戴伦·帕特尔：英伟达的新护城河及中国为何“被半导体药物强化”

AMD试图在玩英伟达的游戏，但AMD在硅片工程方面极其擅长。

You have to AMD is trying to play NVIDIA's game, but AMD is like extremely good at engineering silicon.

对吧？

Right?

其他所有人都必须尝试一些奇特或不同的东西。

Everyone else has to has to has to try something weird or different.

对吧？

Right?

所以当你观察Etched、Maddox、Positron、Cerebris或Tensor这些公司时，你会审视所有这些企业。

And so when you look at Etched or Maddox or Positron or Cerebris or Tensor, you go to look at all these companies.

它们所做的事情各有独特之处，但尚不清楚当产品面世时，AI模型是否仍会处于那个范畴内。

There are unique things about what they're doing and it's not clear if AI models will still be within that realm when that comes out.

对吧？

Right?

现在人们是否开始使用像engrams和其他稀疏注意力技术呢？

Does oh, now people use like engrams and other sparse attention techniques.

这会不会改变人们所专注的一些方向呢？

Is that like is does that change like some of the specializations people are doing?

或者，现在人们做的模型已经是稀疏的MOE，而不是密集型模型了。

Or hey, people are now doing like, you know, models are now sparse MOEs instead of being dense models.

这会改变些什么吗？

Does that change things?

在模型层面有这么多优化和变化，至少你很难预测机器学习研究会走向何方。

There's so many optimizations and changes on the model side and you can't predict what's gonna happen with the ML research easily at least.

你确实无法预测。

You can't.

你今天所优化的目标，必须是对两年后AI发展态势的展望。

The thing you're optimizing for today has to be a vision of where AI will be in two years.

而NVIDIA已经完全接受了这一点。

And NVIDIA's fully accepted.

他们也不知道最终会走向哪里。

They don't know where that's gonna be.

这就是为什么他们现在拥有一个芯片组合，而不仅仅是一条GPU产品线。

That's why they have a portfolio of chips now, not just one GPU line.

对吧？

Right?

不仅仅是Hopper或Blackwell Rubin。

It's not just Hopper or Blackwell Rubin.

现在的情况是，你知道，不再是NPR Hop那条线了。

Now it's gonna be, know, it's not NPR Hop, you know, you know, it's not that line.

这就像有各种各样的芯片来服务不同的市场和不同的可能场景。

It's like there's a variety of chips to serve the different markets and different possible scenarios.

他们认为今天每种芯片都有自己的愿景，但哦，结果可能通用型芯片很糟糕，而实际上AI模型的发展方向使得CPX或Grok风格的芯片才是最好的。

They think each of them has this vision today, but oh, it might turn out the general purpose one sucks and and actually AI models have developed in a way where CPX or Grok style chips are the best.

对吧？

Right?

嗯，好吧。

Well, okay.

现在我们为那个市场提供了解决方案。

Now we have a solution for that market.

所以我认为这是初创公司面临的挑战。

And so I think that's the challenge with the startups.

话虽如此，我认为他们都在进行非常有趣的押注。

With that said, think they're all taking very interesting bets.

我认为这比第一波AI硬件押注要令人兴奋得多，比如Graphcore、Rebris、SamaNova、Grok，它们都押注于内存并将内存集成在芯片上。

I think it's I think it's much more exciting than the first wave of AI hardware bets, Graphcore, Rebris, SamaNova, Grok, where they all made the same bet on memory and putting the memory on the chip.

他们基本上就是下了一个赌注，针对某一类模型（所有相似类型的模型）进行了优化，但很长一段时间里都没能成功。

They sort of just made a bet and they optimized for a certain kind of model, all similar kinds of model, and it didn't end up working out for a long time.

对吧？

Right?

他们不得不转向，并且

They had to pivot and

他们不得不着手处理许多耗时的事情。

they had to work on a lot of things that took

很长时间。

a long time.

我认为这些公司对他们认为模型会是什么样子有非常清晰的愿景。

I think these companies have like a really clear vision of what they think models will look like.

对吧？

Right?

没错。

Right.

Etch、Maddox、Positron 都是如此。

Etch does, Maddox does, Positron does.

这正是这三家新兴公司之间真正酷的地方。

And that's what's really cool about it between the three of them, these new age.

所以，我是真的很期待它们。

So I mean, I'm I'm excited for them.

我非常非常怀疑。

I'm very, very skeptical.

我不知道风险投资人认为这些公司成功的可能性有多大，但我认为它们都低于1%。

I don't know what a what a venture capitalist views as like likely chance of succeeding, but I think all of them are less than 1%.

对吧？

Right?

是的。

Yeah.

但你知道，那只是那只是

But, you know, that's that's that's a

但它们成功的那个世界，是不是一个多元化的硅世界，即每个客户都会使用多种不同的GPU？

But the world where they win is a a multi silicon kind of world where any given customer uses a range of different GPUs?

也可能是每个客户只关注某一项特定的工作负载。

It could it could or it could be any given customer has like one workload they care a lot about.

Anthropic 明显对视频生成和图像生成毫不在意。

Anthropic clearly does not give a crap about video gen image gen.

对吧？

Right?

他们根本不在乎。

They just don't care.

另一方面，像Midjourney这样的公司就非常重视图像和视频生成。

On the flip side, company like Midjourney cares a lot about image and video gen.

对吧？

Right?

图像和视频生成，就像我提到的，它非常……嗯，它对内存带宽的要求并不高。

Image and video gen is very very like like I mentioned, like it's a very like it's not very memory bandwidth heavy.

它非常非常非常热爱计算。

It loves loves loves compute.

对吧？

Right?

而像链接大型语言模型的推理，比如，你知道，这些，举例来说，编码代理就非常关心长时间流的解码，这非常消耗内存带宽。

Whereas inference of link large language models in the style of like, you know, this these, you know, say for example, coding agents cares a lot about decoding for long streams of time and that's very memory bandwidth heavy.

对吧？

Right?

所以这就像一个简单的例子，但其中还有更多细微差别，比如矩阵乘法单元的大小、张量核心（也就是你使用的脉动阵列）、网络与内存的比例、内存层次结构是怎样的，以及针对不同类型的注意力机制你做了什么处理等等。所有这些方面都存在大量的专业化设计。

And so there's like that's like a simple example, but there's a lot more nuance there in terms of like even like the size of like the mat matrix multiply units, the tensor cores that you you know, the systolic arrays that you use or the ratios of networking and memory memory and like what's the memory hierarchy look like and you know, what are you doing for different kinds of attention and like, oh, like all these sorts of things like there's a lot of specialization here.

