deployment we do see is in those are in those internal processes and to create the efficiency gains, but also some on revenue uplift.

亚历山德拉，很多问题都关于这些效率提升何时实现。

Alexandra, a lot of this question is about when those efficiency gains come.

它们已经实现了吗？

Have they already come?

是在2026年，还是从你的角度看，这仍然是更长期的？

Are they coming in '26, or is it more still longer term from your perspective?

是的。

Yeah.

有意思的是，我们已经进入了生成式AI旅程的第三年，这个旅程最初充满热情和测试，现在正逐步进入实际应用。

I mean, it's it's interesting, right, because we're three years into the Gen AI journey, and that was a a journey with that started with a lot of excitement and testing and now use cases moving into production.

但如今，似乎正在发生一场真正的结构性转变，即开始全面将AI嵌入到银行的各个部门、各项职能和业务线中。

But there seems to be a real sort of shift in the tectonic plates now where it's looking at fully embedding it across the banks and every across the bank and every function and line of business.

要做到这一点，你需要一个为规模化而设计的平台架构。

And with that, you need to have platform architecture that's built for scale.

我们明确看到，效率方面已经开始产生投资回报。

And what we are definitely seeing is is some return on investment coming through on the efficiency side.

因此，自动化KYC流程，以及进入资产管理、投资银行等领域等等。

So automation of KYC processes and, you know, going into the asset management side, the investment banking, and so on and so forth.

所以实际上是贯穿整个银行的每一个环节。

So really across the entirety of the bank.

因此，从实际回报和投资回报率的角度来看，这仍然处于早期阶段。

So it's still early days in terms of the the actual sort of return and the ROI.

我认为，要真正全面实现这些效果，还需要几年时间。

I still think that there's some years to it would take for this to really fully come through.

此外，我们还有AgenTik AI用例正在进入生产阶段。

And then we've also got AgenTik AI use cases coming, know, going into production.

而我认为，真正的影响力将在这里显现。

And that's where I believe we're gonna see the real impact.

但AgenTeq和完全自主的AgenTeq用例要完全落地并实现我们所知的那些根本性、大规模的投资回报，还需要三到四年时间；Gen AI的全面影响需要几年，而Gen Tech AI的影响则可能需要三到四年才能完全显现。

But it is gonna take another three to four years for AgenTeq and fully autonomous AgenTeq use cases to be fully embedded and to see that, you know, really fundamental and sizable ROI that we know is coming, but it is gonna take a couple of years for the Gen AI to fully come through and maybe three to four years for Gen Tech AI impact to fully come through.

有趣的是，迈克·梅奥关于AI的问题基本上出现在电话会议的末尾。

It was interesting that basically that question about AI from Mike Mayo was at the tail end of the call.

在此之前，有很多关于Jamie Dimon与苹果公司信用卡合作的问题，但我们也在更广泛地讨论他们投资银行业务的失误以及信用卡业务。

Before that, there was myriad of questions for Jamie Dimon around the Apple onboarding with the credit cards there, but we're also talking more broadly about their investment banking miss and indeed credit cards.

但是，亚历山德拉，让我们继续向前推进。

But, Alexandra, push us forward.

我们还有大量其他财报即将密集发布。

We've got a whole host of other earnings coming thick and fast.

在哪些地方他们会最频繁地谈到人工智能？

Where will they be talking about AI most abundantly?

嗯，你知道，现在有很多事情正在发生。

Well, so, you know, it's it's a it's it's a time when there are a lot of things going on.

因此，这次人工智能可能不会成为焦点。

So AI might not be front and center this time.

但过去几年它一直是焦点。

It has been over the years.

在财报电话会议上，一些被提出的问题涉及股东对银行现状的强烈关注。

Some of the questions that have been asked in the earnings calls, there's a lot of shareholder pressure to understand, like, where the bank is.

我们确实看到，银行已经明显将人工智能列为高度优先事项，它已成为高层管理团队的一部分，首席执行官们也充分认识到，人工智能必须贯穿整个银行并成为核心。

We've definitely seen a shift in terms of banks being much more clear that it is a high priority, that it is part of the senior leadership team and the CEOs are, you know, fully understanding that it has to be embedded across the bank and has to be core.

因此，我预计不仅在财报电话会议上，而且在今年早些时候的投资日、投资者日以及银行发布合作伙伴关系等新闻稿中，这一趋势仍将持续。

So I expect that there will continue to be not only in the earnings calls, but as we've seen throughout the year in investment days and or investor days and in, you know, press releases when partnerships and so on are getting, you know, produced by the banks.

