跨国深谈

Scale

14分钟阅读

跨国深谈← 返回←Scale
SC

Scale

公司
本站收录 5 集 · 75 条金句 · 关联 10 个

集里怎么说它

① 提到它的金句

75 条

对我来说，这很多就像是一个自动驾驶的迷你版本，就是，最终你在大规模上证明这件事的唯一方式，是让这个东西谈判足够多的 NDA，以至于人们就是信任它。
Like to me, a lot of this is like, this is like a mini version of self-driving where it’s like, the only way eventually you prove this at scale is you have this thing negotiate so many NDAs that people just trust it.
—— Gabe · [42:18]

指向原始笔记的链接

他们不会去想什么永久底层阶级之类的事情，只想着如何把模型扩展得更大，他们的驱动力是工程，而不是政治、地位或别的什么。
They don’t think about things like the permanent underclass but only how to scale even bigger models and they are driven by engineering and not by politics or status or whatever.
—— Florian · [24:12]

指向原始笔记的链接

Cloudflare 在大规模上做很多事情，但我们现在必须忍受我们自己的规模反作用于我们。就像行星计算机将开始攻击自己。
Cloudflare does a lot of things at scale, but we now have to tolerate our own scale back at us. It’s like the planetary computer is going to start attacking itself.
—— Matt Carey · [10:48]

指向原始笔记的链接

问题在于，在医疗健康领域，在不同国家打造跨国业务是如此困难，以至于它成了一道不可逾越的屏障，这意味着没有参与者能够实现国际化扩张。
The issue is that, in health, as it is so difficult to create something international in the different countries, it has become an insurmountable barrier, which means that no actor has been able to scale internationally.
—— Charles Gorintin · [11:56]

指向原始笔记的链接

也没有互联网规模的分发能力，你就是历史上贬值最快的资产。
And internet scale distribution, you’re the fastest depreciating asset in history.
—— David George · [21:11]

指向原始笔记的链接

客户愿意付钱给他们，也非常看重他们的产品，但他们无法扩展，因为一旦扩展，他们可能会扩展到破产。
Customers want to pay them, and they really value their product, but they cannot scale because once they scale, they could scale into bankruptcy.
—— Lin Qiao · [16:27]

指向原始笔记的链接

在整个开源模型、替代模型、模型优化市场真正起飞之前，这家公司就已经在一个可能会让大多数人震惊的规模上疯狂盈利了。
This company is wildly profitable at a scale that would probably shock most people before this whole open source model, alternative model, model optimization market really takes off.
—— Matt Murphy · [28:41]

指向原始笔记的链接

如果我们扩展规模，仅靠扩展规模并没有帮助。你需要增加模型的多样性。
If we scale, scaling alone doesn’t help. You need to increase the diversity of the models.
—— Damian Borth · [24:25]

指向原始笔记的链接

对其中很多平台来说,你可以非常快地搭建出一个东西,但如果你真的想把它投入生产或扩展规模,平台就到顶了。
With many of them you could build something very fast, but if you actually wanted to bring it into production or scale it, the platform maxed out.
—— Carlos González de Villaumbrosia · [06:35]

指向原始笔记的链接

所以我真的认为，对于初创公司来说，从来没有比现在更好的时机去销售，并让自己的产品从一开始就获得相当有规模的采用。
And so I really think there’s never been a better time for startups to sell and to have their products get adopted at a pretty meaningful scale right out of the gate.
—— Patrick Collison · [28:36]

指向原始笔记的链接

所以这现在是一个验证装置,但它不再需要一个晚上,他们制造了一个速度快 300,000 倍并且几乎与运行全尺寸模拟器一样准确的东西。
So this is now a validation device, but instead of it taking a night, they made something that was 300,000 times faster and nearly as accurate as running the full scale simulator.
—— Jeff Dean · [45:17]

指向原始笔记的链接

本质上，对抗规模的结构将总是输掉，而引导规模的结构将总是获胜。
Essentially, structure that fights scale will always lose, and structure that channels scale always wins.
—— Dmitri Dolgov · [31:25]

