AWS 新融资下的AI未来

由米里亚姆·马雷达创立的Thinking Machines公司发布了他们的首个开源权重AI模型，名叫Inkling。这家公司因她在萨姆·阿尔特曼被罢免后担任CEO而闻名，之后便离开了OpenAI。

Thinking Machines by Miriam Maradi, who famously left OpenAI after she was the CEO when Sam Altman got kicked out, have released their first open weight AI model called Inkling.

OpenAI开发了一个叫GPT-RED的系统。

OpenAI has built something called GPT-RED.

这是一个超级黑客级别的大语言模型，他们打造它的目的是为了提升自家模型的安全性。

It's an LLM super hacker, and they built this to harden their own models.

AI Box已经推出了一款MPC服务器，能把80多个不同的AI模型整合进Claude、ChatGPT和Gemini里。

AI Box has shipped an MPC server that brings 80 different AI models into Claude, Chat Chat GPT, and Gemini.

AWS决定投入十亿美元，组建一支前沿工程团队。

AWS is committing $1,000,000,000 to a forward deployed engineering organization.

OpenAI和Anthropic都在扩大各自的规模。

OpenAI and Anthropic are both scaling theirs.

Apple Intelligence已获准在中国上市，合作方是阿里巴巴的通义；而Meta也计划效仿SpaceX的做法，开始销售和转售他们的AI算力资源。

Apple Intelligence has been cleared for a China launch with Alibaba's Quen, and Meta plans to sell and resell their AI compute following what SpaceX is doing and kind of copying their playbook.

咱们先从Thinking Machines说起。

Let's kick this off with Thinking Machines.

这家公司是我特别看好的一家。

This is a company that I'm really rooting for.

他们融了超过十亿美元，但一直没推出什么特别颠覆性的产品。

They've raised, you know, over $1,000,000,000, but they haven't come out with anything super groundbreaking.

不过说到这些大模型，成本太高、开发周期太长，像Anthropic这样的公司，当初感觉OpenAI已经甩开他们太远，根本追不上了。

But what I will say when it comes to a lot of these models, they're so expensive, and they take so much time that when you think of something like even Anthropic, it felt like OpenAI had just run away from them and they were never gonna catch up.

但他们一点一点找准方向，不断打磨产品，慢慢抢回市场份额，现在Anthropic已经爆发了，收入甚至超过了OpenAI。

And little by little, they they pick their lane, they make their product better, and they're able to get market share until the point where Anthropic has now exploded and is making more money than OpenAI had surpassed them in revenue.

我觉得很多AI公司如果不想被淘汰、被拆卖，或者被收购走人，很可能也会走同样的路。

I think we might see a lot of that same strategy played out by a lot of different AI companies, if they don't kinda shrivel up and die or get sold off for parts or, you know, an acquihire or something like that.

所以Thinking Machines就是我特别期待的公司之一，真希望他们能大有作为。

And so Thinking Machines is one of these companies that I'm really excited about, and I hope that they, really go places.

这家公司由Miriam Radhi领导，她就是Sam Altman离开时OpenAI的CEO，现在她推出了一个叫Inkling的东西。

But this is run by Miriam Radhi, who is famously the CEO of OpenAI when Sam Altman left, and she has created this thing called Inkling.

这是一个开放权重的AI模型。

It's an open weight AI model.

这个模型有9750亿个参数，企业可以下载后自行定制。

It has 975,000,000,000 parameters, and companies can download this and customize it themselves.

所以你不用再去花钱买OpenAI或者Anthropic的API接口。

So you don't have to go and pay for API access to OpenAI or Anthropic.

你可以直接自己动手去微调这个模型。

You can actually just go and fine tune this all on your own.

还能用你自己的数据来做微调。

You can fine tune on your own data.

他们希望人们相信，这种个性化调整的效果，会比OpenAI、Anthropic、Gemini这些大厂卖的通用模型更好。

And the bet that they want people to take is that that is going to be better than a one size fits all model that OpenAI or Anthropic or Gemini or any of these other big labs are selling.

Inkling是用45万亿个文本、图像、音频和视频令牌训练出来的。

Inkling was trained on 45,000,000,000,000 tokens of text, image, and audio and video.

