章节 1：AI 经济的企稳以及开发者首选大模型（LLM）的风向转变

📝 本节摘要：

OpenAI 一家独大的局面已经结束，开发者正根据具体需求（如编程能力、实时搜索准确性）在 Anthropic、Google 和 OpenAI 之间进行务实的选择。

1. AI 经济趋于稳定与创业常态化

2. 统治地位的更迭：Anthropic 超越 OpenAI

3. Google Gemini 的崛起

章节 2：模型“人格化”差异与用户的使用偏好迁移

📝 本节摘要：

本节深入探讨了开发者和用户选择模型的感性因素。除了编程能力，Google 的 Gemini 模型凭借其强大的搜索落地能力（Grounding）和准确性，在 YC 批次中的使用率飙升至 23%，甚至被部分用户视为 Perplexity 的替代品。对话中出现了一个生动的比喻：OpenAI 像高冷的“黑猫”，而 Anthropic 则像热情的“金毛猎犬”。尽管如此，ChatGPT 凭借其“记忆功能（Memory）”建立了强大的护城河，因其能深度理解用户个性而难以被替代。

[原文] [Speaker A]: How about Gemini How's Gemini doing in those rankings

[译文] [Speaker A]: 那 Gemini 怎么样？Gemini 在这些排名中表现如何？

[原文] [Speaker B]: Gemini is also pretty much has been climbing up pretty pretty high I think last year was probably singledigit percent or even like two 3% and uh now for winter 26 is about 23% And uh we personally been using also a lot of Gemini 3.0 and we've been impressed with the with the quality of it I think it's really really working

[译文] [Speaker B]: Gemini 的排名也爬升得相当、相当高。我想去年它可能只是个位数百分比，甚至像 2% 或 3%，而现在对于 Winter 26 批次来说，大约是 23%。而且我们个人也一直在大量使用 Gemini 3.0，我们对它的质量印象深刻，我觉得它真的非常有效。

[原文] [Speaker A]: I mean they have all different personalities don't they

[译文] [Speaker A]: 我的意思是，它们都有不同的“个性”，不是吗？

[原文] [Speaker B]: They got too Yeah It's uh it's kind of the classic where OpenAI sort of has the uh black cat energy and almost like uh Anthropic is kind of more the happy golucky a bit more very helpful golden retriever At least that's what I feel when I talk to them

[译文] [Speaker B]: 它们确实有。是的，这有点经典：OpenAI 有某种……“黑猫能量”，而 Anthropic 简直就像是那种乐天派、更加乐于助人的“金毛猎犬”。至少这是我跟它们交谈时的感觉。

[原文] [Speaker A]: And how about Gemini It's kind of like in between Har you prefer Gemini actually

[译文] [Speaker A]: 那 Gemini 呢？它有点介于两者之间。Har，其实你更喜欢 Gemini 对吧？

[原文] [Speaker B]: Yeah I switched to Gemini this year as my just go-to model I think even before 2.5 Pro came out and just seemed better at reasoning For me it was just like the increasingly I replaced my Google searches with Gemini and I just sort of trusted that Google's I think like the groundings API and its ability to actually like use the Google index to give you like real time information correctly I just found it was better than I personally I found it was better than all the other tools for that and it was better than Plexity on it too Like Plexity would be fast but not always accurate and Gemini was not quite as fast as perplexity but was always pretty accurate if I asked it about something that happened today

[译文] [Speaker B]: 是的，我今年把 Gemini 换成了我的首选模型。我想甚至在 2.5 Pro 发布之前我就换了，它似乎在推理方面表现更好。对我来说，就是越来越多地用 Gemini 取代了我的 Google 搜索。我有点信任 Google 的——我想是那个基础关联 API（Groundings API）——以及它利用 Google 索引为你提供正确实时信息的能力。我只是觉得在这方面它比……我个人觉得它比其他所有工具都好，甚至比 Perplexity 还要好。比如 Perplexity 可能很快，但不总是准确；而 Gemini 虽然不像 Perplexity 那么快，但如果我问它今天发生的事情，它总是相当准确。

[原文] [Speaker A]: Even if you use Gemini as the reasoning engine in Perplexity

[译文] [Speaker A]: 即使你在 Perplexity 里选择 Gemini 作为推理引擎也是这样吗？

[原文] [Speaker B]: I have not done that Interesting Yeah So it's hard to know like how much of it is the tooling and how much of it is like the base LLM That's fair

[译文] [Speaker B]: 我还没试过那样做。这很有趣。是的，所以很难知道这其中有多少是工具层面的原因，有多少是底层大语言模型（LLM）的原因。这很公允。

[原文] [Speaker A]: Yeah I mean what are your guys' tools of choice I haven't switched off of Chat GPT I mean I find the memory very sticky It knows me It knows my personality it and knows the things that I think about And so I'll use Perplexity for fast web searches or things that um you know I know was like a research task cuz I think Chetchup PT is still like a little bit of a step behind for searching the web

[译文] [Speaker A]: 是的，我是说你们大家的首选工具是什么？我还没有从 ChatGPT 切换走。我觉得它的“记忆功能（Memory）”非常有粘性。它了解我，它了解我的个性，也了解我在思考的事情。所以我会在需要快速网络搜索或者做一些……你知道，像研究任务之类的事情时使用 Perplexity，因为我觉得 ChatGPT 在网络搜索方面还是稍微落后了一步。

[原文] [Speaker A]: I don't know I think memory is turning into um an actual moat for like that consumer experience And I don't expect Gemini to ever have the personality that I would expect from Chat GPT It just feels like a different like entity you know

[译文] [Speaker A]: 我不知道，我认为“记忆功能”正在变成这种消费者体验的一条真正的护城河（Moat）。我不指望 Gemini 能拥有我所期待的 ChatGPT 那种个性。它只是感觉像是一个不同的……你知道，不同的实体。

