Almaty, Six Months Ahead

Good afternoon to everyone, dear guests, and a big thank you to the organizers for such a cool event.

各位嘉宾，大家下午好，也非常感谢主办方办了这么棒的活动。

Yerzat, thank you for coming, for taking the time.

Yerzat，谢谢你肯来，肯抽出时间。

You work a huge amount, you work through the nights.

你们工作强度非常大，晚上也在干活。

You just had a coffee or something right now, so thank you very much for setting aside the time.

你刚刚是喝了杯咖啡还是什么吧，所以真的非常感谢你腾出这个时间。

Uh, you know what I wanted to start with?

你知道我想从哪儿开始聊吗？

The audience we have here, uh-uh, is not all IT and startups.

我们这里的听众，并不全是做 IT 和创业的。

You and I meet periodically, probably once a month, uh, once every 2 months we make some public statements, some interviews, podcasts, and, well, you're growing so fast that everything changes every month.

我们俩隔一阵子就碰一次，大概一个月一次，每一两个月我们会做些公开发言、接受一些采访、播客，而你们长得太快，每个月一切都在变。

But I'd like to start with you telling the broad audience about Higgsfield.

但我想先请你给在座的广大听众介绍一下 Higgsfield。

Everyone knows that it's Gen- a bit more detail about the business model and about the essence of your business, a bit more detail.

大家都知道这是 Gen——再具体讲讲商业模式，讲讲你们业务的本质，讲得再细一点。

Let's start with that.

我们就从这儿开始。

Mhm.

嗯。

Uh, yes, Yelzhan, thanks for the question.

好的，Yelzhan，谢谢你的问题。

Uh, yes, Higgsfield is a platform that from the moment it was founded has been fully oriented toward, uh, AI video.

Higgsfield 是一个从创立之初就完全聚焦于 AI 视频的平台。

When we started, AI video was in this embryonic state.

我们起步的时候，AI 视频还处在萌芽状态。

And now, probably, everyone here who sits on social media, they constantly get all kinds of videos coming up in Reels, on YouTube and so on.

而现在，在座凡是刷社交媒体的人，在 Reels、YouTube 上应该都会不停刷到各种各样的视频。

Uh, and when we started Higgsfield, we understood that AI video, well, basically the video market — it's probably the biggest market on the internet, because at this point, uh, all direct-to-consumer, B2C sales in the market at all happen through video.

我们开始做 Higgsfield 的时候就明白，AI 视频——其实整个视频市场，大概是互联网上最大的市场，因为到今天，市场上所有 direct-to-consumer、B2C 的销售都是通过视频完成的。

That is, if before businesses had to sell through Google, some kind of optimization, then now it's direct-to-consumer, any busine— any businesses in the world, they sell their services, offerings, products through Reels, through TikToks.

也就是说，以前企业得靠在 Google 上做各种优化来卖东西，而现在是直接面向消费者，全世界任何企业都在通过 Reels、通过 TikTok 卖自己的服务、产品。

And it's obvious that this market, every single year — the video market just keeps growing.

而且很明显，这个市场每一年——视频市场就是在不断增长。

And when we realized that in the near future it would be possible to generate video with neural networks, then, uh, in this field you'd be able to build a very big company for the world economy, one that, well, we hope, will make it into the top 10 biggest companies in the world.

当我们意识到不久的将来可以用神经网络生成视频，那么在这个领域就能为全球经济造出一家非常大的公司，我们希望它能进入全球最大公司的前十。

Uh, and since we started doing this very early, we — we managed to become this kind of household name.

而因为我们动手非常早，所以我们才做成了一个家喻户晓的名字。

And right now we're the biggest platform for video generation.

而现在我们是最大的视频生成平台。

And we — we managed to achieve this in a year.

而这一切我们用一年就做到了。

Right now, for example, colleagues from ByteDance came by, and they shared the numbers that we're the biggest both outside China and inside China.

比如说，最近 ByteDance 的同行来过我们这儿，他们分享了数据：不论是中国以外还是中国境内，我们都是最大的。

And we're also the, uh, biggest among all the American video generation companies.

而且在所有美国的视频生成公司里，我们也是最大的。

And here we got lucky in that we were the very first to appear in this market at all and to start working on it seriously.

