人工智能作为新的全球力量？

欢迎收看《市场观点》。

Welcome to Thoughts on the Market.

我是迈克尔·泽祖斯，摩根士丹利全球研究副主管。

I'm Michael Zeezus, Morgan Stanley's deputy head of global research.

我是史蒂芬·伯德，全球主题与可持续研究主管。

And I'm Stephen Byrd, global head of thematic and sustainability research.

今天，人工智能是否正成为地缘政治权力的新支柱？

Today, is AI becoming the new anchor of geopolitical power?

现在是纽约时间2月27日，星期五，中午十二点。

It's Friday, February 27 at noon in New York.

所以，史蒂文，在最近的印度人工智能影响峰会上，美国提出了一个愿景，旨在通过与所谓的美国人工智能栈整合，推动全球人工智能的采用，即所谓的‘真实人工智能主权’或战略自主。

So, Steven, at the recent India AI Impact Summit, The US laid out a vision to promote global AI adoption built around what it calls, quote, real AI sovereignty or strategic autonomy through integration with the American AI stack.

但来自全球南方的几个国家，以及可能部分欧洲国家，似乎对依赖专有系统持怀疑态度，他们担心控制权、可解释性以及数据所有权问题。

But several nations from the global South and possibly parts of Europe, they appear skeptical of dependence on proprietary systems, citing concerns about control, explainability, and data ownership.

目前面临的关键问题不仅仅是技术政策，更是全球权力结构、经济分层的未来，以及主权国家能否切实构建出超越美国和中国的竞争性替代方案。

And it appears at stake isn't just technology policy, it's the future structure of global power, economic stratification, and whether sovereign nations can realistically build competitive alternatives outside The US and China.

所以，史蒂文，你当时在场。

So, Steven, you were there.

你一直描述着美国与全球南方乃至部分欧洲国家在AI战略获取方面日益扩大的鸿沟。

And you've been describing a growing chasm in the AI world in terms of access to strategies between The US and much of the global South and possibly Europe.

那么，根据你在峰会上的所闻，推动这种分歧的核心分歧点是什么？

So from what you heard at the summit, what are the core points of disagreement driving that divide?

确实存在一些共识，就在过去几天，美国政府和印度政府达成了几项高调协议。

There definitely are areas of agreement, and we've seen a couple of high profile agreements reached between the US government and the Indian government just in the last, several days.

因此，双方的重叠之处确实很多。

So there certainly is a lot of overlap.

比如，PACS硅协议对保障供应链和获取AI技术至关重要。

Point to the PACS silica agreement that's so important to secure supply chains, to secure access to AI technology.

我认为，对印度而言，重点正如你所说，是可解释性与开放获取。

I think the focus, for example, for India is, as you said, it is, you know, explainability, open access.

我特别印象深刻的是，莫迪总理强调要确保所有印度人都能使用AI工具，以改善他们的日常生活。

I was really struck by, prime minister Modi's focus on ensuring that all Indians have access to AI tools that can help them in their everyday life.

有一个非常具体的例子让我印象深刻：印度一个偏远村庄里，有人患有某种疾病，附近没有医生或护士，他们通过AI拍照识别病情，获得诊断和建议，从而确定下一步该怎么做。

You know, a really tangible example that really stuck with me is someone in a remote village in India who has a medical condition and there's no doctor or nurse nearby using AI to, you know, take take a photo of the condition, receive diagnosis, receive support, figure out what the next step should be.

这非常有力量。

That's very powerful.

因此，我认为开放访问和可解释性非常重要。

So I'd say open access explainability is very important.

如今，美国的超大规模云服务商正积极服务印度市场，满足印度政府的诸多目标。

Now the American hyperscalers are very much trying to serve the Indian market and serve the, the objectives really of the Indian government.

例如，他们推出了开放权重的模型版本，免费提供给印度的卫生机构和政府使用。

And so there are versions of their models that are open weights, that are being made freely available for health agencies in India as an example, to the Indian government as an example.

因此，他们确实在努力实现多个目标，但我认为关键在于开放访问和可解释性，这一点我确实看到了关注。

So there is an attempt to really serve a number of objectives, but I think this key is around open access, explainability, that I do see that, that that there's attention.

那我们再深入谈谈这一点，因为人们提出的一个担忧是，会被锁定在专有的大语言模型中。

So let's talk about that a little bit more because it seems one of the concerns raised is this idea of being captive within proprietary large language models.

这可能包括未来需要支付更多费用，或失去对公民数据的控制权。

And maybe that includes the risk of having to pay more over time or losing control of citizen data.

但与此同时，你提到这些国家希望采用人工智能确实有一些实实在在的好处。

But at the same time, you've described that there are some real benefits to AI that these countries want to adopt.