章节 3：应用层的反击——模型编排层与套利机会

📝 本节摘要：

本节揭示了应用层初创公司如何通过“模型套利（Model Arbitrage）”夺回主动权。如同消费者手动在浏览器标签页中对比不同模型的答案一样，初创公司正在构建自动化的“编排层（Orchestration Layer）”。他们不再忠诚于单一模型，而是根据特定任务的表现和成本，实时切换模型（例如用 Gemini 做上下文处理，用 OpenAI 执行任务）。这种趋势将大模型推向了类似 Intel 和 AMD 的“大宗商品化”境地，预示着应用层将迎来丰收的一年。

[原文] [Speaker A]: Did you see Karpathy release like sort of a LLM arena of a sort which I mean I do by like hand right now using tabs It's like you have uh Claude open you have Gemini open you have Chetchip open and you give it the same task and then you take the output from each and then I usually go to Claude at that point and I'm like all right Claude this is what the other ones said what do you think and check each other's work

[译文] [Speaker A]: 你看到 Karpathy 发布的那种类似 LLM 竞技场（LLM Arena）的东西了吗？我的意思是，我现在是手动用浏览器标签页来做这件事的。就像你打开了 Claude，打开了 Gemini，又打开了 ChatGPT，你给它们同样的任务，然后提取每一个的输出。接着我通常会去问 Claude，我会说：“好了 Claude，这是其他模型说的，你怎么看？”让它们互相检查对方的工作。

[原文] [Speaker B]: think that that particular behavior at the consumer that level that we're doing startups are doing as well they are actually arbitrageing a lot of the models I had some conversations with a number of founders where before they might have been loyalist to let's say open AAI I models or entropic and I just had some conversations recently with them and these are founders that are running larger companies like series B level type of companies with AI they're actually abstracting all that away and building this orchestration layer where perhaps as each new model release comes out they can swap them in and out or they can use specific models that are better at certain things for just that

[译文] [Speaker B]: 我认为我们在消费者层面做的这种特定行为，初创公司也在做，他们实际上是在对很多模型进行套利（Arbitraging）。我之前和一些创始人谈过，以前他们可能是 OpenAI 模型或 Anthropic 的忠实拥护者。但我最近和他们交流——这些都是运营着较大规模公司、比如 B 轮融资级别的 AI 公司创始人——他们实际上正在把底层模型抽象化，并构建这种“编排层（Orchestration Layer）”。这样一来，也许每当有新模型发布时，他们就可以随意通过替换；或者针对某些特定任务，使用在这些方面表现更好的特定模型。

[原文] [Speaker B]: For example I heard from the startup they use Gemini 3 to do the context engineering which they actually then fed into OpenAI to execute it and they keep swapping it as new models come up and the winner for each category or type of agent work is different and ultimately they can do this because it it is all grounded based on the evals and the evals are all proprietary to them because they they're a vertical AI agent and they just work in a very regulated industry and they have this data set that just works the best for them

[译文] [Speaker B]: 例如，我听说有一家初创公司，他们使用 Gemini 3 来做上下文工程（Context Engineering），然后实际上将其输入到 OpenAI 去执行。随着新模型的出现，他们会不断进行替换。对于每一类任务或每一类代理工作，获胜的模型都是不同的。最终他们能做到这一点，是因为这都是基于评估（Evals）建立的。而这些评估对他们来说是私有的，因为他们是垂直领域的 AI 代理，在一个监管非常严格的行业工作，他们拥有的数据集对他们自己来说效果最好。

[原文] [Speaker A]: I think this is the new normal right now where people are expecting yeah the it's cool that the model companies they're spending all this money and making intelligence faster and better and we can all benefit Let's just do the best It's almost like the era of um Intel and AMD with new architecture would come up People could just swap them right

[译文] [Speaker A]: 我认为这就是现在的新常态，人们期望……是的，模型公司花这么多钱让智能变得更快、更好是很酷，我们都能从中受益。那就让我们用最好的吧。这简直就像是 Intel 和 AMD 的时代，当新架构出现时，人们可以直接把它们换掉，对吧？

[原文] [Speaker A]: Yeah It feels at the highest level that angst around where's the value going to acrew Is it going to go to the model companies or like the application layer are either startups feels like that es and flows in either direction a little bit throughout the year to me like I feel there are moments where like claude code amazing launch and it was like oh okay like the model companies are actually going to play at the application layer but then to me at least is all vibes based like Gemini surge especially over the last few months just feels like it returns us to a world of where exactly that like the models are all essentially commoditizing each other and it's just like the application layer and the startups are going to are set up to have another fantastic year if that continues

[译文] [Speaker A]: 是的。在最高层面上，感觉关于价值将流向何处的焦虑——是流向模型公司，还是应用层，或者初创公司——感觉这一年中这种焦虑在两个方向上起起伏伏。对我来说，有些时刻，比如 Claude Code 惊艳发布时，你会觉得：“噢，好吧，看来模型公司实际上也要涉足应用层了。”但随后的感觉，至少对我来说完全是基于“氛围（Vibes）”的，比如 Gemini 在过去几个月的激增，感觉就像把我们带回了一个世界：在这个世界里，模型本质上都在互相大宗商品化（Commoditizing）。如果是这样的话，应用层和初创公司似乎准备好迎接又一个丰收年了。

章节 4：AI 泡沫论辨析——以史为鉴的“安装期”与“部署期”

📝 本节摘要：

针对网络上关于“AI 泡沫”的质疑，演讲者将其与 90 年代电信泡沫进行了对比。当时光纤的过度铺设虽然导致了电信公司的崩溃，但廉价的带宽却催生了 YouTube 等应用。同理，现在的 GPU 和算力过剩将降低初创公司的成本。对话引用了经济学家 Carlota Perez 的理论，指出我们正处于高资本支出的“安装期（Installation Phase）”，而真正的应用爆发将在随后的“部署期（Deployment Phase）”到来，这对应用层创业者来说是巨大的利好。

[原文] [Speaker A]: I'm curious what you think Jared with some a lot of um perhaps the negative comments on Twitter around is this a bit of a bubble AI bubble

[译文] [Speaker A]: 我很好奇你是怎么想的，Jared，关于推特上那些……很多负面评论，说这是否有点泡沫？AI 泡沫？

[原文] [Speaker B]: Yeah When I talk to undergrads this is like a common question that I get is like oh like I heard it's a big AI bubble because like there's all this like crazy roundtpping going between Nvidia and OpenAI and like is it great is is it all fake