这一点上我们的运气在于，我们是最早出现在这个市场、并且最早认真去做它的。

And we managed to take the leadership, but the market, you have to understand, is still in an only just-beginning state.

我们拿下了领先位置，但也得明白，这个市场仍然处在刚刚起步的状态。

And Jensen Huang, the CEO of Nvidia, said that 100% of the pixels you'll be seeing on your phone will be AI generated, and we're rapidly heading toward that, and we want to hold the leadership in it.

Nvidia CEO 黄仁勋说过，你在手机上看到的像素将有 100% 是 AI 生成的，而我们正在飞速逼近那一天，我们也想在这件事上保持领先。

So, that is, our main focus turns out to be video for social media, but also everyone, well, should understand that Hollywood too, and movies, series — all of that can now be generated with AI.

也就是说，我们的主要焦点是社交媒体视频，但大家也该明白，好莱坞、电影、剧集，这些现在也都能用 AI 生成了。

And in this market we've also started, well, we're broadly represented among the AI companies.

在这个市场我们也已经动手了，在众多 AI 公司当中我们的布局很广。

We're pioneers there too.

在这个领域我们同样是先行者。

We star— started generating the first series.

我们开——开始生成第一部剧集。

Right now, uh, for Cannes we're generating the first full-length AI film.

现在我们正在为戛纳生成第一部 AI 长片。

Uh-uh, that is, we're trying to hold the leadership in the AI video market.

也就是说，我们在努力保持 AI 视频市场的领先地位。

Thank you, Yerzat.

谢谢你，Yerzat。

In a moment we'll talk about films, about niches in video, in more detail.

待会儿我们会更详细地聊电影，聊视频里的细分赛道。

So right now it turns out Higgsfield is first — correct me if I'm wrong — Higgsfield is first place in the world.

现在看下来，Higgsfield 是第一——说错了你纠正我——Higgsfield 是全球第一。

uh, the startup people come to if you want to make some video, doesn't matter whether you're B2C, B2B, some complex film, some complex ad, or just some, uh, fun video for your personal TikTok, Instagram — everyone comes to you, right?

就是大家想做视频时会去找的那家创业公司，不管你是 B2C 还是 B2B，想做复杂的影片、复杂的广告，还是只想给自己的 TikTok、Instagram 发个好玩的视频，大家都来找你们，对吧？

And the second one, the competitor that's behind you in this, in this space, uh, it's more than two times smaller.

而排在你们后面的第二名，这个赛道里的那个竞争对手，体量还不到你们的一半。

And I also want to highlight this: in video generation there are a huge number of models, and there, probably, the competition going on is no less fierce, if not fiercer, than in the chat models that we all know, like ChatGPT, Anthropic and so on, I mean in video there are Chinese models, there are American ones, there are Google's ones.

我还想特别点一下：视频生成这边模型特别多，那里的竞争激烈程度恐怕不亚于、甚至超过我们都熟悉的聊天模型，比如 ChatGPT、Anthropic 这些；也就是说视频这边有中国的模型，有美国的模型，还有 Google 的模型。

Well, and basically these are the two countries.

基本上就是这两个国家在做。

Could you talk about the landscape of LLM competition and how you interact with them, what place you occupy there, why everyone wants to distribute through you and why you're pulling it off?

你能不能讲讲 LLM 竞争的格局、你们跟它们怎么打交道、你们在里面占什么位置、为什么所有人都想通过你们做分发，以及你们为什么能做成？

Why is it more advantageous for Seedance to work through Higgsfield than directly?

为什么 Seedance 通过 Higgsfield 走，比自己直接做更划算？

Yeah, well, Seedance is a model from ByteDance, and there's also a geopolitical question there, why, well, a Chinese company can't freely operate on international markets.

是的，Seedance 是 ByteDance 的模型，这里面还有地缘政治的问题——中国公司没法在国际市场上自由地开展业务。

And globally, yeah, I mean training models, training a model from scratch is very expensive.

从全局看，训练模型这件事——从零开始训一个模型，成本非常高。

It takes hundreds of millions of dollars just for the training.

光是训练这一项就要好几亿美元。

And right now only giant corporations can afford this, such as ByteDance, Google, OpenAI, and, uh, Anthropic, which right now is raising at a one trillion dollar valuation on the private markets ahead of its IPO.