关于人工智能的讨论是一个持续进行的过程，贯穿全年。

The discussion on on AI is an ongoing one and throughout the year.

但接下来我们将迎来几家大型银行的财报电话会议。

But we've got some earnings calls coming through from the big banks.

你知道，高盛去年底也发布了重大公告，推出其'One GS 3.0'计划，旨在从上到下、从内到外全面转型银行，彻底融入人工智能。

You know, Goldman Sachs has also made a big announcement late last year about their one GS three point three point zero program, is really about fully transforming the bank end to end, top down, bottom up for, you know, fully embedding AI.

因此，我很期待看到这一计划将带来哪些具体数据。

So it's gonna be really exciting to see what numbers might be associated with that.

我认为，领先的银行将重点讨论人工智能的投资回报率。

The leading banks are going to be talking about ROI, I believe.

在过去一年中，已经有很多与之相关的具体数据被提及。

There has been a lot of of talk with numbers associated with it in the last year.

我认为这些数字可能只是冰山一角，我们将在2026年看到它们被更新和提升。

I think that these numbers are probably tip of the iceberg, and we're gonna see these being updated and upgraded in 2026.

亚历山德拉，我们只剩下三十秒了，但摩根大通在吸引和支付AI顶尖人才方面有多具竞争力？

Alexandra, we just have thirty seconds, but how competitive is JPMorgan in attracting and paying top talent in AI?

非常有竞争力。

Very yeah.

它在吸引顶尖人才方面非常出色。

It's it's very good at getting top talent.

它是一家以AI为优先的知名机构。

It is a place that is known for putting AI first.

早在2017年，杰米·戴蒙就明确表示，我们将成为一家AI优先的企业。

Jamie Dimon was very clear about that all the way back in 2017 where he said we're gonna be an AI first enterprise.

他们能够吸引顶尖人才。

They are able to attract top talent.

他们正与科技行业展开竞争。

They're in competition with the tech sector.

每个人都想寻找那些能够真正重新设计复杂系统的顶尖人才。

Everyone's looking for that top talent that could really do the difficult systems rethink.

大家都在关注那些从科技公司挖走人才的现象，这些人才能够端到端地重新思考流程，彻底改变整个系统。

Everyone's thinking about the shark tanking, taking talent from tech companies that can rethink the processes end to ends and completely change the system.

请记住，这里的科技只占问题的10%。

Remember, the technology here is just 10% of the problem.

90%的问题在于彻底重新思考流程。

90% of the problem lies in the the sort of rethinking of the processes entirely.

你几乎需要先建立一个数字孪生体，重新设计流程，然后再将其重新投入运行。

You almost have to build a digital twin, rethink the process, put it back in.

这种人才正是摩根大通能够吸引到的。

That talent is what JPMorgan can attract.

亚历山德拉·穆索维扎德。

Alexandra Musovizadeh.

很高兴能再次和你交流。

Always great to catch up with you.

Evident的联席首席执行官兼联合创始人。

Co CEO, cofounder of Evident.

感谢您。

We thank you.

接下来，我们将为您带来彭博社对百度首席财务官何凯的独家访谈，内容涉及中美竞赛。

Coming up, we'll bring you Bloomberg's exclusive conversation with Baidu CFO, Henry He, on The US China race.

来自纽约和旧金山，这里是彭博科技。

From New York, from San Francisco, this is Bloomberg Tech.

每天，数百万客户与我这样的AI代理互动。

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

我们全天候工作，随时掌握最新信息。

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

我们快速高效，但极其耐心。

We're fast and effective, but incredibly patient.

我们基于Sierra平台构建，这是领先的AI驱动客户体验平台。

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

没有等待音乐，只有答案和行动。

No hold music, just answers and action.

访问 sierra.ai 了解更多信息。

Visit sierra.ai to learn more.

那就是 sierra.ai。

That's sierra.ai.

国防部长佩特·赫格塞斯昨晚在参观SpaceX星舰基地时，批评了国防工业的规避风险文化，并赞扬了埃隆·马斯克。

Defense secretary Pete Hegseth lamented the defense industry's risk averse culture and praised Elon Musk during a visit to the SpaceX Starbase launch site that was last night.

赫格塞斯宣布计划将马斯克的Grok人工智能平台整合到国防部系统中，并使美军成为所谓的‘人工智能优先’的作战力量。

Hankseth announced plans to integrate Musk's Grok AI platform into the defense department system and to make the US military a, quote, AI first war fighting force.

请听一下。

Take a listen.