[译文] [Speaker B]: 是啊。当我和本科生交谈时，这是一个我常被问到的问题，就像：“噢，我听说这是一个巨大的 AI 泡沫，因为 Nvidia 和 OpenAI 之间有这种疯狂的循环交易，这真的好吗？还是一切都是假的？”

[原文] [Speaker B]: Yeah No this is fantastic Right Like people look at the telecom bubble it's like there's just you know billions of dollars like tens of billions hundreds of billions just like sort of sitting in a bunch of telecom back in like the you know '9s Actually that's why YouTube was able to exist right Like if you just have a whole bunch of extra bandwidth that isn't being used and is relatively cheap the cost is low enough for like something like YouTube to exist Like if there wasn't a glut of telecom then like maybe YouTube would have happened It just would have happened later

[译文] [Speaker B]: 是的，不，这太棒了。对吧？就像人们看电信泡沫，就像你知道的，有数十亿、数百亿资金就像是闲置在 90 年代的一堆电信设施里。实际上，这就是 YouTube 能够存在的原因，对吧？如果你有一大堆没有被使用的额外带宽，而且相对便宜，成本低到足以让像 YouTube 这样的东西存在。如果当时没有电信资源的过剩，那么 YouTube 可能还是会出现，只是会晚一些。

[原文] [Speaker B]: And then that isn't that like sort of what we're talking about here Like how do we we have to accelerate right We have the age of intelligence The rocks can talk they can think and they can do work And you just have to zap them more And you get like smarter and smarter stuff At this point I think the argument to college students is actually like because there will be a glut there is an opportunity for you And if there was not a gluten there wouldn't be as much competition the prices would be higher the margins lower in the stack would be higher right

[译文] [Speaker B]: 这不就是我们在这里讨论的情况吗？比如我们要如何加速？我们要加速，对吧？我们处于智能时代。石头会说话，它们会思考，它们能工作。你只需要给它们更多电力，你就会得到越来越聪明的东西。在这一点上，我认为对大学生的论点实际上是：正因为将会出现过剩（Glut），所以你们才有机会。如果没有过剩，就不会有那么多竞争，价格会更高，技术栈底层的利润率会更高，对吧？

[原文] [Speaker B]: And then you know what's one of the big stories this year Like Nvidia suddenly is on the outs Like I think their stock is today is like around 170s or something You know I think I'm still at long-term buy and hold honestly But for the moment people are like "Oh well Gemini is so good and all the you know nobody seems to be Nvidia only now and everyone's buying AMD and everyone's you know and TPUs are working." So you know at the moment it looks like there's you know what does that mean like there's competition and uh it means that there will be more compute not less and then that means that probably a little bit better things for all of the big LLM companies like sort of the you know the AI labs uh they get a little bit of power but you know they too are in competition with one another so then what does that mean well then it's you know go up another level in the stack right like as long as there are a great many AI labs that are in uh deep competition with one another then uh that's even better for that college student who's about to start a company at the application level

[译文] [Speaker B]: 然后你知道今年的一大新闻是什么吗？比如 Nvidia 突然失宠了。我想他们今天的股价大概在 170 左右。你知道，老实说我仍然坚持长期持有。但在目前，人们会说：“噢，Gemini 太棒了，而且你知道，现在似乎没人只用 Nvidia 了，大家都在买 AMD，而且 TPU 也能用了。”所以目前看起来……这意味着什么？这意味着有竞争，意味着会有更多而不是更少的算力。这也意味着对于所有大型 LLM 公司，比如那些 AI 实验室来说，情况可能会稍微好一点，他们获得了一点议价权。但你知道，他们之间也在相互竞争。那么这意味着什么呢？那就往技术栈的上一层看，只要有许多 AI 实验室在进行激烈的相互竞争，那么对于那些准备在应用层创办公司的大学生来说，这就更好了。

[原文] [Speaker A]: Yeah I think that's exactly right It's like people are asking this question like is it a bubble That's maybe a question that's really relevant if you're like the equivalent of like Comcast Like if you're Nvidia that's a very relevant question like oh are people overbuilding GPU capacity But like the college students they're not Comcast they're actually like YouTube If you're doing a startup in in your dorm room it's like the AI equivalent of like YouTube and like kind of doesn't really matter that much Maybe Nvidia's stock will go down next year I don't know But like even if it does that doesn't actually mean that it's like a bad time to be working on an AI startup

[译文] [Speaker A]: 是的，我认为完全正确。这就像人们在问这是否是一个泡沫。如果你是像 Comcast 那样的角色，这可能是一个非常相关的问题。如果你是 Nvidia，这是一个非常相关的问题，比如：“噢，人们是否过度建设了 GPU 产能？”但像大学生，他们不是 Comcast，他们实际上就像 YouTube。如果你在宿舍里创办一家初创公司，那就相当于 AI 时代的 YouTube，这对你来说其实没那么重要。也许 Nvidia 的股票明年会下跌，我不知道。但即使它跌了，这实际上并不意味着现在是做 AI 创业的坏时机。

[原文] [Speaker A]: Yeah To what Zach said on a podcast survey this year I think right It's like Meta may end up overinvesting like a significant amount in like the capex and infrastructure but like they essentially have to the big companies have to do it because they can't just like sit on the sidelines And in the case like demand falls off a cliff for some reason it's their capex not the startup's capex And there's still going to be tons of infrastructure and ideas to still continue building

[译文] [Speaker A]: 是的。正如 Zach 今年在播客调查中所说的，我想是对的。就像 Meta 最终可能会在资本支出（Capex）和基础设施上过度投资一大笔钱，但本质上他们必须这么做。大公司必须这么做，因为他们不能只是在场外观望。而且万一因为某种原因需求断崖式下跌，那是他们的资本支出，不是初创公司的资本支出。而且仍然会有大量的基础设施和想法可以继续构建。