他们只用了大约九个月就完成了，比OpenAI的五年或者Anthropic的三年快太多了。

They did it in about nine months, which is way faster than OpenAI's five year timeline or Anthropix's three years.

在与桥水基金的测试中，基于该基金自身专业知识训练的金融模型，在金融推理基准测试中达到了84.7%的准确率，成本却只有Anthropic和OpenAI顶级模型的约十四分之一。

In a test with Bridgewater Associates, the financial model trained on the hedge fund's own expertise scored 84.7% on financial reasoning benchmarks at roughly one fourteenth the cost of the top models from Anthropic and OpenAI.

所以当你用这些模型，并用自己的数据进行微调时，效果提升非常明显。

So we're seeing some massive improvements when you're taking this kind of these kind of models and you're fine tuning it on your own data.

这个模型在处理每个任务时，只激活9750亿参数中的410亿个，用户还可以调高计算强度，以牺牲速度换取更高准确率。

The model activates only 41,000,000,000 of its 975,000,000,000 parameters per task, and users can dial up thinking efforts to trade speed for accuracy.

我支持他们，但最终效果如何，还得看时间检验。

So I'm rooting for them, but time will tell how well this model does.

我们刚得知，OpenAI 开发了一个叫 GPT-RED 的系统。

We just learned that OpenAI built something called GPT-RED.

这是一个专门用来攻击自家系统的 AI 模型，目的是在发布前发现安全漏洞。

This is an AI model trained to attack their own systems, and they use it to catch security flaws before release.

这个模型成功将针对 GPT-5.6 的攻击成功率从90%以上降到了23%，堪称 OpenAI 迄今最安全的版本之一。

So the model cut successful attacks on GPT 5.6 from over 90% down to 23%, which basically makes it one of OpenAI's most secure releases yet.

其中一个特别有意思的攻击方式，叫‘新型虚假思维链’，就是诱使模型编造出根本不存在的推理步骤。

One particularly interesting attack that it was able to discover is called a novel fake chain of thought, which basically is tricking a model into making up fake reasoning steps.

这有点像让人相信一加一等于三，然后还说：‘我早就算过了。’

So it's kind of similar to convincing someone that, you know, one plus one equals three, and then, you know, saying that I already checked the math.

这就是关键所在。

That's what equals.

如果真是这样，那你就可以趁机在下一轮骗过它。

And if it equals that, then therefore and you, you know, go trick them on the next thing.

他们发现，当把同样的任务交给人类红队成员测试GPT-5时，GPT-RED找到的攻击方式比人类更有效。

So they found that when they gave the same task as a human red teamer who tested GPT-five in 2025, GPT-RED found more effective attacks than the humans.

他们还让它去攻击一个第三方自动售货机系统，这还挺搞笑的。

They also got it to, go and hack, like, this third party vending machine agent, which is kinda funny.

它确实有不少局限，比如不太擅长进行来回对话式的攻击，也不太能利用嵌入在恶意文本里的图片。

It does have a bunch of limitations, so it's not very good at back and forth conversation attacks, and it's not very good at exploiting images that are embedded into malicious text.

这个设计的核心思路是，让GPT-RED自动化发现所有安全漏洞——把攻击模型和防御模型放在一起，在模拟的真实环境中互相较量。

The idea behind this is actually functionally working is that GPT-RED is automating how all of the security holes are found because it puts an attacker model against a defender model, and they put them in these kind of simulated real world environments, and they're battling it out.

OpenAI并没有把这套系统对外公开，也不让其他人来测试，这点挺有意思的。

OpenAI is not releasing this externally, so they're not giving this to other people to test, which is interesting.

对吧？

Right?

因为我们手上有Anthropic的Mythos模型，它在安全漏洞利用方面表现非常出色，所以他们把它分发给了所有顶尖实验室，说：嘿。

Because we had the whole Anthropic Mythos model, which was really good at security exploits, they gave it out to all of the top labs and said, hey.

赶紧用这个来强化你们的安全系统。

Like, go harden all your security with this.

所以OpenAI并没有把这套工具对外公开，只供自己内部使用。

So OpenAI is not giving this out externally beyond just using it for themselves.

而且，这些漏洞的攻击成功率已经大幅下降了。

And, I mean, it's got a huge massive drop in the success rate of a lot of these exploits.