我们需要在这里直言不讳。

We need to be blunt here.

我们再也承受不起等待十年，让我们的传统主承包商交付下一个‘完美’系统，结果却发现它不仅多年延迟，成本还高达应有价格的十倍。

We can no longer afford to wait a decade for our legacy prime contractors to deliver the next perfect system, only to find that it's delivered years behind schedule and cost 10 times what it should.

胜利需要一套新策略。

Winning requires a new playbook.

埃隆用他的算法写下了这套策略。

Elon wrote it with his algorithm.

质疑每一个需求，删除那些愚蠢的要求，然后全力加速。

Question every requirement, delete the dumb ones, and accelerate like hell.

你知道吗，卡莉，人们都在说为什么PEGCEF会在SpaceX？

You know, Carrie, people are saying why is PEGCEF at SpaceX?

嗯，仅在开发方面，SpaceX就拥有40亿美元的政府合同，而且它是少数几家被授权将五角大楼的敏感卫星送入轨道的运载工具之一，比如猎鹰九号和猎鹰重型火箭。

Well, SpaceX has $4,000,000,000 in government contracts alone just for development, but one of few carriers, Falcon nine, Falcon Heavy, that's authorized to take Pentagon sensitive satellites to orbit.

关于Grok的时机也很有意思。

It's also interesting timing regarding Grok.

对吧？

Right?

我的意思是，谷歌AI已经整合进国防部系统了，但在这个时候加入Grok却意义非凡，因为此前人们对它生成的图像等内容一直存在诸多担忧，这标志着在这方面关系的显著推进和深化。

I mean, already Google AI is integrated within defense department, but but to add Grok at this moment when there has been a lot of concern about what it has been producing in terms of imagery and the like, a notable step forward in that respect and that relationship and deepening at least.

一周前，总统在True Social上表示，他希望对那些CEO表现不佳的国防公司实施薪资上限。

A week ago, the president put on True Social that he wanted to cap salaries of defense companies who weren't performing for their CEOs.

今天，我们得知政府正收购一家顶级五大连锁国防承包商——L3哈里斯的大量股份。

Today, we have news that the government is taking a significant stake in a top five defense contractor, l three Harris.

我们需要了解什么？

What do we need to know?

是的。

Yeah.

而且他们打算将导弹解决方案业务分拆上市。

And the idea that they're gonna be spinning off an IPO of the unit, the missile solutions business.

L3哈里斯将重新掌控导弹解决方案业务，但到2026年，他们计划将其分拆出去。

They're gonna l three Harris is gonna retrain control of the missile solutions business, but 2026, they're looking at spinning it off.

我们了解到，国防部将投资10亿美元购买可转换优先股。

And we understand that the Department of Defense is set to invest $1,000,000,000 in convertible preferred security.

我的意思是，这推动股价上涨了2%，一度上涨了6%，埃德。

I mean, it's lifted the shares up 2%, up 6% at one point, Ed.

与此同时，在亚洲，百度首席财务官何俊杰与彭博社的史蒂文·恩格尔讨论了该公司超过140亿美元的AI支出。

Meanwhile, in Asia, Baidu CFO, Henry He spoke with Bloomberg's Steven Engel about that company's AI spending of over $14,000,000,000.

听听这段采访。

Listen to this interview.

在去年11月的财报电话会上，我们实际上披露了相关数据。

In the November earning call last year, we actually disclosed one numbers.

自2023年财报发布以来，我们在AI领域的投资已超过1000亿元人民币，这是一项巨大的单方面投入。

Since earning came out, back in 2023, we have invest over RMB100 billion in terms of AI investment, which is a huge investment just on the one side.

另一方面，我们也看到了这笔投资的丰厚回报。

On the other end, we do see a great return on that.

例如，今年第三季度，我们的AI云收入同比增长了128%。

For example, now, our cloud revenue increasing about 128% on a Y o Y basis in Q3 for the AI cloud.

此外，我们在AI搜索转型方面也实现了超过200%的增长。

And also, we are seeing over kind of 200% growth on AI search transformation.

例如，去年，我们传统搜索中由AI生成的内容仅占3%。

So, for example, last year, we only have 3% of content we are generating for the AI for our traditional search.

但目前，这一数字正在以18%的速度增长。

But right now, this number is increasing by 18%.

我们看到传统搜索向AI新搜索的转变呈现出三位数的增长。

And we're seeing kind of three digits increasing on the traditional search to the AI new search.

此外，关于自动驾驶出租车，正如我提到的，每周已实现超过二十五万次行驶。

And also for the robotaxi, as I mentioned, it's actually delivered over a quarter million drives every week.