[原文] [Speaker A]: There was this book written by this economist called Carlo Perez who studied a lot of uh tech trends and it studies a lot of um technology revolutions and it summarizes that there's really two phases There's the phase of uh installation which is where a lot of the very heavy capex investment come in and then there's the deployment phase where really ripples it where it rips and then everything explodes in terms of abundance and during the initial phase of installation is where it feels like a bubble There's a bit of a frenzy because it starts first with a there's this new technology that's amazing which happened with the chatb moment in 2023 Everyone got super excited about the tech and then everyone got super hyped and got into investing into a lot of the infrastructure with buying a lot of GPUs and all the giant gigawatt data center built out and then people's like but what is the demand What are going to be all the applications to be built out

[译文] [Speaker A]: 有一本由经济学家 Carlota Perez 写的书，她研究了很多科技趋势和技术革命。书中总结说，实际上有两个阶段。一个是“安装期（Installation Phase）”，这是大量重资产资本支出进入的阶段；然后是“部署期（Deployment Phase）”，在这个阶段影响会真正扩散，一切都会在丰富性方面爆发。在安装期的初始阶段，感觉就像一个泡沫。会有一点狂热，因为它首先始于一项令人惊叹的新技术，正如 2023 年 ChatGPT 时刻发生的那样。每个人都对这项技术超级兴奋，然后每个人都被超级炒作，开始投资大量基础设施，购买大量 GPU，建设所有那些巨大的吉瓦级数据中心。然后人们会问：“但是需求在哪里？要把什么应用建立起来？”

[原文] [Speaker A]: I think right now we're in that transition which is actually really good news for startup founders because they are not involved into the building the data centers but they're going to build the next generation of applications in the deployment phase when it really proliferates and what happened just going back to the analogy with with the era of uh the internet before the 2000 there was a lot of heavy capex investment into the telos right those were giant projects that college students wouldn't be involved but they were very heavily invested and in some cases were overinvested I mean this is the whole thing with dark fiber and some pipes that are not used And that's fine The internet ended up being still a giant economic driver And what that means is startups like the future Facebook or the future Google are yet to be started because those come in in the deployment phase because right now I think this things things are still getting built up I I do think the foundation lab companies and GPUs very much are falling into the bucket of infrastructure

[译文] [Speaker A]: 我认为现在我们就处于那个过渡期，这实际上对创业公司的创始人来说真的是个好消息。因为他们不参与建设数据中心，但他们将在部署期构建下一代应用，那时技术会真正普及。就像回到 2000 年前互联网时代的类比，当时对电信公司有大量的重资产资本支出投资，对吧？那是大学生不会参与的巨大工程，但它们获得了大量投资，在某些情况下是过度投资。我的意思是，这就像“暗光纤（Dark Fiber）”和一些未被使用的管道。但这没关系。互联网最终仍然成为了巨大的经济驱动力。这意味着像未来的 Facebook 或未来的 Google 这样的初创公司还没有开始，因为它们会在部署期出现。因为现在，我认为这些东西还在建设中。我确实认为基础实验室公司和 GPU 很大程度上属于基础设施这一类。

章节 5：突破物理瓶颈——太空数据中心与核聚变愿景

📝 本节摘要：

本节讨论了 AI 基础设施建设面临的物理极限——电力和土地短缺。面对这一严峻挑战，科技巨头和初创公司正在探索极端的解决方案：从 Boom Supersonic 利用喷气发动机发电，到将数据中心发射到太空（如 StarCloud 和 Google 的计划），甚至是在太空中实现核聚变（Zephyr Fusion）。这些看似科幻的举措实则是为了绕过地球上（特别是加州）繁琐的监管和资源瓶颈，确保未来三到五年的算力供应。

[原文] [Speaker A]: Yeah I mean it's interesting to watch uh how the stuff is evolving a little bit So do you remember summer 24 there was a company called StarCloud that came out and was one of the first to come out and say we're going to make data centers in space and what was the reaction when you know people laughed at them on the internet

[译文] [Speaker A]: 是的，我是说观察这些东西是如何演变的很有趣。你还记得 24 年夏天吗？有一家叫 StarCloud 的公司横空出世，他们是第一批站出来说“我们要把数据中心建到太空中去”的公司之一。当时的反应是什么？你知道的，人们在互联网上嘲笑他们。

[原文] [Speaker B]: Yes They said that's the stupidest idea ever you know I guess 18 months later uh suddenly Google's doing it Elon's doing it In every interview now apparently it seems to be like his top talking point

[译文] [Speaker B]: 是的，他们说那是史上最蠢的主意。你知道，我猜大概 18 个月后，突然间 Google 在做这事了，Elon 也在做这事了。现在显然在每一次采访中，这似乎都成了他的首要话题。

[原文] [Speaker A]: Yeah And so I mean why is that Like I feel like one of the aspects is that like part of the um infrastructure buildout right now that's so intense is like we literally don't have power generation

[译文] [Speaker A]: 是的。那么，为什么会这样呢？我觉得其中一个方面是，目前基础设施建设如此紧张，部分原因是我们真的没有足够的发电量。

[原文] [Speaker B]: Boom Supersonic instead of making supersonic jets right now is on this good quest to create enough power for a bunch of these AI data centers that are being built right now They use jet engines and even those like are so bad you know the supply chain for jet engines to generate power for data centers is so backed up that you know you would have had to have ordered these things you know two or three years ago just to even have it in two or three years from now you know these constraints uh end up like influencing like fairly directly what the giant tech companies need to do to win the game three or five years out

[译文] [Speaker B]: Boom Supersonic 现在不造超音速喷气机了，而是正在致力为目前正在建设的一堆 AI 数据中心创造足够的电力。他们使用喷气发动机，但即使是这样情况也很糟糕。你知道，用于数据中心发电的喷气发动机供应链已经严重积压，你必须在两三年前就订购这些东西，才能在两三年后拿到货。你知道，这些限制最终会相当直接地影响那些科技巨头为了在三五年后赢得比赛而需要做的事情。

[原文] [Speaker B]: Like suddenly there's not enough land You know in America we can't build The regulations are too high In California we have SQA which is totally abused by the environmental lobby to stop all innovation and building housing By the way we just don't have enough terrestrially like to just do the things that society sort of needs right now So you know the escape valve is like actually let's just do it in space