我觉得这说明，由AI驱动的红队测试，完全能和人类一样有效，甚至更胜一筹。

I think it shows AI powered red teaming can be just as effective or more effective than humans.

毕竟，AI能比人类多做无数倍的测试尝试。

I mean, you can just brute force way more, tests than humans.

它虽然是从人类的行为中学来的，但现在它的表现已经超过了大多数人类。

I mean, it's trained off of what humans are doing, but at this point, it's doing better than what a lot of the humans are doing.

AI Box——坦白说，这是我自己的创业公司——刚刚推出了一个MCP服务器，现在允许用户直接把80种不同的AI模型接入Claude、ChatGPT、Gemini或Cursor。

AI Box, which full disclosure, is my own startup, has just released an MCP Server, and we are allowing people to plug 80 different AI models straight into Claude, ChatGPT, Gemini, or Cursor.

简单来说，它让你能在自己平时用的助手里直接调用任何模型。

Basically, it lets you use any model inside of whatever assistant you already are used to using.

我平时基本全天都用Claude，不过最近也在慢慢转用ChatGPT的新应用。

Personally, I use, Claude all day long, although I'm kinda switching over to ChatGPT with their new ChatGPT app.

真的特别厉害。

It's really amazing.

但不管用哪个，只要接入AI Box的MCP，你就能使用OpenAI能生成的所有图像。

But both either way, you get the AI Box MCP, and it allows you to access all of the images that OpenAI can generate.

如果你用的是Claude或ChatGPT，还能生成Google V3能做的所有视频。

It allows you if you're on Claude or ChatGPT to generate all of the videos that Google v o three can make.

不管你在哪个平台，都能调用Eleven Labs的音频功能。

And no matter what platform you're using, you can access what Eleven Labs can do with audio.

今天一整天我都在为AI Box即将上线的大规模Facebook广告活动做准备。

I spent basically the entire day today getting ready for a big Facebook campaign that we're launching here at AI Box.

以前做Facebook广告，通常得专门请一个广告团队。

And in the past, doing Facebook campaigns usually meant hiring a Facebook ads team.

以前做Facebook广告得大量制作素材，特别耗时，还得花很多时间弄落地页。

It meant creating tons of creative, which just takes a lot of time, and it meant, you know, spending a ton of time on landing pages.

用了Claude和我们内置的AI Box MCP之后，我其实几乎搞定了所有内容——我就是用手机拍了一堆自己讲产品的自拍视频。

With Claude and with the AI Box MCP that we built into it, I actually got almost all of the so, basically, what I did is I went and recorded a bunch of just selfies of me talking about the product on my phone.

我把这些视频丢进一个文件夹，让Claude帮我全部剪辑好。

I dropped those into a folder, and I had Claude go and edit all of them.

然后我对它说：帮我生成一批动态图片广告。

And then I was able to say, hey, need to create a bunch of dynamic, so like image generated ads.

我告诉了它想要的风格。

I gave it the style that I wanted.

我脑子里有几个不同的创意。

I have these different ideas.

其中一些看起来像突发新闻的图片风格。

Some of them, it looks like kind of like breaking news images.

还有一些看起来像是我在开FaceTime通话，屏幕上弹出一条关于AI Box的短信。

Some of them, they actually look like it's like me doing a FaceTime call, and there's like a text message popping up on screen, and it's some sort of text message about AI Box.

总之，我收集了大量来自不同人的创意点子。

So anyways, I had all these different creative ideas that I went and got from a bunch of people.

要是每回都在Canva上做模板，再一个个换标题、换背景，那真得花不少时间。

And to go and, you know, create a template on Canva and then switch out tons of different variations of the title and the the backgrounds and stuff, like, that takes a lot of time.

以前Claude做不了，因为它根本没法生成图片。

And before, Claude couldn't do it because it couldn't do any of the image generation.

但现在AI Box已经集成到Claude里，能生成图片了，我就试了一个：我正在演示Claude生成图片的功能，需要把这个过程做成广告。

But now that AI Box is embedded inside of Claude and it can do image generation, one in particular that I did is I was showing, I was showing the ability to generate images inside of Claude, and I needed that generated as an ad.

所以你得设计出一个Claude界面的样子，还要能展示生成的图片，再加上AI Box的logo之类的元素。

So you had to have, like, basically, the design of a Claude interface, and then you had to be able to show images generated and, you know, have, like, AI Box logos and stuff.