指向原始笔记的链接

一个如此智能、无所不知、可以扩展并分发到宇宙四角的 AI，不可能只待在地球上的数据中心里，不可能属于一个人。
An AI that is so intelligent, that knows everything, that can scale and be distributed to the four corners of the universe cannot sit in a data center on Earth, cannot belong to one person.
—— Paolo Ardoino · [16:05]

指向原始笔记的链接

也许等我们规模扩大 10 倍时这就不适用了，但现在我们只在乎用最好的工具产出最高质量的成果。
Maybe that doesn’t scale when we’re like 10x the size, but right now we just care about using the best tools to produce the highest quality output.
—— Ali Haghani · [11:17]

指向原始笔记的链接

这就是为什么履约是极少数随着规模扩大、单位成本实际上变得更贵的事情之一。
And that’s why fulfillment is one of the only things that as you scale, it actually gets more expensive on a per unit cost.
—— Kevin Gibbon · [17:08]

指向原始笔记的链接

四面墙的诊所加上一屋子设备，无法扩展到今天的需求和未来的需求。
The four wall clinic with a room full of equipment does not scale to the demand today and the demand to the future.
—— 嘉宾 · [45:15]

指向原始笔记的链接

所以归根结底，付费是验证你拥有 PLG、验证你拥有一个能以任何方式、形状或形式扩展的产品最简单的方法。
So at the end of the day, paid is the easiest way to validate that you have PLG, that you have a product that can scale in any way, shape, or form.
—— Matt Swulinski · [08:01]

指向原始笔记的链接

我们的态度是,你把我们打得落花流水,来告诉我们你做对了什么,来替我们运营这个业务,因为在没有我们的资金、我们的资源、我们的规模的情况下你都做得这么好,所以你一定悟出了什么。
Our attitude is, you kicked our ass, come tell us what you did wrong, come run this for us because you did well without our money, our resources, our scale, so you must have figured something out.
—— Nikesh Arora · [21:04]

指向原始笔记的链接

我们模式的美妙之处在于，我们主要是一个可变成本模式。这意味着当需求下降时，很好，我们的成本随之下降。但这也意味着我们确实没有固定的产能可以送出去。
The beauty of our model is we’re primarily a variable cost model. It means when demand drops, great, our costs scale down with it. But it also means that we just don’t have a fixed capacity to give away.
—— Andrew MacDonald · [11:46]

指向原始笔记的链接

你需要花哨的算法，但你想要那些将随计算扩展的花哨算法。
You need fancy algorithms, but you want fancy algorithms that will scale with computation.
—— Rich Sutton · [09:33]

指向原始笔记的链接

最值得一提的是，我创建了 Folding at Home 这个大规模超级计算机项目，它基本上让我们预览到了如今在 AI 中看到的许多问题，尤其是「苦涩的教训」——算力如此重要这一点。
Most notably, I built Folding at Home, the large scale supercomputer project, which basically allowed us to preview a lot of the issues that we now see in AI, especially the bitter lesson that compute is such a big deal.
—— Vijay Pande · [04:11]

指向原始笔记的链接

他们创造的奖励信号还不够纯粹，无法支持前沿公司运行 RL 的规模。
And the reward signals that they are creating are just not pure enough to support the scale at which the frontier companies are running RL.
—— 嘉宾 · [03:38]

指向原始笔记的链接

你可以把这些护栏直接内置到钱包里，而你作为人类不必逐笔批准每一笔交易，因为那样根本无法扩展
You sort of build these guardrails into the wallet, and you don’t have to, as a human, approve every single transaction, because that would just not scale
—— Harshal Bhangale · [09:32]

指向原始笔记的链接

我认为这需要信心——相信如果你把东西造好，解决以前没被解决的问题，规模化会随后而来。
I think it takes confidence to believe that if you build something well, you can solve problems that have never been solved before that will scale later.
—— Nick Noone · [41:22]