[译文] [Speaker B]: 就像突然之间没有足够的土地了。你知道在美国我们没法搞建设，监管门槛太高了。在加州我们有 SQA（注：此处应指 CEQA，加州环境质量法案），它完全被环保游说团体滥用来阻止所有的创新和住房建设。顺便说一句，我们在地球表面就是没有足够的资源来做社会目前需要的事情。所以你知道，逃生阀实际上就是：我们干脆去太空做吧。

[原文] [Speaker A]: Yeah Come to think about we we kind of have the trifecta of YC companies that are solving the data center buildout problem

[译文] [Speaker A]: 是的。想一想，我们好像有一个解决数据中心建设问题的 YC 公司“完美组合（Trifecta）”。

[原文] [Speaker B]: Well you need fusion energy

[译文] [Speaker B]: 嗯，你需要核聚变能源。

[原文] [Speaker A]: Yeah Yeah Well we have the company that's solving the no land problem by building the data centers in space We have Boom and Helion which are solving that we don't have enough energy problem

[译文] [Speaker A]: 是的，是的。我们有通过在太空建立数据中心来解决土地短缺问题的公司；我们还有 Boom 和 Helion 来解决我们能源不足的问题。

[原文] [Speaker B]: Just fun today uh a space fusion company that just graduated called Zephr Fusion And they actually had a great seed round out of Demo Day They're in their 40s They're national lab engineers who their entire careers they were building you know Tokamax and Fusion Energy And they came into the lab one day They looked at the physics They you know looked at the math and the models and they said "You know what If we did this in space it would actually pencil."

[译文] [Speaker B]: 有意思的是，今天刚好有一家刚毕业的太空核聚变公司叫 Zephr Fusion。实际上他们在演示日（Demo Day）之后完成了一轮很棒的种子轮融资。他们都是 40 多岁的人，是国家实验室的工程师，整个职业生涯都在建造托卡马克（Tokamaks）和研究核聚变能源。有一天他们走进实验室，看着物理原理，看着数学计算和模型，然后他们说：“你知道吗？如果我们在太空中做这个，实际上是行得通的（Pencil）。”

[原文] [Speaker B]: And so that's they're on like this sort of grand next 5 10 year quest to actually manifest it to actually create it in space uh because the equations say that it is possible and uh if they do it it's actually the only path to gigawatts of energy uh up there in space So you know it might be you know an even more perfect trifecta uh shortly

[译文] [Speaker B]: 所以他们正踏上未来 5 到 10 年的宏伟征程，试图将其变为现实，在太空中真正创造出它。因为方程显示这是可能的，而且如果他们做到了，这实际上是在太空中获得吉瓦级能源的唯一途径。所以你知道，这可能很快就会成为一个更完美的组合。

章节 6：AI 创业门槛降低与“微调陷阱”

📝 本节摘要：

本节讨论了 AI 创业技能的普及化（Democratization）。十年前，像 OpenAI 创始团队那样兼具研究、工程和商业头脑的人才极其罕见，但现在这种技能组合已变得普遍。这催生了大量针对垂直领域的初创公司（如医疗），它们通过在开源模型上进行微调（Fine-tuning）和强化学习（RL），在特定任务上击败了通用的巨头模型。然而，演讲者也发出了严厉警告：微调是一条危险的护城河。许多公司在旧模型（如 GPT-3.5）上辛辛苦苦取得的微调优势，往往会在新一代基础模型（如 GPT-4.5 或 5.1）发布时瞬间化为乌有。

[原文] [Speaker A]: Something else I feel like happened over the course of this year is the um interest in starting model companies like I guess maybe at both ends There's like the people who can raise the capital to go and actually try and build a head-on competitor to open AI which there are very very few like maybe have Ilio with SSI but then more so within YC people trying to build like smaller models

[译文] [Speaker A]: 我觉得今年发生的另一件事是，人们对创办模型公司的兴趣增加了。我想这可能体现在两个极端。一端是那些能够筹集资金、真正尝试去建立一个与 OpenAI 正面竞争的公司的人，这类人非常非常少，也许像 SSI 的 Ilya；但更多的是在 YC 内部，人们试图建立更小的模型。

[原文] [Speaker A]: Um I've certainly had more of those in the last few batches than before like whether it's sort of like a models to run on edge devices or maybe like a voice model specific to a particular language And I'm curious to see if that trend continues going back to this early era of YC Actually we sort of saw the explosion of just startups being created and maybe especially SAS startups Partly what what um fed that was just knowledge about startups became more dispersed there wasn't like a cannon of library information on the internet about like how to start a startup how to build software and then over like a decade that just became more common place and that just exploded like society's knowledge of startups and how to build things and it's may feels like maybe we're going through that moment in sort of the AI research meets like actually building things with with training models

[译文] [Speaker A]: 在过去几批次中，我确实看到了比以前更多的这类项目，无论是运行在边缘设备上的模型，还是针对特定语言的语音模型。我很好奇这种趋势是否会持续下去，就像回到了 YC 的早期时代。实际上我们当时看到了初创公司数量的爆发，特别是 SaaS 初创公司。部分原因是关于创业的知识变得更加分散了。以前互联网上并没有关于如何创业、如何构建软件的系统性知识库，但在大约十年间，这些变得更加普遍，社会对创业和构建事物的认知爆发了。感觉我们现在可能正经历这样一个时刻，即 AI 研究与实际通过训练模型来构建产品相融合的时刻。

[原文] [Speaker B]: I think we are absolutely going through that right now Yes where where it's going from being a very rare skill set to a more common one cuz like open AI a decade ago was like a rare like you needed you need like a a unique combination of skills right you need like your researcher brain your sort of like engineering brain maybe like your sort of finance business brain

[译文] [Speaker B]: 我认为我们现在绝对正在经历这个过程。是的，这正从一种非常罕见的技能组合变成一种更普遍的技能。因为像十年前的 OpenAI，那是非常罕见的，你需要一种独特的技能组合，对吧？你需要研究人员的大脑，某种程度上的工程师大脑，也许还需要一点金融商业头脑。