那么，在被专有模型束缚与选择开放免费模型之间，真正的张力或权衡是什么？

So what is effectively the tension between being captive to a model or the trade off instead for pursuing open and free models?

是两者之间存在显著的质量差异吗？这种权衡可以接受吗？

Is it that there's a major quality difference and is that trade off acceptable?

你看，这正是如此引人入胜的地方，迈克，我们需要思考的不仅是技术现在的状态，而是六个月、十二个月、二十四个月后它会变成什么样。

See, that's what's so fascinating, Mike, is, you know, what we need to be thinking about is not just where the technology is today, but where is it in six months, twelve months, twenty four months.

从我的角度来看，很明显，美国的专有模型将会强大得多。

And from my perspective, it's very clear that the proprietary American models are going to be much, much more capable.

让我们用一些数据来说明这一点。

So let's put some numbers around that.

美国五大公司为训练当前的大语言模型所投入的算力，是它们之前模型的约十倍，这意义重大。

The big five American firms have assembled about 10 times the compute to train their current LLM compared to their their prior LLMs, and that's a big deal.

如果规模定律成立，那么训练算力增加十倍，将使模型的能力大约提升一倍。

If the scaling laws hold, then a 10 x increase in training compute to result in models are about twice as capable.

现在，请花一分钟好好想想这一点。

Now just let that sink in for a minute.

从现在开始能力翻倍，这非同小可。

Twice as capable from here, that's a big deal.

因此，当我们考虑部署这些模型的好处时——无论是在生命科学还是其他众多领域——这些好处可能会变得非常巨大，而开源模型面临的挑战将是：它们能否在计算资源、训练机会、数据获取等方面跟上步伐？这是一个重大问题。

And so when we think about the benefit of deploying these models, whether it's in the life sciences or any number of other disciplines, those benefits could start to get very large and the challenge for the open models will be, will they be able to keep up in terms of access to compute, to training, access to data, to train those models, that is a big question.

当然，这两种模式都有其空间，印度政府完全有可能继续探索，找出哪种方式更能服务好他们的民众。我特别印象深刻的是，印度政府如此专注于服务所有公民，尤其是他们国家中最贫困的群体。

Now again, there's room for both approaches, and it's very possible for the Indian government to continue to experiment and really see which approach is gonna serve their citizens, the best, and I was really struck by just how focused the Indian government is on serving all of their citizens, most notably, you know, the the poorest of the poor in their in their nation.

所以，我们只能拭目以待，但纯粹的技术人员会认为，这些专有模型的能力提升速度将远超开源模型。

So I you know, we'll just have to see, but the the pure technologist would say that these proprietary models are gonna increase in capability much faster than the open source models.

所以，迈克，让我们从技术层面转向地缘政治层面，因为峰会上公布的美国战略远远超越了创新本身。

So, Mike, let's pivot from the technology layer to the geopolitical layer because The US strategy unveiled at the summit goes way beyond innovation.

是的。

Yeah.

这是个很好的观点。

It's a good point.

在这场关于其他国家是选择追求开源模型，还是更紧密地追随美国模式的讨论中，核心问题实际上是美国如何在全球范围内行使权力，以及它将如何构建未来的联盟。

And within this discussion of whether or not other countries will choose to pursue open models or more closely adhere to US based models is really a question about how The United States exercises power globally and how it creates alliances going forward.

显然，这一战略的一部分在于，美国假设如果其技术对其合作伙伴具有吸引力，这些国家就会希望与美国的全球总体目标保持一致，并愿意成为支持这些目标的伙伴，而这些目标自然与人工智能发展息息相关。

Clearly, some part of the strategy is that The US assumes that if it has technology that's alluring to its partners, that they'll want to align with The US's broad goals globally and that they'll wanna be partners in supporting those goals, which of course are tied to AI development.

因此，你之前提到的PACSILICA就是一个有趣的例子，这显然是美国战略的一部分，旨在与其他国家建立关系，使这些国家能够获得美国的模型和美国的人工智能技术。

So the PACSILICA, which you mentioned earlier, is an interesting point here because this is clearly part of The US strategy to develop relationships with other countries such that the other countries get access to US models and access to US AI in general.

而美国获得的回报则是对供应链、关键资源、劳动力的访问权——所有这些都是推进人工智能发展的必要条件，尤其是当美国正越来越努力地与中国脱钩，并切断中国可能为人工智能发展提供的资源时。

And what The US gets in return is access to supply chain, critical resources, labor, all the things that you need to further the AI build out, particularly as The US is trying to disassociate more and more from China and the resources that China might have been able to bring to bear in an AI build out.