因此，有些人正大力押注于不同类型的专业化。

And so some people are betting big on on different types of specialization.

而且我认为，你可以清楚地看到一个世界，在这个世界里，不同公司确实关心不同的东西。

And I I think like you could clearly see a world where companies do care about different stuff.

对吧？

Right?

比如说，如果今天存在一款专为视频和图像生成优化的芯片，并且它比英伟达的芯片更好，或者就是英伟达自己制造的，我认为Midjourney绝对会只用它来进行推理。

Like like if for example a chip optimized for video and image generation existed today and it was better than Nvidia or Nvidia made it, I think mid journey would absolutely only use that for inference.

我认为在训练方面，他们仍然会使用通用芯片，就像Meta和谷歌那样——他们应该这样做。

I think for training they'd still use the general purpose thing and as would like meta and Google would like they should do that.

对吧？

Right?

而且，Meta实际上有两条AI芯片产品线。

And hey, meta actually has two lines of AI chips.

他们的MTIA，有一条产品线专注于推荐系统，还有一条产品线专注于生成式AI。

Their MTIA, there's a line that's focused on recommendation systems, and then there's a line that's focused on Gen AI.

生成式AI那条是新产品线，但推荐系统芯片那条线仍在继续发展。

The Gen AI one is a new line, but that recommendation systems chip line is still continuing.

对吧？

Right?

这听起来不够酷。

It's not sexy.

没人关心这个，因为字节跳动也有一条推荐系统芯片产品线，它并不专注于生成式人工智能，这没关系，毕竟这是一个价值约2000亿美元的业务，主要就是决定给我展示什么广告。

No one cares because there's no and and ByteDance also has a recommendation system line of chips, and it's not really focused on Jet AI, which is fine because, you know, this is a $200,000,000,000 business or something which is just deciding what ad to serve me.

对吧？

Right?

以及如何排列我朋友的故事顺序，诸如此类的事情。

And what order to put my friends stories and, you know, things like this.

所以我认为，只要目标市场足够大，出现专门的AI芯片是完全合理的。

So so I think like it's perfectly fine for there to be specialized AI chips given the target market is big enough.

你需要有远见才能知道目标市场是什么，除非你是超大规模云服务商，那样的话你可以先用通用芯片，直到市场明确存在，然后再制造你的专用集成电路。

And you have to have vision to know what that target market is unless you're hyperscaler, then you can, like, just, like you can just use general purpose until you've, like it's clearly there, and then you can make your ASIC.

对吧？

Right?

真有意思。

Fascinating.

再谈谈这一切的地缘政治层面，这个话题总是很有趣。

Turning to the geopolitical aspect of of all of this, which is always fun.

华为和英伟达在中国的情况。

Huawei and NVIDIA in in China.

去年，这部分收入约占他们总收入的10%或12%。

Last year, there was, like, 10 or 12% of their overall revenue.

而今年，他们表示其市场份额已基本降至非常低的水平。

And this year, they they were saying that their market share has basically dropped to not very much.

这是指华为的芯片吗？

Is that Huawei chips?

是限制措施吗？

Is that restrictions?

是关税问题吗？

Is that tariffs?

那边到底发生了什么？

What what's happening over

是多种因素造成的。

variety of things.

实际上，在去年的一些季度里，我记得甚至超过了20%。

Actually, in in some in some quarters last year, it was even north of 20, I think.

但我不太记得具体数字了。

But I I don't remember exactly.

不过无论如何，你知道，如果看2022年的数据，中国在购买服务器硬件方面的规模几乎与美国相当。

But anyways, you know, if you look at 2022, China was almost the size of The US in terms of buying server hardware.

对吧？

Right?

差不多。

Almost.

还没完全达到，但正在接近。

Not quite, but getting there.

看起来他们在一两年后就会和美国达到同等规模。

And it looked like they were gonna be the same size as American like a year or two after that.

对吧？

Right?

如果你看看全球数据中心容量、全球云容量等等等等，美国公司和中国公司，对吧，它们主导了世界。

And if you look at like global data center capacity, global cloud capacity, etcetera etcetera etcetera, American companies and Chinese companies, right, that dominate the world.

美国公司在这方面显然做得更好，但这两者都主导了世界。

American companies obviously doing a lot better here, but both of those dominate the world.

而且如果你看看每个行业，对吧，很明显中国想要实现自给自足。

And if you look at like every industry, right, you know, it's it's it's very clear that like China wants to insource stuff.

对吧？

Right?

所以2015年，他们制定了2020年和2025年的五年计划，设定了本国生产的半导体比例目标。

So in 2015, they made these five year plans for '2 2020 and 2025 where they set the percentage of semiconductors they wanted domestically produced.

但两次都没能达到目标，这没关系。

And they've missed the goal both times, which is fine.

对吧？

Right?

他们设定了非常激进的目标，即使没达到，至少也能取得不错的结果。

They set really aggressive goals and even, know, shoot for the moon even if you miss hit the stars.

对吧？

Right?

事情就是这样发展的。

And that's sort of what's happened.

对吧？

Right?

你看，中国在先进半导体领域还没赶上，但中国生产的微控制器几乎和德州仪器、意法半导体等的产品一样好，而且更便宜。

Like, look, China is not caught up on, you know, leading edge semiconductors, but microcontrollers from China are almost as good as the microcontrollers are as good and cheaper than the ones from Texas Instruments or ST Micro or, you know, etcetera.

对吧？

Right?

或者像这种电源随机芯片，比另一家公司的更好或一样好。

Or like this power random power chip is better than or the same as the one from, like, another company.

对吧？

Right?

所以他们确实建立起了半导体产业，并开始更多地采用内部供应。

And so they've really built up a semiconductor industry and started insourcing a lot more.

我不明白为什么中国不会购买全球40%的AI芯片。

I don't see why China wouldn't be buying, you know, 40% of the world's AI chips.

而美国，大概五六十个百分点，然后世界其他地区，嗯，我说的美国是指美国本土公司。

And The US, like, fifty, sixty percent, and then the rest of the world, like, you know, not and when I say US, I mean US origin companies.

这似乎才是世界更自然的状态，但存在各种限制，而且，嘿，这可能是人类历史上最大的变革——知识工作以及随之而来的一切，最终还有机器人技术等等。显然这里涉及很多地缘政治因素，所以存在限制。

That seems like a more natural state for the world, but there are restrictions and and, hey, this is the biggest change in human history maybe ever, knowledge work and you know everything that's gonna happen there and and then eventually like robotics and all these things like, you know, obviously there's there's a lot of geopolitical stuff and so there are restrictions.