[原文] [Speaker A]: wait so did you just describe Ilia Greg and Sam you got it there was like a rare team right there just wasn't that configuration of team around very much and Now a decade later there's like a plethora of people who have like the research background the engineering background um the startup capital raising um background or at least going to be taught how to do all of that kind of stuff

[译文] [Speaker A]: 等等，你刚才是在描述 Ilya、Greg 和 Sam 吗？你说对了，那是一个罕见的团队，对吧？当时这种配置的团队并不多见。而现在，十年过去了，有太多人拥有研究背景、工程背景、创业融资背景，或者至少会被教导如何做所有这类事情。

[原文] [Speaker A]: And I'm curious if that would just mean we'll just see more applied AI company starting and maybe there'll be like even more models to choose from for all the various specific tasks

[译文] [Speaker A]: 我很好奇这是否仅仅意味着我们将看到更多应用型 AI 公司的诞生，也许会有更多模型可供选择，以应对各种特定的任务。

[原文] [Speaker B]: I think so So I think the other thing that's even contributing and making this a ever even bigger snowball is because of RL I think there's all these new open source models that people are doing the fine tune on top of it with a particular RL environment and task So it is very possible that you can create the best domain specific let's say healthcare model train on a generic open source model by just doing fine-tuning on it and doing arl it beats the regular big model actually I've heard and seen a number of startups where their domain specific model beats u openai let's say on healthcare there's this particular yc startup that told me that they collected the best data set for for healthcare care and they ended up performing better than OpenAI and a lot of the benchmarks for for healthcare with only 8 billion parameters

[译文] [Speaker B]: 我想是的。我认为另一件促成此事并让雪球越滚越大的事情是强化学习（RL）。我想现在有很多新的开源模型，人们在这些模型之上针对特定的 RL 环境和任务进行微调。所以非常有可能的是，你可以通过对通用开源模型进行微调和 RL，创建出最好的领域特定模型，比如说医疗模型，它实际上能击败常规的大模型。实际上我听说也看到过一些初创公司，他们的领域特定模型在医疗方面击败了 OpenAI。有一家特定的 YC 初创公司告诉我，他们收集了最好的医疗数据集，结果他们仅用 80 亿参数（8B parameters）就在许多医疗基准测试上表现得比 OpenAI 还要好。

[原文] [Speaker A]: I guess what's funny is uh you do need to have a post-raining infrastructure You we've also had YC companies where uh they had something that beat OpenAI uh you know GPT 3.5 and they were doing fine-tuning with RL but then uh yeah G GPT 4.5 and then 5.1 came out and uh you know basically blew their finetuning out of the water have to keep going

[译文] [Speaker A]: 我想有趣的是，你确实需要拥有“后训练基础设施（Post-training infrastructure）”。我们也遇到过这样的 YC 公司，他们做出了某种能击败 OpenAI（比如 GPT-3.5）的东西，他们利用 RL 进行了微调。但是后来，是的，GPT-4.5 甚至 5.1 出来了，然后你知道，基本上把他们的微调成果炸得粉碎。你必须不断前进。

[原文] [Speaker A]: Yeah Yeah You got to keep going Yeah I mean you actually have to continue to uh get to the edge

[译文] [Speaker A]: 是的，是的，你得继续前进。是的，我的意思是实际上你必须持续不断地去触达前沿。

章节 7：2025 年的行业复盘——经济企稳与技术“祛魅”

📝 本节摘要：

本节是对 2025 年 AI 行业的深度复盘。与 2024 年那种“脚下土地不断移动”的恐慌感不同，2025 年的 AI 经济已趋于稳定，模型层、应用层和基础设施层的格局日益清晰，构建 AI 原生公司的“剧本”也已成熟。

>

此外，演讲者驳斥了“快速起飞（Fast Takeoff）”的末日论调（如《AI 2027》报告）。他们指出，Scaling Law 的对数线性性质意味着技术进步需要指数级的算力堆叠，速度实则在放缓。更有趣的是，“人类不喜欢改变”和企业低效的 IT 现状，反倒成为了防止社会因技术剧变而崩溃的“刹车片”，给了政府和文化足够的时间去适应。

[原文] [Speaker A]: Yeah I was thinking about things that surprised me in 2025 And I think perhaps the thing that most surprised me is the extent to which I feel like the AI economy stabilized Like I feel like when we did this episode at the end of 2024 it felt like we were still in the middle of a period of incredibly rapid change where the ground was shifting under our feet and like nobody knew when the other shoe might drop and like what exactly was going to happen with startups and AI and the economy

[译文] [Speaker A]: 是的，我在想 2025 年让我感到惊讶的事情。我想最让我惊讶的可能是 AI 经济企稳的程度。感觉当我们 2024 年底做那一期节目时，我们还处于一个变化极其剧烈的时期，脚下的土地都在移动，没人知道另一只靴子什么时候会掉下来，也没人确切知道初创公司、AI 和经济会发生什么。

[原文] [Speaker A]: Now I feel like we've kind of settled into like a fairly stable AI economy where we have like the model layer companies and the application layer companies and seem and the infrastructure layer companies seems like everyone is going to make a a lot of money and there's kind of like a relative playbook for how to build an AI native company on top of the models I feel like things really kind of matured in that way which feels is all downstream of like the models themselves have incrementally improved this year but there haven't been like major steps forward that have shaken everything up which is has a knock on effect

[译文] [Speaker A]: 现在我觉得我们已经进入了一个相当稳定的 AI 经济体，我们有模型层公司、应用层公司，还有基础设施层公司，看起来大家都能赚很多钱。而且对于如何在模型之上构建一家 AI 原生公司，已经有了一个相对成熟的剧本（Playbook）。我觉得事情真的在这一方面成熟了，这感觉主要是因为模型本身今年只是在进行增量改进，而没有出现那种能颠覆一切的重大飞跃，这产生了一系列的连锁反应。

[原文] [Speaker A]: Many episodes ago we talked about how it was felt easier than ever to pivot and find a startup idea because if you could just survive if you could just wait a few months there was likely going to be some like big announcement that would completely make a new set of ideas possible and create more opportunities to build things It certainly feels like that has slowed down and so like finding ideas is sort of returning to sort of normal levels of difficulty in my experience in office hours