我们看到这一数字正处于加速增长的态势。

And we're seeing that number is on accelerated growth paces.

因此，我认为我们的投资正在看到良好的商业化趋势和回报回报。

So I think the investment, we are seeing the good trends of the monetization and the computing on the return.

您如何看待中国AI的发展与我们一直密切关注的‘七巨头’以及美国OpenAI等公司的发展之间的差异？

How do you see difference between how AI is evolving in China versus, you know, what we've all been following very closely in the Mag seven and what's happening OpenAI and others in The United States?

实际上，人们担心资本配置过度，无法获得相应的回报。

Essentially, there are concerns about over allocation of capital and not getting the kinds of returns.

因此，可能存在潜在的泡沫。

So there could be potentially a bubble.

中国也存在泡沫担忧，但中国的AI领域同样呈现出分化态势。

China, there are there are bubble concerns as well, but there's a bifurcated AI space in China as well.

像你们、腾讯、字节跳动和阿里巴巴这样的大公司，以及像MiniMax等在香港市场崭露头角的新锐企业，这些大语言模型已经开始产生影响。

The big boys like you and Tencent and ByteDance and Alibaba, and then there are new up and comers that are hitting the market in Hong Kong like MiniMax and others, those LLMs that are starting to make a dent.

但你们拥有雄厚的资金基础。

But you guys have the capital base.

由于过度投资，中国在AI领域的投资是否也在形成泡沫？

Is there a bubble forming as well in the investment in China because of the over investment?

我想称之为城市的两极。

I want to call it two tails of the city.

首先，我认为如果要分解驱动因素，大概有四个方面，对吧？

Firstly, I think if you want to decompose the drivers, I think there are four things, right?

数据、模型、算力，以及应用，对吧？

The data, model, computing power, as well as applications, right?

所以，如果你分析这四个方面，中美确实有着非常不同的背景，对吧？

So I think if you analyze these four different things, US China does have a lot of different contexts, right?

例如，目前在美国，大量投资都集中在基础设施上。

For example, right now in The US, there's a lot of investment on infrastructures.

但今天你看，中国的电力和电缆网络在过去几年里已经建设完毕。

But today, you look at electricity, cable network in China is already being built in the past few years.

所以如果你真的想比较基础设施方面的投资回报率或收益率，我认为中国在这方面确实有一定的优势。

So if you really wanna compare on IOI or ROE on on the infrastructure side, I think China does have certain advantage on that.

而且在过去十年的移动互联网时代，中国积累了大量的数据。

And also in the past ten years of the mobile internet age in China, there's a huge accumulation of data.

这些数据对今天的推理变得至关重要，因为几年前，大家都在竞争基础模型。

That data become important for today's inferences because a few years ago, everyone was competing for foundation model.

关于你的问题，我认为，在我看来，任何拥有计算能力、数据和应用场景所有权闭环或全栈能力的公司，都能在人工智能领域实现可持续发展。

To your question, I think in my view, any company have a closed loop or full stack of the computing power, data and ownership of application user cases will be sustainable on AI.

百度CFO亨利·海在那里。

Baidu CFO, Henry Hay there.

说到全栈，我们接下来要谈谷歌母公司Alphabet。

Talking of full stack, we're gonna talk about Alphabet next.

接下来，Wedbush的丹·艾维斯将加入我们，讨论苹果选择谷歌AI来驱动Siri的这笔交易及其影响。

Coming up, Dan Ives of Wedbush joining us to talk about that deal, the impact of Apple choosing Google's AI to power Siri.

接下来是这个话题。

That's next.

这是彭博科技。

This is Bloomberg Tech.

今天发布了关于苹果计划使用谷歌AI驱动Siri语音助手及其他相关内容的新研究。

New research out today on Apple's plan to use Google's AI to power Siri voice assistant and much more.

彭博情报指出，这笔交易凸显了谷歌张量芯片相比竞争对手的成本优势。

Bloomberg Intelligence is writing that the deal highlights the cost advantages of Google's tensor chips over rivals.

分析师们也对这一消息表示欢迎，而彭博去年年底就曾报道过此事，记得吗？

Analysts also welcome confirmation in the news, which Bloomberg reported late last year, remember.

这笔交易后，Wedbush重申了其对苹果350美元的目标股价，这是业内最高的。

Now Wedbush reaffirmed its $350 price target as the highest on the street for Apple after the deal.