于是我跟它说：

So what I what I did is I was like, hey.

你就在Claude里帮我做个这个模拟界面。

Go go create this mock up inside of Claude.

至于图片部分，直接用AI Box生成，再把生成的图嵌到它正在画的界面里。

But for the image part, just go use AI Box to generate the image and pull that inside of it's like the image inside of the image that it's generating.

几秒钟就能搞定，我觉得太棒了。

Is able to do that in a couple seconds, and I said, this is awesome.

你想想看，给我列出十种不同的使用场景吧。

Go think of, like, 10 variations or, like, 10 different use cases of my product.

这次呢，产品展示的就是你能生成图片这个功能。

And, you know, and and in this case, it's the product is showing that you can create images.

我知道这有点自指了，有点元。

This is kind of a meta example, I know.

我就说，你再给我想想十种不同的变体。

But I was like, go think of, like, 10 different variations.

于是它就说了，嗯。

So it's like, okay.

你可以用它来做用户生成的内容。

Well, you could use it for, like, UGC content.

还可以用来做产品图、Logo、通讯简报封面之类的。

You could use it for, like, coming up with product images, for logos, for newsletter banners, whatever.

想出了这么多点子。

Came up with all these ideas.

在它生成的广告图片里，每需要制作一个实际的图形，就会调用AI Box。

And inside of the image ad that it's creating, it would go to AI Box anytime it had to generate one of the actual graphics inside of its image.

不管是Logo的图形，还是UGC人物的图片，只要是需要的图像，它都会调用AI Box生成，然后自动嵌入到自己的内容里。

So, the, you know, the graphic of the logo or the graphic of the UGC person, whatever the actual image was, it would hit AI Box, it would generate the image, it would pull inside of its own thing.

所以我根本不用动手。

So I didn't have to do anything.

我就是说了一句：咱们有这个想法了。

I literally just said, go you know, we have the concept.

我也有自己喜欢的模板。

We have the template that I like.

去生成20个不同版本吧。

Go create, 20 variations.

想一些不错的使用场景，所有图片都交给AI Box来处理。

Think of a bunch of good use cases and use AI Box for all the images.

厉害的是，当Claude在做这个特定的界面原型时，它不是像我去问ChatGPT生成一张孤立的图片那样，那种图可能不完美，文字歪歪扭扭，人物也有点不对劲，对吧？如果你只是随便弄张图的话。

And the cool thing is when when Claude was making this specific kind of, like, UI mock up, because it's UI and it's not just like I went to to chat GPT and got it to generate like one singular image, isn't perfect and the text might be funky and someone might be a little bit off or whatever, right, if you're just doing an image.

因为Claude其实是用代码直接构建的，再把图片嵌进去。

Because Claude was kind of just creating it with code and then pulling an image into it.

我说，明白了。

I said, okay.

这太棒了。

This is awesome.

现在你把这个实际的图形放到我们网站上，再加个动画效果。

Now go put this actual graphic onto our website and animate it.

比如那个聊天气泡的地方，你就让一个人物在里面打字的样子。

So, like, where the chat bubble is, I'm like, have someone typing that out.

让气泡慢慢出现。

Have the bubble appear.

再加个加载动画，让图片弹出来。

Have a loading screen that makes the image pop up.

这太棒了，因为Claude能搭建这些东西，而我们通过AI Box直接引入图片，它就不只是张图，而是一个完整的可动画元素，能用在落地页上，还能直接转成Facebook广告，甚至做成视频广告。

So it's so cool because Claude is able to to build these things and because we're just pulling the image in with AI Box, it's not just a picture, but it's a full element that can be animated for the landing page, then it can be turned into, an ad for the Facebook ad, you can actually have the animated as a video ad.

总之，如果你在用Claude，但又没音频、视频或图像功能，赶紧去试试AI Box MCP。

Anyways, so many possibilities, but if you're doing anything with Claude which doesn't have audio, video, or image capabilities, just go get the AI Box MCP.

每月只要7.99美元，或者8.99美元，就能访问80多个不同的AI模型。

It's like $7.99 a month, think, or $8.99 a month, and you get access to over 80 different AI models.