[译文] [Speaker A]: 很多集之前我们谈到过，那时感觉转型（Pivot）和寻找创业点子比以往任何时候都容易。因为只要你能活下来，只要你能等上几个月，很可能就会有一个重大发布，让一套全新的点子成为可能，并创造更多构建事物的机会。现在这种感觉肯定慢下来了，所以在我的办公时间（Office Hours）经验里，寻找点子的难度正在回归到某种正常水平。

[原文] [Speaker B]: I agree I'll tell you what's not a surprise Do you remember that report AI 2027 where it was just sort of like this doomer piece that said like oh well society is going to start falling apart in 2027 But you know at some point they quietly revised it to say that it wasn't 2027 but they kept the title Maybe it's not a surprise Like I was always a little bit of a skeptic um of like this fast takeoff argument because even with the scaling law it is uh log linear So it is slower It requires like 10x more compute and it's still sort of you know topping out right

[译文] [Speaker B]: 我同意。我告诉你什么是不令人惊讶的。你还记得那份《AI 2027》报告吗？那简直就是一篇末日论调的文章，说什么“噢，社会将在 2027 年开始崩溃”。但你知道，后来某个时候他们悄悄修改了说法，说不是 2027 年了，但标题还没改。也许这并不令人惊讶。我一直对这种“快速起飞（Fast Takeoff）”的论点持怀疑态度，因为即使是缩放定律（Scaling Law），它也是对数线性的（Log Linear）。所以它比较慢，它需要 10 倍以上的算力，而且你知道，它还是在某种程度上趋于平缓，对吧？

[原文] [Speaker B]: And uh that's one form of good news Another form of it's weird to call this good news but human beings uh don't like change in our previous episode where we sort of blew up that uh MIT report that said that you know 98% or 90% of uh enterprise AI projects fail Well it turns out that 90% of uh enterprises don't know how to do you know it let alone AI It's weird to say that that's a good thing but in the context of fast takeoff like that is a real break on the ability of this new really insane technology from actually permeating society

[译文] [Speaker B]: 这是好消息的一种形式。另一种形式——虽然把这称为好消息有点奇怪——是人类不喜欢改变。在我们之前的一集里，我们抨击了那个 MIT 的报告，报告说 98% 或 90% 的企业 AI 项目都失败了。好吧，事实证明，90% 的企业连 IT 都不知道怎么搞，更别说 AI 了。把这说成是好事很奇怪，但在“快速起飞”的背景下，这实际上是一个真正的刹车（Brake），阻碍了这种疯狂的新技术真正渗透到社会中。

[原文] [Speaker B]: I love to accelerate but like it's weird to say like oh well actually in this case maybe that's a good thing right like it is a shockingly powerful technology but you know between being log linear scaling and human beings really don't like change like organizationally speaking society will absorb this technology everyone will have enough time to sort of process it like culture will catch up governments will be able to respond to it not in like a frantic SP 1047 sort of like you know let's stop all the compute past 10 the 26 right Like just these knee-jerk responses to technology

[译文] [Speaker B]: 我喜欢加速，但说“在这种情况下这可能是件好事”确实很奇怪。这是一种极其强大的技术，但在“对数线性缩放”和“人类极其厌恶改变”这两个因素之间，从组织层面来说，社会将能够吸收这项技术。每个人都将有足够的时间来消化它，文化会跟上，政府也能够对此做出反应，而不是像疯狂的 SB 1047 法案那样——你知道，那种“让我们停止所有超过 10 的 26 次方算力”的膝跳反应式（Knee-jerk）技术监管。

章节 8：初创企业的扩张真相——人效比、融资护城河与未来预测

📝 本节摘要：

本节通过回顾 Harvey、Gamma 等案例，揭示了 AI 初创公司在扩张期的真实面貌。演讲者观察到，虽然早期初创公司能以极少人数达到百万营收，但为了满足客户不断提高的期望并应对激烈的同质化竞争（如 Harvey 对战 Lora），它们最终仍需回归传统的团队扩张模式。此外，对话犀利地指出了“微调（Fine-tuning）”往往是无效的资本燃烧，真正的护城河有时仅仅是巨大的融资额本身。最后，Gamma 仅用 50 人实现 1 亿美元营收的“反向炫耀（Reverse Flex）”，被视为未来企业追求高人效比的新标杆。

[原文] [Speaker A]: And something that surprised me sort of relate to that with the startups is I remember around this time last year we were talking about how companies are getting to a million dollars AR and raising series A's without hiring like some cases not hiring anyone just the founders maybe hiring one person which just felt very unusual I feel like a year on that hasn't translated into okay and then they went and hit like 10 million AR or like they they scaled without adding any more people to No they turned around and started and started hiring like actual teams

[译文] [Speaker A]: 还有一件让我惊讶的事与初创公司有关。我记得去年这个时候，我们还在讨论公司如何在不招聘任何人的情况下——有些情况下真的只有创始人，或者只招了一个人——就达到了 100 万美元的 ARR（年经常性收入）并完成了 A 轮融资，这感觉非常不寻常。我觉得一年过去了，这种情况并没有转化为“好的，然后他们达到了 1000 万 ARR”或者“他们没加人就扩张了”。不，他们转过身来，开始招聘真正的团队了。

[原文] [Speaker B]: Yeah like post series 8 it actually largely feels like the playbook is the same and the companies might be smaller for the same amount of revenue but it feels it's entirely because they hit the revenue so fast and there's still just bottleneck on how long it takes to hire people versus they have demand for less people

[译文] [Speaker B]: 是的，就像在 A 轮（原文误识别为 Series 8）之后，感觉剧本在很大程度上是一样的。虽然对于同等收入规模来说，公司人数可能变小了，但这完全是因为他们达到该收入的速度太快了，瓶颈在于招聘需要时间，而不是因为他们对人员的需求减少了。