我很高兴邀请到Wedbush的董事总经理、高级股票分析师丹·艾维斯，他正是提出这一数字的人。

I'm pleased to welcome Dan Ives, managing director, senior equity analyst at Wedbush, who has that number.

我们先来谈谈苹果的好处。

So let's talk about the benefits for Apple first.

为什么再次确认350美元的目标价？

Why the reaffirmation of 350?

这对苹果智能有什么好处？

Why is this good for them in Apple Intelligence?

你看，他们一直有一个隐形的AI战略。

Look, they've had an invisible AI strategy.

我的意思是，回溯到最近几次的WWDC，你感觉就像迈克尔·J·福克斯。

I mean, you go back to, like, the last WWDC's, you felt like Michael J.

回到未来。

Fox, Back to the Future.

对吧？

Right?

当时根本没有任何关于AI的讨论。

There was really nothing there talking about AI.

最后，这是一个巨大的进步。

Finally, this is a huge step forward.

而且这不会在内部发生。

And it's not going to happen internally.

必须由谷歌的Gemini来推动。

Had to happen from Google, Gemini.

你看，如果他们从未赢得D或J诉讼，这件事就不会发生。

Look, and it comes down to, like, if they never win D or J suit, this doesn't happen.

所以这说明他们正在一砖一瓦地逐步构建。

So it speaks to just brick by brick, they're finally building it.

我认为这将对苹果今年的估值起到关键作用。

And I think this is gonna be instrumental to the valuation for Apple this year.

如果你看看发布的声明，它明确表示，目前这将嵌入到苹果的基础模型中。

If you look at the statement that was put out, it was very much for the time being, this is gonna be embedded in Apple foundation models.

他们的想法是，要与之拉开差距。

The idea is that they they pull away.

他们能够掌握自己的基础模型，事实上让Siri变得真正有用。

They are able to get a grip on their own foundational models and indeed Siri becoming actually useful.

你认为他们最终会与谷歌拉开距离吗？

Do you think they will distance themselves from Google over time?

你看。

Look.

我觉得我参加NFL季后赛的概率，都比苹果现在内部做点什么的可能性要大。

I think there's a better chance of me playing NFL playoffs than Apple doing something internally at this point.

看吧，现实是，这还是一场高风险的扑克式谈判，关于苹果与谷歌之间更广泛的交易，我认为他们最终不得不达成协议。

Look, the reality is this is also a game of high stakes poker negotiation that's going on too in terms of a broader deal with Google that I think they're ultimately going to have to do.

我认为这将是独家的，因为我的观点是，未来会有一种订阅服务，免费增值模式，你也需要为开发者提供一个平台，这一切都归结于人工智能革命。

Look, I think this will be exclusive, and because my view is that there's going be some sort of subscription service, freemium, you also need a platform for developers, and it comes down to the AI revolution.

苹果正从场边、看台上观察着这一切。

Apple's watching it from the sidelines, from the stands.

他们必须入场参与，这就是为什么我们说，每股7500美元的价值，正随着苹果最终在消费级人工智能革命中落地而被加到这个故事里。

They need to get into the game, and that's why we talk about $7,500 per share that this adds to the story as they execute on the consumer AI revolution finally going through Cupertino.

丹，你曾多次主张苹果收购Perplexity，如今确认Gemini将成为Siri的基础，你还认为这样的交易有必要吗？

Dan, you had advocated on more than one occasion that Apple buy perplexity with this confirmation of Gemini being the underpinning of Siri, is such a transaction still necessary to your mind?

是的，埃德，这是个很好的观点，但很多这种想法都基于当时谷歌正在面临的司法部诉讼。

Yeah, and Ed, it's a great point, but a lot of that was based on as the DOJ suit was going on with Google.

那时有一段时间，苹果需要采取一些行动。

There was a sort of period there where Apple needed to do something.

现在，显然Perplexity已经来去匆匆了。

Now, obviously, perplexity came and went.

他们显然不会这么做，但一旦谷歌赢得司法部诉讼，这就像是绿灯亮起、全面启动，现在正是大型科技公司行动的时刻，我认为这解释了苹果为何达成这笔交易。

They're clearly not going to do that, but once Google won the DOJ suit, that was sort of like green light, lights on, and now it's sort of go time for big tech, and I think that speaks to why Apple did this deal.

你看谷歌Gemini所做的事情，回溯到一年前，真是令人惊叹，我认为苹果正在下这个赌注。

And look, you look at Google Gemini, what they've done, you go back a year ago, it's phenomenal, and I think Apple is making that bet.