你在用Claude时需要的任何功能，它都能帮你调用、整合，省下大量时间。

And anything you need while you're using Claude, it can go and grab that and pull it in and save you so much time.

再也不用来回折腾了。

No more back and forth.

赶紧去了解一下。

So go check it out.

描述里有个链接，aibox.ai/mcp，想试试的话可以直接点进去。

There's a link in the description to aibox.ai/mcp if you wanna get started with that.

亚马逊云正在投入十亿美元，组建一支全新的工程师团队，他们会深入客户公司，为每家企业定制专属的AI代理，项目完成后还会把完整系统交到客户手里。

AWS is committing $1,000,000,000 to a new team of engineers who are going to be embedded inside of customers or in their customers' companies, and they're gonna be building custom AI agents tailored to each business, and they're then gonna hand off the working system when the project ends.

Anthropic 和 OpenAI 这些公司最近冒出来不少。

There's been a ton of these companies being spun up by Anthropic and OpenAI.

他们其实是在派一批前线工程师，直接驻到客户公司里。

They're spending they're they're basically making these, like, forward deployed engineers.

你把工程师派到客户公司去。

So you send the engineer into your customer's company.

他们帮客户搭建一些自动化流程。

They build some sort of automation workflow.

我觉得这其实是 OpenAI 和 Anthropic 在说：嘿。

And I think it's kind of like OpenAI and Anthropics way of saying, hey.

你看。

Look.

我们知道你们想要 AI，但又不知道怎么用。

We know you you guys want AI, but you also are, like, don't know how to use it.

我们才是懂行的。

We are the experts.

我们直接派几个人过去，帮你把这事儿搞定。

We'll just send some people in there to build this thing for you.

他们完全是照搬了OpenAI花了40亿美元、Anthropic至今花了15亿美元才摸索出来的这套打法。

They're directly copying the playbook that that OpenAI spent $4,000,000,000 on and Anthropic spent $1,500,000,000 on so far.

在这两家企业里，OpenAI和Anthropic都跟一些私募基金合作，专门来资助这些现场工程师团队的开支。

And they both have like, in those cases, OpenAI and Anthropic both partnered with some private equity firms to actually fund the staff, for all of this FDE teams.

但Anthropic不一样，他们全靠亚马逊内部的资源来运作。

Anthropic is doing it entirely with internal Amazon resources instead.

说实话，如果我是这些大AI公司里的任何一个——比如Meta——正打算裁员的话，这招还挺有吸引力的。

And I'm gonna be honest, this might be an interesting strategy if I was any of these big AI companies and, you know, like let's say Meta, for example, and you were thinking about doing layoffs.

与其裁员，不如Meta干脆组建一支现场工程师团队，拉一笔钱，把Meta的产品深度嵌入客户业务里，直接把工程师派过去驻点支持。

Well, maybe instead of Meta doing layoffs, they should build one of these forward deployed engineer teams, get a bunch of money, and go and get, you know, Meta embedded into their customers' businesses and just deploy the engineers over there.

不过我也说不准，这到底是亚马逊在变相规避裁员，但这个思路确实值得琢磨，毕竟OpenAI和Anthropic当时根本没多余的人手，才不得不找外部合作。

Anyways, I'm not sure if this is Amazon getting around layoffs, but it is an interesting strategy you could imagine because OpenAI and Anthropic definitely didn't have any extra engineers that had partner with people for this.

AWS在结构上有个明显优势：项目结束后，客户还是会把AI应用跑在AWS的云上，这本质上就是在锁定长期收入。

AWS has a really big advantage, I would say, structurally, because when the project ends, the customer runs AI agents on AWS' cloud, and that's basically just creating long term revenue.

OpenAI 和 Anthropic 是按使用的 token 数量收费的。

OpenAI and Anthropic make money per token used.

每次互动的收益虽然少，但云服务对 AWS 来说却是巨大的回报。

That's a lot smaller of a payoff per engagement, but the cloud is a really big win for AWS.

再说，就算有人说 AWS 在模仿 OpenAI 和 Anthropic，毕竟这些模式是他们先搞起来的，但其实 Palantir 过去二十年一直在做类似的事。

Also, say all of this like an AWS is copying OpenAI and Anthropic, like they pioneered this, but this is actually something that Palantir has spent the last twenty years doing.