英伟达在向中国销售其顶级芯片方面受到了限制，这显然对销量产生了很大影响，因为为什么要这么做呢？

NVIDIA has been handicapped handicapped from selling their best chips to China and so that's obviously impacted the sales a lot because like why would you do that?

所以，当你看看全球谁租用GPU最多时，有三家公司。

And so when you look at who rents the most GPUs in the world, it's three companies.

对吧？

Right?

其中一家显然是OpenAI。

So one of them is obviously OpenAI.

第二家，实际上他们比OpenAI规模更大。

Second one, actually they were bigger than OpenAI.

它们现在比OpenAI规模更大，或者说，不，它们之前比OpenAI大，但最近OpenAI超越了它们，我说的是字节跳动。

They are bigger than OpenAI today, or no, they were bigger than OpenAI than OpenAI eclipsed them recently is ByteDance.

字节跳动从甲骨文、谷歌以及许多其他云公司租用了大量芯片，因为他们在国内无法获得所需的芯片。

ByteDance runs rents tons of chips from Oracle and Google and and, you know, many other cloud companies because they couldn't get the chips they needed in China.

他们主要只是为TikTok提供服务。

They're mostly just serving TikTok.

对吧？

Right?

好的。

Okay.

他们不被允许购买，这很糟糕，但他们被允许租用。

Well, they're not allowed to buy them and that sucks, but, know, they're they're allowed to rent them.

所以，如果我无法获得最好的芯片，我就去外部租用。

And so, okay, if I'm not allowed to get the best ones, I'm gonna rent externally.

如果字节跳动是全球第二大GPU租用者，那它替代了本应在中國内部产生的需求。

And if ByteDance is the second biggest renter of GPUs in the world, that's substituting demand that would have been built in China in many cases.

这些需求转而被部署在马来西亚，甲骨文在马来西亚拥有超过一吉瓦的算力，将被字节跳动使用。

It's instead being built in Malaysia, and Oracle has over a gigawatt of capacity in Malaysia that ByteDance is gonna take.

对吧？

Right?

像这样的情况，涉及的是数十万而非数百万颗芯片，是本应投入中国的数百亿美元算力，但现在却没有。

So things like this are, you know, you know, hundreds of thousands, not millions of chips, tens of billions of dollars of cap capacity that would go to China, but it's not.

比如，这些算力现在转而流向了马来西亚。

It's going to Malaysia instead, as an example.

关于这一点，另一个类似的观点是，中国的情况是这样的，你知道，他们有这些五年计划。

Another sort of point around this is China's like, you know, they've had these five year plans.

所以，而且，你知道，中国这些倡议的运作方式是，虽然有自上而下的指令，但之后他们基本上就放手让大家去干。

So and and, you know, the way these initiatives work from China is there is like some top down ordering, but then they just kind of whip the hole.

就像，每个人都会参与进来，这真的很酷。

Like, everyone just kind of gets into it and it's really cool.

就像，我认为它并不像很多人想的那样完全是自上而下的。

Like, I don't think it's as top down as many people think.

就像，我觉得整个国家都像是被半导体给迷住了。

Like, I think the entire country is like semiconductor pilled.

嗯哼。

Mhmm.

对吧？

Right?

有些电视剧里，人们在晶圆厂里谈恋爱，或者剧情是人们坠入爱河，而他们是光伏——也就是太阳能电池的研究人员和工程师。

There are dramas where people fall in love in the fab or dramas where people fall in love and they're photovoltaic, like, solar cell researchers and engineers.

这就像背景一样，但实际上，你的另一半是半导体工程师或光伏太阳能电池研究员，这简直太酷了。

And it's like it's like this is just the backdrop, and it's like, actually, this is it's like super cool for your, like, significant other to be that semiconductor engineer or to be that photovoltaic, you know, solar panel researcher.

而不是一个网红。

As opposed to an influencer.

而不是一个网红。

As opposed to an influencer.

对吧？

Right?

抱歉。

Like, I'm sorry.

《爱岛》我看了大概十分钟，因为被迫看的。

Love Island is I I I watched, like, for ten minutes because I was forced to.

我觉得这简直太糟糕了。

I was like, this is freaking terrible.

但你知道的，像

But, you know, like

我们真是没救了。

We are so cooked.

不。

No.

说真的，我们没救了。

You know, seriously, we're cooked.

我们没救了。

We're cooked.

所以我觉得，当你想到这种情况时，它甚至已经渗透到戏剧中了。

And so I think I think, like, when you think about, like, this happens, it's like it's diffused into drama even.

人们喜欢，有很多关于半导体行业的戏剧，它们涵盖了浪漫、喜剧等各种类型。

People like like, there's multiple dramas, like, taking place about semiconductor industry, and and they're like romance, comedy, like, the the entire spectrum.

对吧？

Right?

戏剧或者说，这到底是怎么回事？

Drama or really, it's like it's like, what the heck is going on?

总之，你有这么多省份，有这么多地方城市，都在制定规章、发放补贴，各种各样的措施。

Anyways, you have all these provinces, you have all these local cities, setting out ordinances and giving out subsidies and all sorts of stuff.

对吧？

Right?

真的太疯狂了。

It's truly like crazy.

比如有些国家级的措施，比如这个不征税。

Like there's some national level stuff like, oh, no taxes on this.

哦，我们要禁止几样东西。

Oh, we're gonna ban a few things.

但据我所知，国家政府并没有禁止英伟达的H20或H200。

But as far as I understand, the national government has not banned NVIDIA's h 20 or h 200.

但地方上已经禁止了。

But the local ones have.

对吧？

Right?

很多地方都表示不行，你们必须使用中国制造的芯片。

A lot of local ones have said no, you know, you must use China manufactured chips.

这话说得，谁告诉你的，你在这儿是来维护这个的？

It's like, who told you that, you know, you're here to uphold this?

这根本无所谓。

It's like, doesn't matter.

对吧？

Right?

我的意思是，这其实挺有意思的，因为这就成了物竞天择。

I mean, like, it's it's it's cool because then you have this like survival of the fittest.

所有这些省份和城市都在试图通过各种补贴、拨款、产业园区等方式吸引不同的公司。

All these all these provinces and cities are trying to attract different companies with different types of subsidies and grants and industrial parks and, like, all these different things.

然后，那些成功的地区真的会发展出自己的产业，并最终占据主导。

And then, like, the ones who succeed actually develop an industry and and they take over.

这确实就是人们看待中国的方式。

That's really how one thinks of of China.

对吗？

Right?

这听起来更像是美国的情况，或者说，就像一个联邦政府和各州的关系，各省份拥有

It almost sounds like more like The US or, like, it was a federal government and states where the provinces have

是的。

Yeah.

自主采购的权力。

Authority over their purchasing.