相对于2025年，2026年苹果在人工智能领域将会有截然不同的面貌。

It's gonna be a much different 2026 than 2025 relative to Apple when it comes to AI.

卡拉刚刚指出，苹果在声明中非常谨慎地说明了Siri的哪些功能将由Gemini提供支持。

Again, Kara just made the point that Apple was very careful in the statement to explain which parts of Siri's functionality would be underpinned by by Gemini.

另一种看法是，他们是否仍需做大量工作，才能让Siri成为一个全方位的AI助手？

The other way of looking at is to say, have they still got a lot of work to do to make Siri a sort of all encompassing AI assistant?

我的意思是，这就像攀登珠穆朗玛峰一样艰难，因为现实是，他们已经失去了大量开发者和人才。

I mean, look, it's an Everest like uphill battle because the reality is they've lost so many developers, they've lost so much talent.

你已经看到很多变化了。

You've seen a lot changing.

而且，埃德，你也和任何人一样清楚，库比蒂诺的DNA，很大程度上是希望内部自主研发的。

And look, Ed, you know it as well as anyone, the DNA of Cupertino, a lot of that is supposed to come internally.

我认为库克已经意识到，形势已明，这也印证了我的观点：库克不会在任内留下这个过渡期的烂摊子。

And I think Cook recognized, look, writing on the wall, it speaks to my point that Cook is not going to leave his CEO and leave this sort of in transition.

他必须把AI战略理顺。

He has to get the AI strategy sorted out.

这将是一个持续改进的过程，但有24亿台iOS设备，15亿部iPhone。

It's going to be a work in progress, but 2,400,000,000 iOS devices, 1,500,000,000 iPhones.

在我看来，消费级AI革命最终将源自库比蒂诺，而这就是朝这个方向迈出的第一步。

It's my view that the consumer AI revolution ultimately comes through Cupertino, and this is a first step in that direction.

我们显然会在春天看到更多关于它的信息，当然还有WWDC大会。

We'll obviously see more about it in the spring and, of course, more WWDC.

有些人可能会说，保持轻量级AI策略非常精明，不必在数据中心上投入巨额资金，还能强调隐私优先。

Some might say it's incredibly savvy to be AI lite, to not have to make all the investments in the data centers, to be able to say, look, privacy first.

我们将更多地依赖端侧AI。

We're gonna lean much more on edge AI.

这在未来随着模型和复杂性的提升，真的会成为苹果的制胜策略吗？

Is that actually gonna be a winning formula for Apple in the in the future and and as models and sophistication develops?

是的。

Yeah.

你看。

Look.

但苹果的情况也大不相同，因为它拥有无与伦比的用户基础，以及在消费市场的独特定位。

But, also, Apple is in a much different situation because of the unrivaled install base, because of where they play in the consumer.

因此，从资本支出的角度来看，苹果和微软、谷歌以及其他大型科技公司并不一样。

That's why, like from a CapEx perspective, they're not Microsoft, Google, many other big tech players.

但我认为他们终于达到了临界点。

But I think they finally hit the point.

实际行动开始了。

Rubber met the road.

他们必须做点什么了。

They need to do something.

他们将不得不逐步投入资金，但基本上会依赖谷歌作为合作伙伴，因为你不能让第四次工业革命——过去四十年里最大的科技转型——来去无踪，而他们却未能从中获利。

They're going to have to incrementally spend, but they're basically going to rely on Google as a partner because you cannot have the fourth industrial revolution, biggest tech transition in the last forty years, come and go, and they don't monetize.

我认为他们清楚地认识到了这一点，这对股价至关重要。

And I think that's something that they recognize front and center, and it's key to the stock.

我的意思是，现实一点讲，一年后，股价可能会从当前水平跌至3.50美元或更低。

I mean, realistically, like, a year from now, stock's $3.50 or lower from here.

这一切都取决于人工智能以及他们能否从中获利。

It's all based on AI and them monetizing it.

丹·艾维斯，韦德布什公司的董事总经理，对苹果给出了业内最高的目标价3.50美元。

Dan Ives, managing director at Wedbush with the street high price target $3.50 on Apple.

本期Bloomberg Tech Caro就到这里，真是精彩的一期。

That does it for this edition of Bloomberg Tech Caro, and what an addition it was.

确实如此。

It was.

我的意思是，我们一开始听Nasa Tegler讲述这如何展现Alphabet的实力，最后又谈到这关乎苹果的未来。

I mean, we started with Nasa Tegler telling us about how this really shows Alphabet's prowess, then we finished on how it's Apple's future.

别忘了收听我们的播客。

Don't forget to check out our podcast.