现在只有巨型公司才负担得起，比如 ByteDance、Google、OpenAI，还有Anthropic——它正在私募市场上以 $1T 的估值融资，为 IPO 做准备。

And, uh, since they spend giant, uh, prices on training the models, of course they need to make those models pay off.

而既然他们在训练模型上砸了这么巨额的成本，那他们当然得把这些模型的钱赚回来。

And Higgsfield here is the number one distributor of these models.

而 Higgsfield 就是这些模型的头号分发方。

And, uh, since the big corporations have a fierce fight going on among themselves, then, uh, Higgsfield ends up in a very good position, because purely by game theory, uh, we can always negotiate between several big counterparties, for example between, say, Chinese companies, the large ones, uh, and in the GenAI market, well, if people are deeply into it, then they know that constantly, every month, even maybe, like, every few weeks, uh, the models keep taking the lead from each other.

而且，正因为这些大公司彼此之间打得极其凶，Higgsfield 反而处在一个非常有利的位置——单纯从博弈论看，我们随时可以在好几个大交易对手之间谈条件，比如在那几家中国大公司之间；而在 GenAI 市场上，只要是深入了解的人都知道，模型的领先位置每个月、甚至可能每隔几周就要易主一次。

I mean, uh, for example, until recently everyone was talking about Claude Code and there was some kind of unreal hype.

比如说，就在不久前，所有人都在谈 Claude Code，炒作到了一种不真实的程度。

And now Codex from GPT, uh, has again become the number one model for coding.

而现在 GPT 那边的 Codex 又重新成了编程模型里的第一。

And this happens within a single month, I mean the leadership changes.

而这一切就发生在一个月之内——领先位置就是这么换的。

And in video models the fight is even more fierce, because unlike coding models, where there are only American players, in video models there are also Chinese companies, and there are also geopolitical things there.

视频模型这边的争夺更凶，因为不同于只有美国玩家的编程模型，视频模型里还有中国公司，还牵扯到地缘政治的因素。

That's why we, as a distributor, are in an even more advantageous, uh, position, purely by game theory.

所以我们作为分发方，纯从博弈论看，处在更有利的位置。

That's why, uh, well, we manage to, like, always have the best terms on the market.

所以我们基本上总能拿到市场上最好的条件。

But the other part of our business is precisely building distribution there.

但我们业务的另一半，恰恰是把分发渠道建起来。

And, well, off the back of that, that's what we can offer these big companies, and the building of workflows.

靠这个，我们才有东西可以拿去跟这些大公司谈，还有工作流的搭建。

And it's not for nothing that, for example, we're making a feature-length film right now, because we're, uh, roughly 6 months ahead of the entire market.

我们现在做那部长片电影不是没有道理的，因为我们大概领先整个市场 6 个月。

I mean, uh, for us, well, for example, 40% are businesses, uh, among the users.

比如说，我们的用户里有 40% 是企业。

And when we talk, for example, with businesses, we realize that they, since it's enterprise, it lags, like, nine, in general, mon-, by 9 months behind the market.

而我们跟企业客户聊下来会发现，因为是 enterprise，他们比市场落后差不多九个……落后 9 个月。

And some advanced prosumers lag by 6% behind the market, by 6 months behind the market.

而一些比较超前的 prosumer 落后市场 6%……落后市场 6 个月。

And we do a large amount of R&D, a large amount of content and research in video generation, we build workflows and we're, like, at the cutting edge.

我们做大量 R&D、大量内容和视频生成方面的研究，搭建工作流，可以说站在最前沿。

For example, when the big corporations, like Chinese, American ones, make new models, we get access to them already while they're being trained.

比如，中国、美国那些大公司做新模型的时候，我们在他们还在训练的阶段就能拿到访问权限。

I mean, and we already have, like, a whole club, I mean these are all the big names, well, anyone, you can name right now all the companies that are worth 100 plus billion dollars.

也就是说我们这儿已经攒起了一个圈子，全是大名字——你随便点，现在能叫得出名字的 $100B+ 的公司都在里面。

And we work with them directly, I mean literally they train the models, and, uh, they give us access, we give feedback.

我们跟他们是直接合作——就是他们训练模型，把访问权限给我们，我们给反馈。

And for us, as I said, we also lead in Cinema.