我的意思是，这实际上真的很棒。

I mean, it's it's actually, like, great.

有一个抖音，不对，不是抖音。

There's this one TikTok or not TikTok.

是抖音和Instagram上的一个人，他们在唱歌。

TikTok and Instagram, like, person, and they're like they they like singing.

他们唱道，如果你想在中国买东西，一定要去对地方。

They're like, if you wanna if you wanna buy things in China, make sure you go to the right place.

然后他们就开始说些最不着边际的话，还提到城市名字。

And then they just say the most random shit and name the city.

然后你去查一下，就会惊叹，哇，原来这个城市拥有整个产业链。

And then you look into it and you're like, wow, this city has the entire supply chain for this.

比如说是灯罩，然后它就会说出那个城市的名字。

And it's like lampshades, and then it names the city.

这简直让人摸不着头脑，什么鬼？

It's like, what the fuck?

有个城市专门做灯罩。

There's a city that specializes in lampshades.

是的。

Yeah.

对。

Yeah.

比如，就像是麦克风支架那种感觉。

Like, it's like and it's like microphone arms.

像麦克风一样。

Like microphones.

literally，中国有一个城市专门生产吉他

Like, it is like, literally, there's a city in China that specializes in Guitars

也是。

as well.

对吧？

Right?

这个城市成为了吉他之都

This one one city that became the guitar capital of

全世界。

the world.

完全是这样。

Literally everything.

是的。

Yeah.

真的是应有尽有。

Literally everything.

有这么一座城市。

There's a city.

而且它不像那种，比如说，专门生产相机支架臂的城市。

And it's not like, hey, specifically for camera arms, for example.

这里面有滚珠轴承，而这些滚珠轴承呢，有好几家制造商专门为相机支架臂生产滚珠轴承。

There's ball bearings in this, and the ball bearings are like, there's ball bearings there's multiple manufacturers of ball bearings for camera arms.

是的。

Yeah.

然后，世界上大部分的相机臂都来自那个城市。

And then, like, most of the camera arms in the world come from that one city.

这简直让人摸不着头脑，到底是怎么回事？

And it's like, what the hell is going on?

所以，我觉得人们没有意识到半导体行业是极其专业化的。

And and so, like, the semiconductor industry, I think people don't realize is absurdly specialized.

我没有在回答你的问题。

I'm not answering your question.

我只是想发泄一下，因为我觉得人们不了解中国的半导体产业。

I'm just gonna let it out a rant because I think people don't understand China semiconductors.

这真的很糟糕，或者说半导体行业整体都是如此。

It's really sick or semiconductors in general.

这确实很吸引人。

Like That's fascinating.

你知道，在日本，他们专注于几种不同类型的化学品，并且在这方面做得最好。

You know, like, in Japan, they like focus on a few different types of chemicals, and they're the best at it.

这几乎成了一种文化现象。

And it's, like, almost a cultural thing.

对吧？

Right?

就像日本人非常精确，比如做寿司，这完全关乎手艺和技艺。而且，你知道，日本的法国菜比法国本土的还要好，因为日本厨师去那里学习，回来后在日本将其完善，因为他们做事如此精确。

Like, Japanese people are so precise, like, with sushi, and, like, it's all about the trade and the craft, and, like, you know, the French food in Japan is better than the French food in France because the the Japanese chefs went there and then come back and they perfected it in Japan, and, like, because they're so precise.

而且，日本在很多不同的事情上都做得非常出色，因为他们非常精确，并且对技艺有着极高的专注。

And and there's so many different, like, things that, like, Japan is so good at because they're so precise and, like, dedicated to the craft.

这大概源于，嗯，我不知道，可能是武士道文化之类的吧。

And it comes out of, like, I don't know, like, samurai culture or something.

我不太确定。

I don't know.

对吧？

Right?

比如，我不太清楚这种文化是怎么形成的。

Like, I don't know exactly know how that culture came up.

所以当你观察时，你会发现世界各地都有类似的情况发生。

And so when you look at, like, and it's, like, across the world, there's different places where things like this happen.

对吧？

Right?

比如，荷兰制造EUV光刻机。

Like, oh, like, The Netherlands makes EUV tools.

酷。

Cool.

我想是的。

I guess so.

哦，你看整个半导体行业，有一篇著名的经济学文章叫《iPencil》之类的，讲的是铅笔——简单的铅笔是怎么来的，比如橡皮来自印度尼西亚，石墨来自这里的某个矿场，木材来自加拿大的这些白杨树，实际上没有整合整个供应链你就造不出一支铅笔。

Oh, and you look across the semiconductor industry, there's a famous economic essay called iPencil or something like that or talking about how the pencil, like, simple pencil comes from like, oh, the rubber comes from like Indonesia for the eraser, and the graphite comes from this mine here, and and the wood comes from these Aspen trees in Canada, and like you actually can't make a pencil without aggregating this entire supply chain.

半导体行业则更加疯狂，因为我觉得大概有15到20个国家能让整个半导体行业停摆。

Semiconductor industry is like way crazier because like, I would say there's like 15 or 20 countries that could shut down the entire semiconductor industry.

对吧？

Right?

甚至像奥地利也能做到。

Even like Austria could.

对吧？

Right?

然后，这就像是，什么？

And and it's like, what?

就像是，嗯，有两家公司在某些冷门领域各自占据了90%的市场份额。

It's like, well, yeah, there's two different companies there who have like 90% share in like some random niche stuff.

然后呢，好吧。

And it's like, okay.

不错。

Cool.

我想奥地利确实可以，哦，对了。

I guess Austria can and, oh, yeah.

这两家公司年收入加起来还不到十亿美元，但它们掌握着关键环节，而这种关键环节到处都是，因为整个流程太复杂了。

Those two companies only, like, have less than 1,000,000,000 of revenue, but they just happen to have linchpin critical things, and there's linchpin critical things everywhere because the process is so complicated.

所以中国一直在努力复制这种模式。

And so China's been trying to replicate this.

它们目前还缺哪一样东西没搞定吗？

Is there one thing they're missing that they don't have yet?

我觉得缺的东西多了去了。

I think there's a lot of things.

我认为如果你

I think if you were

闭上眼睛说，或者如果切断与所有国家的联系，宣布不再有全球化，中国目前在半导体领域拥有最垂直的产业链，他们是世界上半导体技术最先进的，因为他们的晶圆厂在很多方面还能勉强运转，因为他们已经建立了一些这样的化学供应链。

to close your eyes and say or if were to cut off every country and say there's no more globalism, China has the most vertical stack in semiconductors today and they're the best at semiconductors in the world because their fabs could still run somewhat on a lot of things because they have built some of these chemical supply chains.