所以，迈克，美国将所谓的‘真正的AI主权’定义为战略自主，而非完全自给自足。

So Mike, The US framed, you know, quote, real AI sovereignty, unquote, as strategic autonomy rather than full self sufficiency.

因此，美国实际上是在鼓励各国整合美国人工智能技术栈的组成部分。

So essentially, The US is encouraging nations to integrate components of the American AI stack.

那么，迈克，从宏观和政策的角度来看，你认为这一区别有多重要？

Now from your perspective, Mike, from a macro and policy standpoint, how significant is that distinction?

嗯，我认为这极其重要。

Well, I think it's extremely important.

显然，美国将它的AI战略视为不仅是经济战略，更是国家安全战略。

And clearly, The US views its AI strategy as not just economic strategy, but national security strategy.

过去八十年左右，美国曾利用其在军事和军备上的主导地位，构建了一个其他国家希望加入的安全保护伞，如今在AI领域也可能采取类似做法：如果某种主导技术能带来社会或经济利益，其他国家就会渴望获得它，而这将在你与其他国家就其他你重视的议题（如贸易政策、外交政策、对某国的制裁等）进行谈判时提供助力。

There are maybe some analogs to how The US has been able to, over the past eighty years or so, use its dominance in military and military equipment to create a security umbrella that other countries want to be under and do something similar with AI, which is if there is a dominant technology and others want access to it for the societal or economic benefits, then that is going to help when you're negotiating with those countries on other things that you value, whether it be trade policy, foreign policy, sanctions versus another country, that type of thing.

因此，在很多方面，美国似乎正在将AI视为其权力的支柱，就像军事力量在过去二战后很长一段时间内所扮演的角色一样。

So in a lot of ways, it seems like The US is talking about AI and developing AI as an anchor asset to its power in a way that military power has been that anchor asset for much of the post World War two period.

这正是如此有趣的地方，迈克，因为你之前曾向我强调，你认为AI有可能取代武器，成为美国全球权力的核心支柱，几乎就像一种科技版的防御保护伞。

See, that's what's so interesting, Mike, because you've highlighted before to me that you believe AI could replace weaponry as really the the anchor asset for US global power, almost a tech equivalent of a defense umbrella.

那么，这种战略有多持久？尤其是在一些国家对依赖性表示不安的情况下？

So how durable is that strategy, especially given that some countries are expressing unease about dependency?

是的。

Yeah.

这真的很难说。

It's really hard to know.

我认为，你我之前讨论过的那种紧张关系依然存在：各国是否愿意为了获得更先进的AI模型而做出取舍，放弃那些可能性能较差但更开放和自由的模型。

And I think the tension you and I talked about earlier, Steven, about whether countries will be willing to make the trade off for access to superior AI models versus open and free models that might be inferior.

这将告诉我们这种策略是否可行。

That'll tell us if this is a viable strategy or not.

而且看起来这种情况仍在发展中，因为如果我没说错的话，我们还没有收到印度或其他国家关于是否愿意做出这种权衡的明确信号。

And it appears like this is still playing out because correct me if I'm wrong, it's not like we've received some very clear signals from India or other countries about their willingness to make that trade off.

不。

No.

我认为这是对的。

I think that's right.

进一步谈谈权衡和AI部署标准的问题，美国明确拒绝了集中式的全球AI治理，转而支持与国内价值观一致的国家控制。

And just building on the concept of of the trade offs and sort of the, you know, a standard for AI deployment, you know, The US has explicitly rejected centralized global AI governance in favor of national control aligned with domestic values.

那么，这传递了关于全球技术标准如何演变的什么信号？尤其是在美国，国家标准与技术研究院（NIST）正在努力为代理式AI系统制定可互操作的标准。

So what does that signal about how global technology standards may evolve, particularly as in The US, the the National Institute of Standards and Technology, or NIST, works to develop interoperable standards for agentic AI AI systems.

是的。

Yeah.

史蒂文，我认为这很难说。

Steven, I think it's hard to know.

美国可能并不介意其他国家在使用基于美国的AI模型时拥有相当大的自由度，因为如果将来出现违背美国价值观的使用场景，美国可以通过法律手段事后调整这些模型的使用方式。

It might be that The US is okay with other countries having substantial degrees of freedom with how they use US based AI models because they could use US law to at a later date change how those models are being used if there's a use case that comes out of it that they find is against US values.

这在某种程度上类似于美元作为主导货币和全球主要支付系统，使美国能够根据自身利益实施制裁并限制其他类型的经济交易。

Similar in some way to how the US dollar being the predominant currency and therefore being the predominant payment system globally gives The US degrees of freedom to impose sanctions and and limit other types of economic transactions when it's in The US interest.