指向原始笔记的链接

它从根本上改变了我能做的事情的规模，我过去常常碰壁的时间、精力、能力等限制要低得多了。
It has fundamentally changed the scale of what I can do, the limits of what I use to bump up against time, energy, capacity are way lower.
—— 嘉宾 · [19:12]

指向原始笔记的链接

我们不是在尝试打造或规模化一个专业服务部门。这本质上是为了弄清楚如何以最好的方式把这项技术带进真实世界，然后把所有这些经验模板化，并把它们送出去，要么作为产品，要么作为蓝图。
We’re not trying to build or scale a professional services arm. This is essentially to figure out how to bring this technology into the real world in the best possible way, and then take all those learnings, templatize them, and give them away, either as product or as blueprints.
—— Amandeep Khurana · [13:00]

指向原始笔记的链接

否则，如果它只留在一小撮人手里，AI 就不会在人类的尺度上兑现。它将保持集中在非常小的一群人中，并且仅限于那一小群人所能产生的影响。
Otherwise, if it only remains with a small set of people, AI will not play out at the scale of humanity. It will remain concentrated among a very small group, and limited to the impact that small group can have.
—— Amandeep Khurana · [18:56]

指向原始笔记的链接

但一切都有瓶颈,我认为持续扩展这个东西的主要瓶颈实际上是训练算力。
But everything has a bottleneck, and I think the main bottleneck on continuing to scale this thing is actually training compute.
—— Justin Johnson · [23:04]

指向原始笔记的链接

你比只是写个提示词时更多地把自己从循环中抽离出来。这样你才能真正把自己扩展放大。
So you’re taking yourself out of the loop more than you are with just like a prompt. That way you can actually scale yourself.
—— Tyler Folkman · [05:26]

指向原始笔记的链接

更难的问题在于如何大规模地复刻一家公司的运作方式。
The harder problems are in trying to replicate the way a company works at scale.
—— Sharadh Krishnamurthy · [03:41]

指向原始笔记的链接

如果超过 30 或 40 个工具,我们就无法在不降低上下文质量或系统性能的情况下扩展。
We cannot scale it beyond 30 or 40 tools without degrading the context or degrading the performance of the system.
—— Ajay Prakash · [17:23]

指向原始笔记的链接

瓶颈最终转移到了验证这一侧，因此，我们没有办法大规模地衡量价值，并以同样的速度判断质量。
the bottleneck ends up shifting to the verification side, and thus, we don’t have a way to measure value at scale and to judge quality at the same speed.
—— Maximillian Piras · [12:20]

指向原始笔记的链接

所以这里的要点是，要解决规模化问题，你必须真正地往小处想，想所有尽可能小的事情，想所有可能出错和必将出错的事情，然后提前尝试解决它们。
And so the takeaway here is that to solve the scale problem, you have to actually think small, think all the smallest thing possible, think everything that can go wrong and will go wrong, and then try to solve it before.
—— Vincent Wendy · [15:44]

指向原始笔记的链接

现在，在全球范围内，想象一下类似的事情，只不过庄家是在为某个目的控制你的大脑。
Now, on a global scale, imagine something similar where the house is controlling your brain for a purpose.
—— 嘉宾 · [20:06]

指向原始笔记的链接

规模越大，它们变得越异步，运行时间越长，在真实世界中采取的行动越多，它们创造的价值就越大。
The more scale, the more asynchronous they get, the more longer they run, the more actions they take in real world, the more value they create.
—— Samar Abbas · [03:33]

指向原始笔记的链接

在网络规模下，文本搜索比做向量搜索更困难、计算成本更高。
Text search at web scale is more difficult and more computationally expensive than doing vector search.
—— Simon Eskildsen · [19:06]

指向原始笔记的链接

我不认为它是世界上创办公司的最佳地点,但我认为它是世界上扩展一家公司非常重要的地方。
I don’t think it’s the best place in the world to start a company, but I think it’s a very important place in the world to scale a company.
—— Josh Elman · [53:36]