对吧？

Right?

比如台积电在某些化学品上100%依赖日本供应。

Like TSMC for certain kinds of chemicals 100% share from Japan.

对吧？

Right?

英特尔也是一样的情况。

Or Intel same thing.

对吧？

Right?

或者你知道，某些特定工具可能完全由荷兰公司垄断，或者完全由美国公司、奥地利公司等等掌控。

Or you know, for certain kinds of tools a 100% share from Netherlands or a 100% share from, you know, this American company or that, you know, Austrian company or this or that.

对吧？

Right?

就像，有这么多不同的公司，你知道的，这家瑞士公司。

Like, there's just all these like, you know, this Swiss company.

到处都是这些地方拥有百分之百的份额。

Like, it's just all these different places have 100% share.

可能是一家公司，也可能是三家，但它们在地理上或同一区域。

It might be one company, it might be three companies, but geographically or in the same area.

而中国已经建立了这种能力。

And China's built that up.

对吧？

Right?

因为他们推行了‘中国制造’计划，投入了大量资金，并且形成了这样一种文化：这些省份的人会说，好吧，我决定要专注于这个领域，甚至可能都不是某个城市。

Because they've created this made in China initiatives, which just plowed money into it and they've got this culture of like the diffused like, you know, these provinces are like, yeah, I just decided I'm gonna fucking focus on or it might not even be it might not even be the city.

对吧？

Right?

可能是这样，你知道，有人把它带到那里并决定这么做，然后人们就会说，哦，哇，你也在做这个？

It may be the like, you know, someone brought it there and decided and then people are like, oh, wow, you're doing that?

我也是。

Me too.

就像，我是一个帕特尔，我在汽车旅馆长大，你猜怎么着？

Like, I'm a patel and I grew up in a motel and guess what?

我们，几乎我认识的所有帕特尔人都在汽车旅馆长大，这是因为某个随机的帕特尔移民到美国，在酒店或汽车旅馆工作，然后买下了一家汽车旅馆，然后，这种情况就开始发生了。

We like, almost all the patel's I know grew up in a motel and it's because some random patel immigrated to America and like worked at a hotel motel and then bought a motel and then like, it just started happening.

对吧？

Right?

就像，这些事情在某种程度上是偶然发生的，嗯，我也不知道怎么说。

Like, you sort of like these things are serendipitous in those sorts and like, I don't know.

就像，我认为这是同一种专业化。

Like, and it's like, I I view it as the same kind of specialization.

对吧？

Right?

中国的城市好像也开始做这些事了。

Chinese cities are like starting to do these these things.

中国还缺少很多东西。

China's missing a lot of things.

对吧？

Right?

我想说，如果把科技水平倒退十年，中国是完备的，而其他国家都不完备。

I would say like if you say minus ten years tech, China's complete and no one else is complete.

对吧？

Right?

台湾是不完整的。

Taiwan is not complete.

他们的晶圆厂没有国外供应就会停工，你知道的，然后你要么缩减规模，要么在整个产业链上发展壮大。

Their the fabs are shut down without foreign supply, you know, and you go down or you grow across the stack.

但如果你去看十年技术，或者更可能是二十年技术，中国可以建立完全垂直的供应链，我认为没有其他国家能做到这一点。

But if you go to ten year tech, maybe maybe more like twenty year tech, you could get a fully vertical supply chain in China, which I do not think any country could do.

比如美国没有来自其他地方的设备，根本无法建成一个完全垂直的晶圆厂。

Like America could not build a fully vertical fab without stuff from elsewhere.

是的。

Mhmm.

即使是二十年前的技术也是如此。

Even if it's twenty year old tech.

对。

Yep.

对。

Yep.

恐怕即使是四十年前的技术也做不到。

Probably not even forty year old tech.

所以这很有趣。

And so so that's interesting.

但另一方面，你确实需要专业化。

But then when the flip side is like, well, like you kind of do need specialization.

这就是那种化学品能达到最纯净、最优质、最精密工程化的原因，或者说那种化学浆料、那种气体或那种工具，因为那个国家的每个聪明人或者很多人都在那种文化环境中成长，整个供应链都在那里，大家都有所了解，就像开车就能到达一样，这大概就是供应链运作的方式。

That's how that chemical gets the purest best, you know, most engineered, know, or that that slurry of chemicals or that, you know, that gas or like that tool because every smart person or a lot of them in that country grew up around that culture and like every the supply chain is there and like everyone kinda knows and like it's like a a drive away and like sort of like this is what makes supply chains work.

是的。

Yeah.

关键在于存在这种专业化，而只有当你拥有这种高度专业化时，才能产生最顶尖的东西。

Is that there is this specialization and the best of the best only comes when you have that hyper specialization.

所以中国没有光刻技术。

So China doesn't have lithography.

他们的光刻技术落后了大约十年，而且我认为再过几年，差距会缩小到五年。

Their lithography is like ten years behind, and I think it'll be five years behind in a couple of years.

对吧？

Right?

他们正在快速追赶。

They're catching up fast.

我认为在很长一段时间内，他们都无法达到ASML的水平。

I don't think they'll be as good as ASML for a long time.

你知道，也许吧。

You know, maybe.

我不确定。

I don't know.

也许他们会，你知道，中国，你永远不应该低估中国，而且中国的工程师们，你知道，但至少在一段时间内。

Maybe they will be, you know you know, China you shouldn't ever underestimate China, but like and Chinese engineers are, you know, but like for a while.

对吧？

Right?

或者说，你知道，我认为他们无法像许多中国公司或美国公司那样制造领先的化学品及其工具，你只需审视整个供应链。

Or like, you know, I don't think they'll be able to make leading edge chemicals like many Chinese companies or many American companies and their tools and like you just go across the supply chain.

他们并非，嘿，在制造供应链的任何方面都处于前沿地位。

They're not, hey, forefront on really anything in the manufacturing supply chain.

在设计供应链方面，他们开始在某些领域达到相似水平，但价格更便宜，或者落后一两年但更便宜，这对很多产品来说是可以接受的。

On the design supply chain, there's some things that they're starting to be similar par but like cheaper or like a year or two behind but cheaper and that's like fine for a lot of stuff.

华为就是一个例子。

An example of that is Huawei.

对吧？

Right?

华为在手机领域曾与苹果完全不相上下。

Huawei in mobile phones was on par with Apple like entirely.

是的。

Yeah.

而且当他们设计出最棒的产品时，他们已成为苹果和台积电的最大客户，他们在电信领域排名第一，他们的技术确实更胜一筹。

And they had become Apple TSMC's biggest customer when they were designing the best thing and they are number one in telecom and their tech is just literally better.