而像我刚才说的，我们在 Cinema 这块也是领先的。

I mean we give feedback on how to make a cinematic model, how to do, uh, marketing, UGC, social networks.

也就是说，我们会反馈怎么把模型做得有电影感，怎么做营销、UGC、社交媒体。

And, literally, well, the major players train their models based on our feedback, on, based on the information that we provide them.

可以说，头部玩家就是基于我们的反馈、基于我们提供的信息在训练模型。

And so we've got such a good symbiosis right now >> Right now there's already some, how to put it, pre-news, right, that the IPO market is opening up.

所以我们现在是一种很好的共生关系 >> 现在已经有一些，怎么说呢，前期风声了，说 IPO 市场正在打开。

And among them there's this company Canva for image generation, which is now expecting to go to market already, right, I mean to IPO at a valuation of 50 to 60 million billion dollars.

其中就有 Canva 这家做图片生成的公司，现在准备上市了，也就是要 IPO，估值 $50–60MB。

Tell me, please, can Higgsfield repeat Canva's path, but in video generation?

那你说说，Higgsfield 有没有可能复制 Canva 那条路，只不过是在视频生成这边？

Yes, yes, of course.

对，当然可以。

And Canva's last public round on the private markets was a 40 billion dollar valuation.

Canva 上一轮公开披露的私募市场融资，估值是 $40B。

And since then Canva's metrics have all grown.

而从那以后，Canva 的各项指标都涨上去了。

Uhh, this — actually, we do have a clear vision there of how to get to a $40 billion valuation, but naturally we're aiming for 100x growth.

其实我们有很清晰的构想，知道怎么走到 $40B 估值，但我们自然是奔着 100x 增长去的。

That is, globally that's what we've committed to as a company.

从全局上讲，这就是我们作为一家公司许下的承诺。

Because unicorns in the world, you have to understand, but in America there are 1,000-plus unicorns there.

因为独角兽——你得明白，光美国就有 1000 多家独角兽。

And just being a unicorn actually isn't all that cool.

所以单纯当一家独角兽，其实没那么了不起。

And we want to become — well, obviously, first we need to pass a milestone, that's decacorn, and then become a $100-plus billion company.

而我们想成为的——当然，先得跨过一个里程碑，也就是十角兽，然后再成为 $100B 以上的公司。

That's what we're — and this is how we look at it, so as to get there from zero to unicorn.

这就是我们——我们是这么看这件事的：要从零走到独角兽，

You need there something on the order of, well, a very strong team, on the order of 15 outright superstars there, killers, the very best ML engineers, product designers, uh, B2B people.

需要一支非常强的团队，大概 15 个真正的超级明星、狠角色，最顶尖的 ML 工程师、产品设计师，还有做 B2B 的人。

And right now we're aimed at scaling, that is, we want to get to 100x, meaning this year do a 10x on the valuation and next year another 10x, because, yeah, in twenty-three Anthropic was worth $4 billion.

现在我们瞄准的就是规模化，也就是要做到 100x：今年估值翻 10x，明年再翻 10x，因为 Anthropic 在 23 年的估值是 $4B。

And now, in twenty-nine, there the second-to-last round was already $400 billion.

而到了 29 年，那边倒数第二轮融资就已经是 $400B 了。

And we roughly understand that in twenty-nine we, like Anthropic, should also be worth $400 billion.

我们大致的判断是，到 29 年，我们也应该像 Anthropic 一样值 $400B。

Uhh, well, we're growing 30% month over month, that is, by the end of the year we should definitely be over a billion, uh, well, get there much earlier, but even if we slow down, we should still get to a billion in ARR.

我们现在每月增长 30%，也就是说到年底肯定能超过 $1B，实际上会比这早得多，但就算增速放缓，我们也一定能做到 $1B ARR。

And that already, at the current multiples of AI companies, means we can already get close to Canva's level in valuation there.

而按现在 AI 公司的估值倍数，我们的估值就已经能接近 Canva 的水平了。

Yerzat, these are some kind of insane numbers.

Yerzat，这些数字简直疯狂。

Look, Anthropic right now is trading at $1.3 trillion on the secondary market.

你看，Anthropic 现在在二级市场上的成交价是 $1.3T。

And there's none of it to be had.

而且根本抢不到。

Everyone wants to buy it.

所有人都想买。

There's none to be had.