指向原始笔记的链接

我觉得这两家公司都展示了那一刻有多神奇，当我们实时互动时那种体验有多神奇，但同时也展示了当你只有实时时，扩张有多难。
And I think what both of these companies showed is how magical that moment is, how magical that experience is when we’re having live, but also how hard it is to scale when you’re only live.
—— Josh Elman · [25:45]

指向原始笔记的链接

我们终于有了一个我们知道可以无限扩展的架构。
We finally had an architecture that we knew could scale infinitely.
—— Michael Lee · [02:52]

指向原始笔记的链接

当你拥有能够 24 小时 7 天工作的机器、能够随电力而扩展的机器、能够完成任何个人或人群都无法做到的事情的机器时,正确的答案是思考如何重组一个组织应该是什么样的,以迎接当今技术所带来的机会。
When you have machines that can work 24-7, that can scale with electricity, that can accomplish what no individual human or group of humans can do, the right answer is to think about how do you start to reorganize what an organization should be to meet the opportunity of what’s available today with technology.
—— Michael Lee · [05:40]

指向原始笔记的链接

但值得记住的是，唯一一个价格下降速度比太阳能电力更快的技术，就是电网级规模的电池。
But it’s worth remembering the only technology that’s come down in price faster than solar electricity is utility scale batteries.
—— Al Gore · [07:10]

指向原始笔记的链接

就我个人而言，太阳能革命的广度和深度是可持续转型中最耀眼的明星。
For me personally, the scope and scale of the solar revolution is the breakout star of the sustainability transition.
—— Al Gore · [26:18]

指向原始笔记的链接

进来的 911 报警电话越多、能进来的警报越多,这其实无法扩展。
The more 911 calls that come in, the more alerts that can come in, it doesn’t really scale.
—— Suchet Bargoti · [08:20]

指向原始笔记的链接

我们真的相信,如果我们大规模训练一个视频模型,最后我们会得到一个很棒的模型。
we really believe that if we trained a video model at the large scale, we would get a great model at the end.
—— Anastasis · [08:53]

指向原始笔记的链接

类似地,Gen2 几乎无法创建连贯的视频,但如果你把它扩展,从某种意义上说没有理由它不会成功。
Similar Gen2 can barely create coherent video, but if you scale it up, there is no reason why it shouldn’t work in a way.
—— Anastasis · [26:20]

指向原始笔记的链接

挑战在于,在电商规模下,标准嵌入很快会模糊掉相关性的区分。
The challenge is standard embeddings at e-commerce scale quickly collapse the relevance distinction.
—— Raghav Saboo · [06:23]

指向原始笔记的链接

第一个是,推荐系统的扩展方式和 LLM 一样,而且这个领域在扩展曲线上还处于非常早期的阶段。
The first is that recommendation systems scale just like LLMs do, and that the field is very early in that scaling curve.
—— Devansh Tandon · [00:30]

指向原始笔记的链接

它会改变你对构建底层技术的思考方式，因为为某个规模构建的技术无法在三个数量级的变化中幸存。
It changes how you think about building the tech underneath because no tech built for a certain scale survives three orders of magnitude.
—— Parag Agrawal · [04:18]

指向原始笔记的链接

扩展判断力,而不是扩展模型。
Scale the judgment, not the model.
—— Andrew Orobator · [18:12]

指向原始笔记的链接

我们正在意识到无法规模化扩展的主要事情，就是让人类去点「是」
the main thing that we are realizing does not scale is having humans click yes
—— Liz Fong-Jones · [32:21]

指向原始笔记的链接

他的一个非常有趣的想法是，科学应该由专业人士以工业化的规模来做。
And one of his very interesting ideas was that science should be done at an industrial scale by professionals.
—— Liam Fedus · [08:13]