那么，当你思考会发生什么时，你知道，中国是否缺少任何东西呢？

And so when you think what happens, you know, is is is China missing anything?

看起来他们……他们并没有……他们并没有在当今AI供应链的许多方面占据最优地位。

Looks like they're they don't they don't they don't have the best of much that, you know, today in the AI supply chain.

他们拥有完整的解决方案，虽然落后几年，但他们会想办法降低成本、提升性能、迎头赶上，并建立起一个强大的产业。

They have a complete package and a couple years behind and they'll figure out how to make it cheaper slash do more slash catch up and and create a robust industry.

但有个原因，就像，我并不认为Jensen真的害怕AMD。

But there's a reason, like, I don't think that, like, Jensen is scared of AMD, really.

他很偏执。

He's paranoid.

我提过他很偏执。

I mentioned he's paranoid.

我肯定他有点害怕

I'm sure he's a little bit scared

他们。

of them.

对吧？

Right?

我觉得有些

Like, I think some of the

他们所做的那些事情，是对AMD或谷歌TPU等竞争动态的回应。

things that they've done are reactions and competitive dynamics with AMD or Google's TPUs or whatever.

对吧？

Right?

今天有个CoreWeave的交易，我认为这直接是谷歌行动的结果。

There's a CoreWeave deal today, and I think that's directly the result of what Google's been doing.

是的。

Yeah.

英伟达宣布的那个20亿管道，在

The 2,000,000,000 pipe that NVIDIA announced in

两个季度内。

two quarters.

向CoreWeave投资了20亿美元，但更重要的是，这就像只是个标签。

Invested 2,000,000,000 in CoreWeave, but what's more important is that that's like sort of just like the sticker.

真正相关的是，英伟达将与CoreWeave合作，去获取、担保以及处理所有这些事务——土地、电力、能源、输电，所有帮助建设数据中心的事情，所有这些资本方面的事务，因为英伟达资金雄厚，他们可以担保CoreWeave去做，这样CoreWeave就能成为那个产生人力的一方。

What's really relevant is NVIDIA is gonna work with CoreWeave to acquire and and backstop and all these things, the the land, the power, the energy, the transmission, all the help build the data center, all this capital side stuff that because NVIDIA has so much money, they can backstop CoreReave doing it because CoreReave then can be the one who generates the man.

总之，这有点像，因为谷歌之前就在这么做，他们和几家公司合作过，比如FluidStack、Terrowolf和Cypher。

Anyways, there's like, cause Google was doing that and they did that with like a couple companies such as FluidStack and Terrowolf and Cypher.

这些是已经宣布的一些公开交易。

These are some public deals that have been announced.

所以谷歌正在用TPU做这件事，而英伟达做出了反应。

And so Google is doing that with TPUs and NVIDIA reacted.

对吧？

Right?

嗯。

Mhmm.

同样地，我认为英伟达对AMD做出了反应，而且我认为关键在于英伟达对华为简直是怕得要死。

And so the same way I think NVIDIA's reacted to AMD and in the same way I think the thing is Nvidia is like deathly terrified of Huawei.

因为华为在遭到禁令之前，已经赶上了苹果，并且实际上成为了台积电最大的客户。

Because Huawei has caught up to Apple and actually surpassed them as TSMC's biggest customer before they got banned.

对吧？

Right?

他们确实刚刚击败了诺基亚、索尼爱立信等等。

They did just crush Nokia, Sony Sony Ericsson, etcetera.

没错吧？

Right?

整个电信供应链都是如此。

Like the entire telecom supply chain.

他们彻底摧毁了这些公司。

They just like completely destroyed them.

还有许多其他领域，比如他们直接推出了折叠屏手机。

And there's so many other areas like they straight up made a folding phone.

对吧？

Right?

你知道，我有一部三星折叠屏手机。

You know, I have a Samsung folding phone.

他们有一款折叠屏手机，比三星的还要好。

They have a folding phone that's better than Samsung's folding phone.

是的。

Yeah.

是的。

Yeah.

是的。

Yeah.

这简直了，兄弟，什么情况？

It's like, bro, what?

就像，你懂的吧？

Like, you know?

要知道，华为真的、真的很厉害。

You know, Huawei's really, really cracked.

所以，他们当然对华为感到恐惧，而且华为是世界上垂直整合程度最高的公司。

And so, of course, they're terrified of and and Huawei is the most vertical company in the world.

没有哪家公司比华为的垂直整合程度更高，这进而催生了巨大的创新。

No company is more verticalized than Huawei, which then leads to huge innovations.

这一点在美国我们并未完全认识到，但当你去欧洲旅行时，你会看到每个人用的都是华为手机。

It's something that we don't fully appreciate in The US, but like when you travel in Europe, you see everybody who's like honors on their phones.

而且，华为在手机领域的影响力非常巨大，以至于人们

And it's, the the the footprint of Huawei is huge in in phones in a way that people

一些旧手机。

A little old phones.

你知道，安防摄像头。

You know, security cameras.

是的。

Yeah.

实际上，他们认为他们拥有，你知道，有

Actually, they they think they have, know There's

像有一大群受过训练的内部测试人员。

like a lot of training on the that a a captive group of testers.

没错。

Exactly.

没错。

Exactly.

我觉得华为真的很可怕。

I think I think Huawei is terrifying.

对吧？

Right?

而且，是的，他们现在的芯片确实没那么好。

And and so, like, yes, their chips are not as good today.

这种情况已经发生了吗？

And is that is is that already happening?

我的意思是，显然美国和是两大市场，但对于其他市场，比如阿联酋、中东、欧洲，英伟达和华为已经在这些地方正面交锋了吗？

I mean, obviously, The US and China are the two biggest market, but, like, for other markets, I don't know, UAE, Middle East, Europe, are are NVIDIA and Huawei already head to head in

是运出了一些，但主要只是贴牌产能，比如没什么，不，不，我会说只是少量，比如几台服务器，而不是价值数十亿美元的东西。

in shipped a little bit, but like mostly just like sticker capacity, like there's nothing like no no like I would say like a little bit as in like a few servers, not like a billion dollars worth of stuff.

对吧？

Right?

关键在于中国的供应链必须加速发展。

The thing is China's supply chain has to ramp up.

对吧？

Right?

中国的明确目标是实现所有内部化，但像阿里巴巴这样的公司却说：我不想用华为。

China China's express goal is to have all inter inter internalized, but then like a company like Alibaba was like, I don't I wanna use Huawei.

对吧？

Right?

我想用英伟达，打造最棒的模型。

Like, I wanna make I wanna use NVIDIA and just make the best freaking models.