[原文] [Speaker B]: I think they're like there might be a case of two fuagra startups like one being Harvey and the other one being open evidence right Harvey the founders are incredible uh they were you know very early and then there's this sort of idea of like for VCs you could just go down Sand Hill Road and like the fixes in like you just sort of block out all of them and then all the people you know there maybe 30 people who could write checks of like 10 to 100 million and if you just sort of get all of their money like there's sort of no one who can actually come in and do the next series A and then basically you're safe like you have capital as a budgeon is capital as a moat in that case right so yeah Harvey is interesting because you know uh Lorra is coming fast for them and obviously we have some skin in the game on Lora but we think that they have uh as good a shot at any

[译文] [Speaker B]: 我觉得可能有这种“填鸭式（Foie Gras）”初创公司的案例，比如 Harvey 和 Open Evidence，对吧？Harvey 的创始人非常不可思议，他们起步很早。然后这就有了某种针对 VC 的策略：你可以沿着沙山路（Sand Hill Road）走一圈，就像是把局做好了，你把他们都搞定。你知道大概只有 30 个人能开出 1000 万到 1 亿美元的支票，如果你把他们的钱都拿了，基本上就没人能进来投下一轮竞争对手的 A 轮了。然后你就安全了，就像你把资本作为一种大棒，在这种情况下资本就是护城河，对吧？所以 Harvey 很有趣，因为你知道 Lora 正在快速追赶他们。显然我们在 Lora 上有利益相关（Skin in the game），但我们认为他们和其他人一样有机会。

[原文] [Speaker A]: I guess that's one trend that we saw in 2025 is that there was like a first wave of like AI hative companies like Harvey who might have wasted a lot of money on finetuning actually totally that like broke out really in 2023 and kind of did a victory lap that you know oh we've won the the space and now we're seeing a second wave of companies like Lora and Giga and it turns out that like oh actually like it isn't so simple

[译文] [Speaker A]: 我想这是我们在 2025 年看到的一个趋势：第一波 AI 原生公司，像 Harvey，可能在微调（Fine-tuning）上浪费了很多钱。实际上完全是这样，他们在 2023 年爆发，像是跑了一圈胜利以此庆祝说“噢，我们赢得了这个领域”。而现在我们看到了像 Lora 和 Giga 这样的第二波公司，结果证明，“噢，实际上事情没那么简单”。

[原文] [Speaker B]: Yeah The weird beneficiary of um you know burning some non-trivial double-digit percentage of your capital stack on fine-tuning that buys you no advantage is like basically the investors are the only winners there because they just own more of your company you know

[译文] [Speaker B]: 是的。你知道，为了那些根本买不到任何优势的微调，烧掉你资本结构中不可忽视的两位数百分比的钱，这种行为的奇怪受益者基本上只有投资者，因为他们因此拥有了你公司更多的股份，你知道的。

[原文] [Speaker A]: Yeah at least as it relates to like the the hiring and team size I feel like of the two camps one being the AI is going to make everything more efficient You will need less people and the other AI is going to reduce the cost of like producing the time to produce things and so then the expectations from your users and customers will just go up and you'll need to keep hiring more people to satisfy like the growing expectations I feel like this year has been more in that second camp

[译文] [Speaker A]: 是的。至少在涉及招聘和团队规模的问题上，我觉得在两个阵营中——一个阵营认为 AI 会让一切更有效率，你需要的人更少；另一个阵营认为 AI 会降低生产成本和时间，因此用户和客户的期望会上升，你需要继续招聘更多人来满足这些增长的期望——我觉得今年更多是属于第二个阵营。

[原文] [Speaker A]: and I think that is what's driving the fact that the companies are still just hiring as many people as they were preai is just like the bar for what the soft what their customers expect and they're all in you know like Lora's racing with Harvey Giga's racing with Sierra Like they're all still competing for the same set of customers and they still ultimately are bottlenecked on like people and like I don't think anyone's bottlenecked on ideas but they're bottlenecked on like people who can execute really well

[译文] [Speaker A]: 我认为这就是导致公司招聘人数仍与 AI 时代之前一样多的原因。这就像是客户期望的门槛提高了。而且他们都在竞争中，你知道，比如 Lora 在和 Harvey 赛跑，Giga 在和 Sierra 赛跑。他们都在争夺同一批客户，最终仍然受制于人。我不认为有人受制于点子，但他们受制于那些能真正出色执行的人。

[原文] [Speaker B]: I agree with you that like the uh era of the one person running a trillion dollar company is not here Not yet

[译文] [Speaker B]: 我同意你的看法，一个人运营一家万亿美元公司的时代还没有到来。还没到。

[原文] [Speaker A]: But I think it's going to trend that way eventually That'll be a wild time Maybe that's a prediction for 2026 Yeah You think it's coming in I mean I don't think it'll happen in 2026 either honestly I mean I think you will have many stories of companies run by you know under a hundred people that are making hundreds of millions of dollars So I mean Gamma was interesting to see like uh one of the biggest things that they said in their launch that I think is a very good trend is they said they got to hundred million dollars in ARR with only 50 employees So which is a very different it's you know such an inversion right like normally you have the big banner and the like little X thing you know image and it's like oh yeah like we raised all this money and look at all the people who work for us It's a good trend to have the reverse flex which is like look at all this revenue and look how few people work for us

[译文] [Speaker A]: 但我认为最终会朝着那个方向发展。那将是一个疯狂的时代。也许那是对 2026 年的预测。是的，你觉得它会来吗？我是说，老实讲我不认为 2026 年会发生。但我认为你会听到很多故事，关于不到 100 人的公司创造了数亿美元的收入。我的意思是，Gamma 很有趣，他们在发布中提到的最重要的事情之一——我认为这是一个非常好的趋势——是他们仅用 50 名员工就达到了 1 亿美元的 ARR。这非常不同，你知道这完全是一种反转，对吧？通常你会看到那种大横幅和某种 X 图片，然后说：“噢是的，我们融了这么多钱，看我们有多少员工。”这种“反向炫耀（Reverse Flex）”是一个好趋势，即：“看我们有这么多收入，再看我们只有这么点员工。”

[原文] [Speaker B]: Well that's all we have time for this time We just wanted to wish you a really happy holidays and happy new year from all of us to you and yours See you next time

[译文] [Speaker B]: 好了，这次我们的时间就到这里了。我们只想祝大家节日快乐，新年快乐，献上我们要给你和你家人的祝福。下次见。