指向原始笔记的链接

然后你又有另一些工人，同样技艺娴熟、同样有成就、同样叙述可靠地讲述他们的经历，他们说，天哪，你简直不敢相信我们正在制造多少技术债，以及我们制造它的规模。
And then you have other workers, equally skilled, equally accomplished, equally reliable, narrating their experience who say, my God, you would not believe how much tech debt we are producing and the scale that we’re producing it at.
—— Cory Doctorow · [11:44]

指向原始笔记的链接

那就是为什么我们在规模化时反而增长得更快。
And that’s why we’re growing faster at scale.
—— Sean · [12:59]

指向原始笔记的链接

如果你必须依赖一个经理向一个经理汇报再向一个总监汇报,我不确定这些层级能否与 AI 长期共存,因为 AI 可以做其中很多工作。
If you have to rely on a manager reporting to a manager reporting to a director, I’m not sure all those layers scale long-term with AI, because AI can do a lot of that work.
—— Faisal Masud · [22:07]

指向原始笔记的链接

我们作为一家初创公司、一家规模化中的公司，必须赢得信任；在位者已经拥有信任，但这也意味着，如果他们过早发布什么产品，而产品不好用、很糟糕，他们就会毁掉信任。
And like we as a startup, scale-up, we have to earn the trust, incumbents already have the trust, but this also means they can destroy the trust if they ship something like too early and the product maybe doesn’t work or it’s bad
—— Vlad Kyle · [11:46]

指向原始笔记的链接

你要让你的解决方案具备与问题相同的特性。也就是在规模、成本、速度上,你要做到阻抗匹配。
You want your solution to have the same characteristics as the problem. Namely in scale, cost, speed, you wanna have an impedance match.
—— Reid Hoffman · [27:43]

指向原始笔记的链接

对我们来说，TypeScript Go 让构建获得了 10 倍提升——在智能体运行自动化 PR、需要 CI-CD 循环来获取反馈改进代码的规模下，这是一个巨大的提升。
And for us, TypeScript Go was a 10 times improvement in our builds, which at the scale of automatic PRs running with agents and them needing that CI, CD loop for feedback to improve the code is a massive improvement.
—— Harald Kirschner · [07:36]

指向原始笔记的链接

这是智能体的时代、推理的时代。而为此,你并不需要一台 1 吉瓦规模的计算机来服务这些模型。
It’s the era of the agent, the era of inference. And for that, you don’t need a gigawatt scale computer to serve these models.
—— Chase Lochmiller · [21:44]

指向原始笔记的链接

如果你想要在规模化下交付更好的客户体验，你就不能成为良好客户体验的瓶颈。
If you want to deliver a better customer experience at scale, you can’t be the bottleneck to having a good customer experience.
—— Matt Lawler · [04:07]

指向原始笔记的链接

这显然非常清楚地表明人才密度真的很重要。如果在一定的规模下你只有一千人，你只能要最好的人。
that obviously makes very clear that talent density is really important. If you have only a thousand people at a certain scale, you can only have the best ones.
—— Jan Oberhauser · [66:41]

指向原始笔记的链接

它实际上变得更差了。大型实验室下的赌注是，扩展通用智能就会扩展教学能力。不幸的是，情况并非如此。
It’s actually getting worse. The bet that the big labs have made is that scaling general intelligence is going to scale pedagogical capacity. Unfortunately, that has not been the case.
—— 嘉宾 · [45:55]

指向原始笔记的链接

但区别首先在于，AI 能做到的规模让人类望尘莫及，大到人类甚至无法理解的程度。
But the differences are, first and foremost, the scale of what AI can do dwarfs humans, like in ways humans don’t even get.
—— Kevin Mandia · [06:13]

指向原始笔记的链接

但在大规模下，当你有数千台服务器时，一年一两次的故障就变成了每周或每月一次的故障，所以你需要有相应的系统来很好地应对这种情况。
But at a large scale, when you have thousands of servers, a once or twice a year failure becomes a weekly or monthly failure, and so you need to have systems in place to deal with this very well.
—— Ben · [05:41]