对吧？

Right?

因为这才是我的生意。

Because that's my business.

我的生意不是用华为的东西，而是被逼着这么做。

My business is not, you know, using a Huawei thing, but it's like, it's being pushed upon me.

还有其他公司，比如海康威视之类的。

There's other companies who like Camera Con and so on and so forth.

所以，这种供应链问题，中国的一些公司并不想用华为。

And so this sort of like supply chain, you know, companies in China don't wanna use Huawei.

他们显然是被鼓励的，被推动的，你知道，你必须这么做。

They're kind of encouraged obviously, pushed, you know, you must.

一些地方政府会说，你在这里做了这么多生意。

Some local provincial government be like, well, you're doing this much business here.

你就得这么做。

You gotta do this.

对吧？

Right?

就像有各种各样疯狂的事情，你知道，推动公司使用华为。

Like there's all sorts of like crazy stuff that, know, pushing of of companies to use Huawei.

挑战在于华为无法生产足够的量。

The challenge is Huawei can't manufacture enough.

对吧？

Right?

我们在这方面做了很多工作，并且免费提供，你知道，而不是像给客户那样，因为这有点像国家安全问题，也就是华为实际上是如何制造芯片的？

We've like done a lot of work on this, and we've just put it for free, you know, instead of like to our customers because it's like something that's like national security, which is how was Huawei actually building chips?

实际上，他们通过空壳公司从台积电获取芯片，并采用各种方法，比如将来自韩国的HBM（即内存）经由台湾偷偷运往中国。

Well, actually, they were using shell companies to get chips from TSMC and using different methods of like sneaking HBM, which is memory from, you know, Korea through Taiwan to China.

对吧？

Right?

我们报道过各种疯狂的事情，这就像打地鼠游戏一样。

Like all sorts of crazy stuff we've reported on and and people it's like a whack a mole.

对吧？

Right?

他们关闭了它，或者像那些运往中国的工具，这些工具本不应该用于制造尖端芯片，但实际上却被用于此目的。

They shut it down or like tools that get shipped to China and they shouldn't be for, you know, making leading edge chips, but they actually are.

所有这些情况的发生，是因为他们无法制造所有东西，如果他们想要制造尖端产品，确实需要在上游供应链方面相当依赖外国供应链。

And all these sorts of things are happening because they can't make everything and if they want to make the leading edge stuff, do need to rely on the foreign supply chain quite a bit in terms of the upstream supply chain.

对吧？

Right?

内存、逻辑芯片、晶圆厂工具、晶圆厂化学品等等。

Memory, logic chips, tools for fabs, chemicals for fabs, etcetera.

华为无法满足市场需求，因为中国国内在内存、逻辑芯片等先进前沿产能方面都不足。

Huawei cannot satisfy the market because there's not enough advanced leading edge capacity in memory, logic, know, and all these other things domestically in China.

他们正试图尽快建立这些产能，但这意味着目前就是无法满足市场需求。

And they're trying to build it as fast as they can, but that means there's just not enough to satisfy the market.

所以英伟达就有市场空间。

So Nvidia has a market.

我认为他们会想办法向中国销售芯片，而且黄仁勋现在应该就在中国，或者昨天刚来过。

I think they'll figure out how to sell chips to China and Jensen's in China I think like right now or was yesterday.

所以他显然是在想方设法把芯片弄进中国，因为英伟达的论点是如果我们卖给他们芯片，那么他们就不会...你知道...就不会有那么大的国内市场。

And so like he's clearly like wheeling and dealing to try and get his chips into China because you know, I think NVIDIA's argument is if we sell them chips then they won't you know, there won't be enough of a as much of a domestic market.

软件和其他一切的反馈循环也就不存在了。

The feedback loop for software and everything else won't be there.

这确实对他们构成了相当大的挑战。

That was sort of like really challenge it.

对吧？

Right?

大多数用于人工智能的开源软件都有很多中国贡献者。

Like most of the open source software for AI has a lot of Chinese contributors.

对吧？

Right?

VLM、PyTorch、SGLang，还有这些其他的库和工具，尤其是底层软件方面。

VLM and PyTorch, SGLang and like all of these other like libraries and things that are just like, you know, and and it goes to low level software especially.

对吧？

Right?

很多最好的开源项目其实是由一些中国公司决定开源的。

Like, a lot of the best open source stuff is actually just from like a Chinese company who decided to open source it.

模型也是如此。

And same with models.

对吧？

Right?

所以，好吧。

And so like it's like, okay.

那么，如果他们不能再使用英伟达芯片，这些开源软件就不会为英伟达芯片设计，而是为华为芯片设计，这难道不会削弱CUDA模式吗？现在不仅是中国国内，他们内部形成了一个反馈循环，然后还能将这种模式推广到世界其他地区。

Well, if they can't use NVIDIA chips anymore, then this open source stuff won't be designed for NVIDIA chips, it'll be designed for Huawei chips, and now does that like weaken the CUDA mode and now like not only is China domestic, now they have like a feedback loop internally and then they can externalize across the rest of the world.

对吧？

Right?

所以这就是英伟达提出的论点。

So this is the like argument NVIDIA makes.

我不太确定自己是否……你知道，我认为我的AI发展时间线非常快。

I'm not sure if I am like I'm like, you know, I think I think my AI timelines are so fast.

我没那么快。

I'm not that fast.

不是说像通用人工智能那种快，而是说，嘿，整个AI行业的收入已经达到1000亿美元了。

Like not in terms of like AGI, but like, hey, AI is a $100,000,000,000 of revenue across the industry.

我认为这个行业到今年年底可以实现1000亿美元的年经常性收入。

I think the industry could hit a 100,000,000,000 ARR by the end of this year.

比如OpenAI大约450亿美元，Anthropic大约354亿美元，然后还有谷歌的Vertex、DeepMind模型，比如Gemini。

Like $45.50 for OpenAI, like $35.40 for Anthropic, and then, you know, Vertex, DeepMinds models at Google, Gemini.

对吧？

Right?

然后是Anthropic模型的Vertex API、Bedrock API和Azure Foundry API。

And then Vertex API for Anthropic models and Bedrock APIs and Azure Foundry APIs.

我觉得，到今年年底，整个行业能达到1000亿美元的收入。

Like, I think a $100,000,000,000, like, end of this year.

嗯。

Mhmm.

这可不少。

That's a lot.

嗯哼。

Mhmm.

那么这1000亿美元的经济价值是多少呢？

And then what's the economic value of that $100,000,000,000?

现在，这其中有多少是在中国？

Now, how much of that is in China?

对吧？

Right?

中国的数字可能低10倍。

Like China's number is probably 10 x lower.