所以我觉得这事儿真不算什么新鲜玩意儿。

And so I don't think this is, you know, something that's just brand new in the situation.

这在过去早就有人做过，而且效果还不错。

It's it's something that's been done in the past, and it's it's kind of something that's been successful.

咱们就看看 Amazon 能把这模式做多大吧。

We'll see how big Amazon is able to scale this.

苹果刚刚获得了在中国推出 Apple Intelligence 的监管批准。

Apple has just won regulatory approval to launch Apple Intelligence in China.

他们正和阿里巴巴合作，为所有 AI 功能提供支持。

They're partnering with Alibaba to power all of the AI features there.

这挺有意思的。

It's interesting.

对吧？

Right?

因为苹果说，嘿。

Because Apple said, hey.

你看。

Look.

他们本来打算在Siri里用自家的AI，但要是这样，在中国肯定特别麻烦。

Like, they were gonna have their own AI in Siri, yeah, that probably would have been a big pain for them with China.

但他们后来又说，算了。

But then they said, look.

只要你是iPhone用户，想用哪个AI模型都行。

We're just gonna say anyone with an iPhone can, go and use whatever AI model they choose.

所以在中国，用户就会用阿里的模型；在美国，大概就是Anthropic、OpenAI这些，或者你平时付费用的那些模型，可以直接接入Siri。

So in China, that's gonna be, something from Alibaba, and in America, that's probably gonna be Anthropic or OpenAI or whatever the model that you probably pay for that you have premium of, you could probably plug that straight into Siri.

我觉得这很重要，因为中国是苹果的第二大市场。

I think this matters a lot because China is Apple's second largest market.

上个季度销售额高达205亿美元，我觉得能缩小这个差距，有助于苹果保住在中国智能手机市场的第二名位置。

20,500,000,000 in sales last quarter, and I think being able to close that gap helps keep Apple their number two smartphone position.

还有华为这样的竞争对手。

There's rivals like Huawei.

阿里巴巴的通义模型将负责所有设备——包括iPad、Mac和Vision OS——上的文本和图像理解与生成。

Alibaba's quen model is gonna handle text and image understanding and generation across all of the different, you know, iPad and Mac and Vision OS.

苹果正在和百度、深度求索以及字节跳动洽谈合作。

And Apple's exploring a deal with Baidu, DeepSeek, and ByteDance.

我觉得在和所有其他选项都谈过之后，他们最终还是选了阿里巴巴。

I think they finally have settled on Alibaba after kind of talking to all of the other options.

这个消息公布后，阿里巴巴在美国上市的股价上涨了6%。

Alibaba's US listed shares went up 6% when that was announced.

这笔交易本质上会让通义直接接入在中国发售的全新iPhone，获得现成的渠道分发。

This deal is basically gonna give Quen some built in distribution to the very new iPhone that will be sold in China.

我实话跟你说吧。

I'm gonna be honest.

我试过几个通义模型，真的挺惊艳的。

I've tried some cool some Quen models and been really impressed.

特别是他们的文本转语音模型，表现非常出色。

Their their text to speech model in particular is really good.

但对我来说，更关键的是，不只是苹果，任何西方公司，如果不用本地AI模型作为合作伙伴来驱动服务，中国都不会批准这些西方AI服务。

But kind of the bigger story for me on all of this is that for not just Apple, but for any of the Western companies, China is not gonna approve any sort of Western AI service if there isn't a local model partner that's locked in and that is powering everything.

Meta正在推出一项名为Meta Compute的云业务。

Meta is launching a cloud business called Meta Compute.

他们要把AI算力和AI模型直接卖给其他公司，就是为了跟亚马逊、谷歌和微软正面竞争。

They're gonna be selling AI compute powered AI models to other companies, and they're doing this directly to compete with Amazon, Google, and Microsoft.

就在Meta宣布投入高达1829亿美元建设AI基础设施之后，这简直是在复制SpaceX的打法。

This and all this is right after Meta committed a whopping a $182,900,000,000 to AI infrastructure, which is basically the SpaceX strategy.

他们正把多余的数据中心产能全部变成收入来源。

They're turning all of their excess data center capacity into revenue.

挺有意思的。

It's interesting.

对吧？

Right?