你可以在终端、苹果和Spotify以及iHeart上找到它。

You can find it on the terminal as well as online on Apple and on Spotify and iHeart.

听好了，Bloomberg。

Listen, Bloomberg.

Bloomberg Tech正在全美直播，Caroline Hyde在纽约，Ed Ludlow在旧金山。

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

这

This

接下来是彭博科技。

is Bloomberg Tech coming up.

Meta和依视路陆逊梯卡计划到今年底将人工智能智能眼镜的产能翻倍，以应对不断增长的需求。

Meta and Essilor Luxottica cub double production capacity for AI powered smart glasses by the end of this year based on growing demand.

此外，微软试图缓解消费者对数据中心电力成本上升的担忧，承诺支付足以覆盖公司成本的公用事业费率。

Plus, Microsoft seeks to quell consumer anxiety of the power price impact of data centers, pledging to pay utility rates that will cover the company's costs.

我们来讨论一下。

We discuss.

Klarna首席执行官塞巴斯蒂安·施穆克沃斯基就特朗普总统呼吁将信用卡年利率上限设为10%发表看法。

And Klarna CEO Sebastian Schmuckowski weighs in on president Trump's call for a 10% interest rate cap on credit cards for the year.

首先，我们关注市场动态，尽管最新公布的通胀数据CPI降至2.6%，市场仍承压。

First, we turn our attention to the markets that despite that cooler inflation print CPI coming in with 2.6, we're looking at markets under pressure.

自然，这延续了昨日的涨势。

Naturally, it raising yesterday's gains.

纳斯达克100指数下跌了百分之零点二。

We're off by two tenths percent on the Nasdaq 100.

这些变动并不剧烈，但我们在地缘政治动荡背景下仍看到一些风险规避情绪。

These aren't massive moves, but we are still seeing some risk aversion amid the geopolitical turmoil.

而且我们是否真的会迎来美联储的暂停加息。

And whether or not we do indeed get any sort of fed pause.

目前看来，市场似乎已将降息计划搁置一旁了，埃德。

This seems as though the rate cut has been put to the back burner by the markets at the moment, Ed.

今天早上有很多关于Meta的新闻。

A lot of meta news this morning.

彭博社的最新消息指出，Meta与Essilor Luxottica的谈判正围绕将Ray-Ban Meta的产能从目前的1000万单位提升至2000万单位。

The latest from Bloomberg is that talks with SL Luxottica are around doubling capacity for Ray Ban Metas to 20,000,000 units from 10,000,000 units current installed capacity.

如果需求强劲，据我们的消息来源称，他们可能在今年达到3000万单位的产能。

If demand's good, they're talking about maybe, according to our sources, 30,000,000 units in the year.

这推动了Essilor Luxottica在巴黎的股价上涨，但与此同时，其他消息显示Meta将削减元宇宙部门的员工人数，卡罗，这给Meta股价带来了下行压力。

Now that drove shares of SLO Luxottica up in Paris, met it to the downside with other news, the confirmation of recent reporting that they're gonna be trimming head count from the metaverse, Caro.

还有微软的这一则消息。

Then there's the Microsoft piece of news.

微软提出了一个五点计划，旨在承担与数据中心相关的电力成本。

Microsoft has a five point plan to basically take responsibility for the cost of electricity associated with data centers.

主要的一点是覆盖其成本，但该计划的其他部分还包括与数据中心可能影响电价的社区保持透明度，我们在这里的彭博社已经对此做了大量报道。

The main one is to cover its costs, but other parts of that are pieces of transparency with the communities where those data centers might impact pricing, and we've done so much reporting on that here at Bloomberg.

让我们连线彭博社的科技与产业政策记者玛吉·伊斯顿，她正在华盛顿。

Let's get out to Bloomberg's tech and industrial policy reporter, Maggie Easton, who's in DC.

你当时参加了微软的发布会。

You were there at the Microsoft presentation.

他们说了什么？这个五点计划的具体内容是什么？

What do they have to say, and what are those five points in the plan?

是的。

Yes.

微软的布拉德·史密斯今天早上在华盛顿就这项五点计划发表了讲话。

So Brad Smith of Microsoft was in Washington this morning talking about this five point plan.

我认为这其中最重要的一点无疑是电力成本。

So I think the largest piece of this is, of course, electricity costs.

因此，他们表示将与当地电力公司合作，确保微软承担其数据中心带来的任何成本增加。

So they've said that they're gonna work with these local power utilities to ensure that Microsoft is paying for any increase in costs that they're responsible for.