对吧？

Right?

因为他们还没能广泛地推广AI。

Because they just haven't been able to pervasively push AI.

对吧？

Right?

ChatGPT大约有十亿用户，然后再加上Gemini，Meta声称他们有五亿用户。

ChatGPT has a billion users roughly, and, you know, then you add on Gemini and Meta claims they have 500,000,000 users.

我不确定。

I don't know.

我觉得人们可能只是不小心点到了生成贴纸之类的功能。

I think people just accidentally click like generative sticker or something.

但不管怎样，西方对AI的使用已经相当广泛，而且还会继续增长。

But like anyways, like there's like there's like a lot of usage of AI in the West already, and it's gonna climb.

它会持续攀升。

It's gonna keep climbing.

而你，某种程度上必须去适应它。

And like you you kind of have to get used to it.

所以问题在于，你知道，这对世界的经济效益是什么？

And so like the question is like, do you you know, what's what's the economic benefit to the world?

对吧？

Right?

归根结底，这是一场经济战争。

And at the end of the day, this is an economic war.

对吧？

Right?

如果美国和西方在人工智能领域获胜并掌控局面，你知道，那些更强大的人工智能系统会形成这种反馈循环，从而促进经济增长、改进武器系统以及其他方面，比如电网工程、网络攻击等等诸如此类的事情，对吧。

If The US and the West win in AI and control, you know, more powerful AI systems that have this feedback loop that improve economic growth and weapon systems and whatever else, right, engineering of grids and cyber attacks and all these sorts of things.

如果他们拥有这种对中国的优势，中国就无法崛起为全球霸权。

They have this like advantage over China, then China will not rise to be the global hegemony.

但如果没有人工智能，中国肯定会崛起为全球霸权。

But without AI, China definitely will rise to be the global hegemony.

他们只是会超越美国。

They're just gonna outrun America.

所以问题在于，我认为这是另一种观点。

And so the question is like, you know, that's I think like the other view.

对吧？

Right?

那么，超级强大的人工智能系统，与中国建立自己落后几年的芯片、模型等国内生态系统，哪个更快？

And how fast are super powerful AI systems versus, you know, China building a domestic ecosystem for chips and models and everything that is a few years behind?

也就是说，真正的价值到底在哪里？

Like, what's the where what's what's actually the value?

对吧？

Right?

这大概就是围绕限制和监管展开的。

Like, that's sort of like around restrictions and regulations.

美国的本土化努力在这方面处于什么位置？

Where where do The US onshoring efforts fall in that category?

从《芯片法案》到正在建设的各种项目，你怎么看待这些举措？

What what do you make of them from the Chips Act to, like, all the thing that is being built?

顺便说一句，所有事情看起来都严重延迟了，这或许并不令人意外。

Everything looks like it's massively delayed by the way, which perhaps is not surprising.

我认为台积电正在制造晶圆，他们确实如此。

I think TSMC's manufacturing wafers and they're like Yeah.

他们在建造真正的晶圆和晶圆厂，而且你知道，还有其他一些已经宣布的晶圆厂，它们进展顺利，还有各种不同类型的工厂，比如一家韩国公司在德克萨斯州为他们的芯片建造一个随机的气体工厂。

Building real wafers and there's real fabs and like, you know, there's some other fabs that have been announced and like they're doing well and there's like a bunch of like different kinds of plants like a Korean company making a random gas plant in Texas for, you know, their chips.

对吧？

Right?

比如为了芯片，所有这些类似的事情正在发生。

Like for chips and all these like sort things are happening.

我认为《芯片法案》用其500亿美元的资金运作得相当不错。

I think the chips acted really well with its $50,000,000,000.

只是我觉得人们并不了解半导体产业的规模。

It's just I don't think people understand the scale of the semiconductor industry.

它是全球最复杂的供应链。

It is the most complicated supply chain in the world.

对吧？

Right?

这比制造飞机要大得多，你知道的。

It's much bigger than, you know, say manufacturing airplanes.

这比任何其他事情都要大得多。

It's much bigger than like, you know, really anything else.

对吧？

Right?

如果你看看全球前十大公司，我认为其中有八家都是设计半导体的。

If you look at the top 10 companies like of the world, I think eight of them design semiconductors.

对吧？

Right?

现在很明显，像谷歌这样的公司也设计半导体，但感觉就像是，哦，等等。

Now obviously like Google designed semiconductors, but it's like, oh, wait.

不对。

No.

但如果没有TPU，他们的搜索成本可能会高出10倍。

But their cost of search would be like 10 x higher if they didn't have TPUs.

而TPU（张量处理单元）是为搜索功能高度优化的。

And TPUs were super optimized for search.

对吧？

Right?

或者，你知道的，顺着这个名单往下看。

Or like, you know, you you you go down the list.

对吧？

Right?

就像Meta用他们自己的芯片来服务推荐系统。

Like Meta serves recommendation systems with their chips.

对吧？

Right?

就像，你顺着名单看下去，每个人都在制造自己的芯片。

Like, you go down the list, it's everyone is making their own chips.

如果苹果没有自己的芯片，他们的设备会明显差很多。

Apple devices would be materially worse if they didn't have their own chips.

是的。

Yep.

对吧？

Right?

你就顺着这个清单往下看。

And you just go down the list.

这就像是，这是最复杂的供应链，而且他们每年向芯片行业提供大约1500亿美元的补贴。

It's like, it's the most complicated supply chain and and they're they're spending something on the order of like $150,000,000,000 roughly in subsidies a year to the chip industry.

嗯哼。

Mhmm.

我们计划在十年左右的时间里投入500亿。

We are doing 50 over like a decade.

是的。

Yep.

这里的规模存在差异。

There's a difference in scale here.

对吧？

Right?

在台湾投入的资本支出总额已经超过5000亿美元。

The collective total amount of like CapEx that has been spent in Taiwan is like 500,000,000,000 plus.

对吧？

Right?

在整个行业，包括台湾所有制造半导体的公司。

Across the industry, across all the companies that are making semiconductors in Taiwan.

而且台湾没有本土产业。

And Taiwan doesn't have a domestic industry.

500亿美元的补贴如何能改变美国的指针？

How is $50,000,000,000 of subsidies gonna change America's needle?

对吧？

Right?

它确实稍微推动了一点。

It does move it a little bit.

对吧？

Right?

我想说清楚一点。

I I wanna be clear.

比如，《芯片法案》非常棒。

Like, the chips act is awesome.

我不太明白为什么电动汽车或者太阳能行业能获得如此庞大、高达万亿美元级别的支持计划。

I don't understand why like EVs or like solar was given this massive massive like trillion dollar package.