我觉得Meta原本希望他们的AI模型能被更多人使用，也更受欢迎。

Because Meta, I think, hopes that or would have hoped that their AI model was way more used and way more popular.

但这么多多余的AI基础设施，说白了简直就是一座金矿。

But having all of the extra AI infrastructure is, I mean, basically a gold mine.

他们本来不用花这笔钱，现在反而靠它赚钱了，有点像Grok的做法。

It's money that they're not spending and now money that they're actually making similar to Grok, I think.

Meta不仅要卖原始的计算能力，就像CoreWeave那样，还会提供自家的AI模型，包括最近推出的Muse Spark模型——这模型最近挺火的，终于让Meta站到了前沿。

Meta is gonna sell both raw computing capacity, basically just like CoreWeave does, and they're also gonna have access to their own AI models that they'll sell, including the recently launched Muse Spark model, which has, you know, gotten a lot of attention because it kind of finally pushes Meta to the forefront.

马克提到的那座俄亥俄数据中心，规模堪比曼哈顿，今年就要投入使用，正好能提供Meta转售所需的全部多余算力。

The Ohio data center that is described by Mark as a Manhattan sized project is expected to open this year, and it's gonna provide all of the excess capacity that Meta needs to resell.

SpaceX已经证明了这个模式是行得通的。

And SpaceX basically showed this model works.

X AI在五月与Anthropic、谷歌和Reflection AI达成了协议，购买算力和使用时间，把他们闲置的资源直接变成了收入，而且客户支付了相当可观的费用。

X AI signed deals in May with Anthropic, Google, and Reflection AI to buy compute, time, and they basically turned all of their idle capacity into immediate revenue, and people paid a lot of money.

我觉得Anthropic每月为使用这些资源支付的费用超过十亿美元。

I think Anthropic is paying over $1,000,000,000 a month for access to that.

对于资金雄厚的公司来说，这似乎是个不错的策略。

For companies with deep pockets, this seems to be a good strategy.

如果哪天Meta的AI模型变得特别特别火，他们就可以自己调用这些算力；但如果没火，他们照样能从其他AI公司身上赚得盆满钵满。

If at any point Meta's AI models get super, super popular, right, they can go and use their own capacity, but if not, they're just gonna make all of the money from all of the other, AI companies.

归根结底，AI的需求不会减少，这步棋其实挺聪明的。

And I think at the end of the day, the demand for AI is not gonna decrease, so it's a pretty smart bet.

好了，今天播客的内容就这些。

Guys, that was everything for the podcast today.

非常感谢大家收听。

Thank you so much for tuning in.

如果你喜欢今天的节目，别忘了去aibox.ai看看，就像我之前说的，那里能拿到我们的MCP，提供80种不同的AI模型。

If you enjoyed today's episode, make sure to go check out aibox.ai, like I mentioned, where you can get our MCP that gives you 80 different AI models.

更重要的是，如果你在用 Claude，你就能直接在里面生成图片、音频和视频。

And more importantly, if you're using Claude, you get images, audio, and video generated right inside of Claude.

用起来真的特别神奇。

It's, like, actually magical to use.

我一整天都在用它。

I've been using it all day.

连接只需要两秒钟，而且只要八美元。

It takes two seconds to connect, and it's only $8.

我真的求你们了，只要你在用 Claude，就一定要试试这个。

Please, I beg you if you have Claude, you have to try this out.

它会让你的生活轻松太多。

It will just make your life so much easier.

另外，如果你想把我在播客里聊的这些内容直接收到邮箱里，就去看看 aichatdaily.com。

Also, if you wanna get all of these different stories that I talk about on the podcast here straight into your inbox, go check out aichatdaily.com.

那是我的网站，和这个播客是连在一起的新闻站。

That's my website, my news site that is tied to this.

我为这里提到的每一个故事都写了深入分析的文章，还提供了链接，方便你进一步阅读。

I have deep dive articles on every story that I covered here, and I also have links over there where you can go and read more about it.

网站顶部有个订阅入口，你可以选择把内容直接收到邮箱，或者去网站上阅读深度文章。

There's a subscribe tab at the top of that website where you can get this all as an email, or you can go get the deep dives on the site.

非常感谢大家收听，我们下一期再见。

Guys, thank you so much for tuning in, and I will catch you in the next episode.