指向原始笔记的链接

所以这是那种——底层现实中是非常复杂的基础设施在扩展你的系统——但只要你为它提供合适的接口，比如简单的配置文件，这些能力对任何想要扩展的公司来说都变得容易接触得多。
So this is one of those things, very complex infrastructure in reality under the hood that is scaling your systems, but you provide the right interfaces for it, like simple configuration files, and those things become much more accessible to any company trying to scale.
—— Ben · [13:55]

指向原始笔记的链接

而且结果某种程度上证明，当你真正专注于开发者体验的核心时，你最终得到的东西同时也是非常好的原语，能让 AI 安全、可靠地与数据库协作并帮助它扩展。
And it does kind of turn out that when you really focus on that core of the developer experience, what you actually end up with is also very good primitives for AI being able to work with a database safely and reliably and helping it scale.
—— Ben · [18:12]

指向原始笔记的链接

这是每秒 PB 级的内存带宽。
That’s petabyte per second scale memory bandwidth.
—— Charles Frye · [51:23]

指向原始笔记的链接

我认为肯定是一天多次的规模，也许取决于具体的配置，每小时多次，总会有东西出故障。
I think that it is definitely going to be at the scale multiple times a day, and perhaps depending on the exact configuration, multiple times an hour, something is going to fail.
—— Amin Vahdat · [10:42]

指向原始笔记的链接

实际上客户每天支付的是几美元几美分，但在 Anthropic 的成本端，补贴高达 150 甚至 750，这种经济模型是没法规模化的，即便是 Anthropic 这个体量的公司也不行。
So effectively what the customers were paying daily was dollars and cents, but on the cost side for Anthropic, it was like 150 to even 750 subsidized, which is effectively not an economics that you can scale, even for the company at the size, at the scale of Anthropic.
—— Dor Sasson · [03:15]

指向原始笔记的链接

但这里的主要结论其实是，它最终归结为你所拥有的数据的规模，以及你愿意投入多少整理成本来维护那套合理的检索系统或数据管理系统。
But the major takeaway here is actually it really comes down to the scale of the data that you have and how much curation effort you’re going to put in in order to maintain that retrieval system or that data management system that makes sense, right?
—— George He · [05:18]

指向原始笔记的链接

是的，我的意思是，规模是市场平台的终极护城河，对吧？
Yeah, I mean, scale is the ultimate moat for a marketplace, right?
—— Alyssa Ravasio · [18:58]

指向原始笔记的链接

所以现在我可以在这个小系统的帮助下，规模化地创建同理心
So now I can go ahead and create empathy at scale with the help of this small system
—— Ankit Shukla · [23:53]

指向原始笔记的链接

所以，我可以三个月就把这东西设计出来，但实际上我没法在九个月内把它大规模做成真正的硅片。
So, I can design the thing in three months, but I can’t actually get it into real silicon at scale for nine months.
—— Pat Gelsinger · [10:20]

指向原始笔记的链接

你知道，它们不会全都赢，它们不会全都赢，因为归根结底，你必须在这些东西上获得规模。
And you know, they’re not all going to win and they’re not all going to win because at the end of the day, you have to get scale on these things.
—— Pat Gelsinger · [18:09]

指向原始笔记的链接

我的信念是，如果你想打造一个非常大的生意，你不能只做一件事然后把它规模化。
My belief is, if you want to build a really big business, you can’t just do one thing and scale it.
—— Daniel Saks · [03:46]

指向原始笔记的链接

那就是，说到底，扩展是行得通的。但在机器人学中，扩展以及如何扩展要难得多，因为没有互联网。
which is like, at the end of the day, scaling will work. But in robotics, to scale and how to scale just much harder because there is no internet.
—— Jason Ma · [42:38]

指向原始笔记的链接

② 出现在这些集

5 集

③ 关联

点进去有真内容 —— 本页主要出口

智能体 · Palantir · Meta · YC · NVIDIA · Anthropic · Garrett Lord · Alexandr Wang · Diana · Michael Lee

反向链接