我不确定这是否确切成立，但这是一个值得思考的有趣问题，也可能是为何自由放任的策略最终仍符合美国利益的潜在动机。

So I don't know that to be specifically true, but it's an interesting question to consider and a potential motivation behind why a laissez faire approach might be ultimately still aligned with US interests.

所以，迈克尔，听起来AI正日益成为全球新的战略基础设施。

So, Michael, it sounds like really AI is becoming the new strategic infrastructure globally.

是的。

Yeah.

我觉得这确实是一个很好的视角。

I think that's actually a great way to think about it.

因此，史蒂文，如果情况如此，我们讨论的是AI可能塑造地缘政治竞争、全球范围内的经济差异，而这些至少在某种程度上与这些模型的进一步发展和算力提升相关。

And so, Steven, if that were the case and we're talking about the potential for this to shape geopolitical competition, potentially economic differentials across the globe, and if that is correlated at least to some degree with the further development and and computing power of these models.

那么，你认为投资者应该关注哪些信号？

What do you think investors should be looking at for signals from here?

对我来说，最重要的是模型进步的速度，不仅仅是美国的模型，还有中国和开源模型。

Number one, by a mile for me is really the pace of model progress, not just American models, but Chinese models, open source models.

对于美国来说，重大突破的时间点预计在四月到六月之间，涉及五大大型语言模型厂商。

And there, the the big reveal for The United States should be somewhere between April and June for the big five LLM players.

这是一些推测，基于对它们芯片采购、电力获取等情况的追踪，但这个时间框架似乎确实如此，几位高管也提到过这个大致的时间范围。

That's a bit of speculation based on tracking their, chip purchases, their power access, etcetera, but that appears to be the, the time frame, and a couple of execs have spoken to that that approximate time frame.

我想提醒投资者，我认为这些模型的实际能力可能会让我们感到惊讶。

I would, caution investors that, I think we're gonna be surprised in terms of just how powerful those models are.

我们已经在2026年初看到，那些并未使用大量算力训练的模型，其表现远远超出了预期，有些情况甚至极为显著。

And we're already seeing in early, 2026, these models that that were not trained on that kind of volume of compute have really exceeded expectations, you know, quite dramatically in some cases.

我举一个例子。

And I'll give you one example.

METR 是一家第三方机构，用于追踪这些模型能够完成的任务的复杂程度。

METR is a third party that tracks the complexity of what these models can do.

METR 一直指出，这些模型能够完成的任务复杂度大约每七个月翻一番。

And METR has been highlighting that every seven months, the complexity of what these models are able to do approximately doubles.

速度非常快。

It's very fast.

但真正引起我注意的是，大约一周前，其中一个大语言模型大幅突破了这一趋势。

But what really got my attention was about a week ago, one of the LLMs broke that trend in a big way to the upside.

如果根据METR所预期的扩展定律，模型应该能够独立运行大约八小时，略多于八小时。

So if the, scaling laws would hold based on what MHR would have expected, they would expect the model to be able to act independently for about eight hours, a little over eight hours.

而我们看到的是，最近推出的最佳美国模型能达到约十五小时。

And what we saw was, the best, American model that was recently introduced was more like 15.

这意义重大。

That's a big deal.

因此，我认为我们正看到非线性进步的迹象。

And so I think we're seeing signs of nonlinear improvement.

我们还将看到这些AI高管就模型的递归自我改进发表更多声明。

We're also going to see additional statements from these AI execs around recursive self improvement of the models.

一位前AI高管曾谈到过这一点。

One ex AI executive spoke to that.

另一位大语言模型高管最近也谈到了这一点。

Another LLM exec spoke to that recently as well.

所以我们开始看到加速的迹象。

So we're starting to see an acceleration.

这意味着我们需要认真权衡开源模型和专有模型之间的利弊。

That means we then need to really consider the trade offs between the open models and the proprietary.

这将变得至关重要，并且应该会在春夏季期间发生。

That's gonna become really critical, and that should happen really through the spring and summer.

明白了。

Got it.

好了，史蒂文，感谢你抽出时间交谈。

Well, Steven, thanks for taking the time to talk.

和你交谈很愉快，迈克。

Great speaking with you, Mike.

也感谢大家的收听。

And thanks for listening.

如果你喜欢关于市场的观点，请在你收听的平台给我们留下评价，并今天将这个播客分享给朋友或同事。

If you enjoy thoughts on the market, please leave us a review wherever you listen, and share the podcast with a friend or colleague today.

前面的内容仅供参考，基于创作时可获得的信息。

The preceding content is informational only and based on information available when created.

这不构成任何要约或招揽，也不是税务或法律建议。

It is not an offer or solicitation, nor is it tax or legal advice.

它未考虑您的财务状况，可能不适合您。

It does not consider your financial circumstances and may not be suitable for you.