当然，这里有个前提，即他们并不保证电费不会因通货膨胀或其他原因上涨。

Now, of course, the caveat there is they're not ensuring bills won't go up due to inflation or other causes.

但根据特朗普总统昨天的要求，他们表示，如果成本增加是由微软建设数据中心引起的，他们将确保为此买单。

But what they're saying in accordance with what president Trump asked for yesterday is that if the cost is due to Microsoft coming in with a data center, they're gonna ensure that they're paying for that.

是的。

Yeah.

特朗普总统在社交媒体上表示，不希望消费者承担这笔费用。

President Trump in a few social saying that they don't want consumers to pick up the tab.

那么，这笔费用还可能落在哪里？

Where else is the tab potentially going to fall?

因为这不仅仅是关于电价的问题。

Because this isn't just all about electricity prices.

还有其他资源也可能面临压力。

There's other resources that are likely to be stretched.

当然。

Definitely.

所以微软今天早上还谈到了另一个话题，即地方税收基础。

So another thing that Microsoft was talking about just this morning is the local tax base.

历史上，当这些数据中心项目入驻时，微软和其他超大规模企业通常会获得某种地方激励措施。

So historically, when these data center projects come in, Microsoft and other hyperscalers have received some sort of local incentive.

而微软现在表示，他们正在放弃这种做法，希望确保能增加这些地方社区的税收基础，以支持学校建设，并且在某些情况下甚至不再接受地方激励，尽管他们仍对州级和联邦级激励持开放态度。

And Microsoft is now saying they're moving away from that, and they want to make sure that they're increasing the tax base in these local communities to support the schools and, yeah, just making sure that they're not accepting local incentives in some cases even though they are still open to state and federal incentives.

玛吉·伊森，来自微软活动的最新报道，非常感谢你的分享。

Maggie Eason, with the latest from that Microsoft event, we so appreciate it.

现在，让我们转向这一问题的商业层面。

Now let's turn to the business side of the equation.

南希·滕勒是拉弗尔·滕勒投资公司的首席执行官兼首席投资官。

Nancy Tengler is with our CEO, CIO of Laffer Tengler Investments.

你看。

And look.

当你考虑对微软以及这些数据中心建设的广泛敞口时，它们现在在多大程度上需要考虑选民和消费者？

When you're thinking about the exposure you have to Microsoft and these data center build outs more broadly, how much they need to factor in the voter, the consumer right now?

嗯，我觉得我觉得

Well, I think I think

它们提前应对这一点非常明智。

it's super smart for the them to get ahead of this.

我肯定它们让总统先宣布，但在纳德拉的领导下，微软一直擅长预判趋势。

I I'm sure they let the president announce it first, but Microsoft has always been good at anticipating under Nadella.

我认为成为社区的好公民很重要。

And I think being a good citizen in the community is important.

但这并不是我们第一次听到反对声音。

But it's not the first time we've heard resistance.

当特斯拉迁往德克萨斯州时，就遭遇了大量抵制。

Tesla got a lot of pushback when they moved to Texas.

它们接管了一座城市。

They took over a city.

他们收购了人们的房屋。

They bought people's homes.

这具有破坏性。

It disruptive.

我认为我们正在看到类似的情况，尽管在数据中心方面程度较轻，但重要的是，这正是在转型经济中会发生的事情。

And I think we're seeing the same thing, I would argue to a lesser extent with data centers, but importantly, you know, this is what happens in a transformative economy.

我们在清洁能源推广中也见过这种情况。

We saw it also with clean energy push.

风力发电机，你知道的，太阳能电池板，人们不喜欢它们出现在自家后院。

Windmills, you you know, solar panels, people don't like that in their backyard.

因此，他们提前行动，开展善意之旅，承担部分成本，帮助社区，我认为这非常重要。

So the fact that they're getting ahead of it, having a goodwill tour, taking on some of the costs, helping the community, I think that's very important.

有趣的是，这恰好发生在Meta全面投入Meta Compute的第二天，他们招聘了新人才，思考如何从主权和领导力角度应对计算需求。

It's interesting that it comes the day after Meta went all in with Meta Compute, hiring new talent, thinking about the way in which they're gonna navigate the need for compute from a sovereign perspective, from a leadership perspective.

但是，南希，你认为我们看到多少公司真正抓住了这个机会？还是说我们在建设热潮中已经进入了某种泡沫状态？

But, Nancy, how much do you think that we are seeing companies dive in on this opportunity enough, or are we in some sort of bubble territory when it thinks about the build out?