但就是没有货。

Yes.

是的。

And you just said 100x from now, right, from your current valuation.

你刚才说的 100x，是从现在算起，对吧，从你们现在的估值算起。

Can you say what your projected valuation is right now, what your round is, at what valuation, like, what the scale of the numbers of this round is, uhh, so we understand what the 100x is off of — off of a billion, off of — here again, yeah, it's good that our CFO isn't here, because, yeah, she really tells us off when we go around telling everyone our numbers.

你能不能说一下，你们现在预期的估值是多少、在融哪一轮、按什么估值、这一轮的数字大概是什么量级，好让人明白这个 100x 到底是从什么基数算起——是从 $1B 起，还是——说到这儿，幸好我们的 CFO 不在场，因为我们到处讲数字，她每次都要把我们狠狠训一顿。

Well, there is private market data, the fact that the secondary market in companies globally is very liquid, and Higgsfield is already trading on the secondary market there.

私募市场是有数据的：全球公司的二级市场流动性非常好，Higgsfield 在二级市场上已经有人在交易了。

Well, I know there were, well, liquid deals at a $4 billion valuation.

我知道确实有过按 $4B 估值成交的流动性交易。

There you go.

就是这样。

And obviously we can't go below that there, since the valuation on the secondary market is already that.

那显然我们不可能再往下走，因为二级市场上已经是这个估值了。

And basically, in my view, it's even understated, because after all we — well, for example, ElevenLabs, a company that has the same ARR as ours, they closed a round at an $11 billion valuation.

而在我看来，这个价其实还偏低，毕竟比如说 ElevenLabs，一家 ARR 和我们差不多的公司，他们这一轮的估值是 $11B。

And that's a company older than us, they were growing much more slowly than we are.

而那是一家比我们老得多的公司，增长速度也比我们慢得多。

So, roughly, yeah, those are the numbers.

大概就是这样的数字。

And, well, we are, yeah, raising a Series B, and the round is plus-minus, yeah, at a valuation that is fair on the secondary market already right now.

我们确实在融 Series B，这一轮的估值大致就是二级市场现在给出的公允价。

So let's, say, state the ratio.

那我们就把这个 比例 说出来吧。

Well, and to reach valuations like that, as I understand it, Higgsfield needs both new markets and new products.

那要拿到这样的估值，我理解 Higgsfield 既需要新市场，也需要新产品。

Can you tell us something about that?

这方面你能讲讲吗？

Yes, well, we — that is, you have to understand that the AI market, it's very, uh, volatile.

是的，我们——首先得明白，AI 市场是非常动荡的。

Everyone, well, understands that one moment OpenAI is at its peak there, and everyone is afraid of OpenAI, they think there'll be a singularity, then Gemini from Google becomes the leader, then Anthropic just shows up out of nowhere, and, well, they ran up to $40 billion in ARR in 4 months.

大家都明白：一会儿 OpenAI 站在巅峰，所有人都怕 OpenAI，觉得奇点要来了；然后 Google 的 Gemini 又成了领跑者；接着 Anthropic 凭空杀出来，4 个月就冲到了 $40B ARR。

I mean, that's, well, just phenomenal growth that happened to them.

他们这个增长真的是现象级的。

That is, the market is very volatile.

也就是说，这个市场波动极大。

And since the market is very volatile, the classic strategy, well, in a volatile market, like, I don't know, for some commodity traders, the classic approach is always an offensive approach.

而既然市场波动这么大，在波动的市场里，经典策略——比如大宗商品交易员那一套——经典打法永远是进攻。

And we're constantly trying to do moonshots, constantly looking for new products.

所以我们一直在做 moonshot（登月式豪赌），不停地找新产品。

We're always in offensive mode in the market, because, well, it's obviously the best strategy in a volatile market.

我们在市场上一直保持进攻姿态，因为在一个波动的市场里，这显然是最好的策略。

And at the same time, even if we don't make new products, it's precisely the core revenue that's growing for us right now — that's prosumers.

而且就算我们不再做新产品，眼下真正在增长的正是那块核心收入——也就是 prosumer 用户。

As I said earlier, 40% is businesses, and Higgsfield's users, they already do 30% overs for us.

像我前面说的，40% 是企业客户，而 Higgsfield 的个人用户已经给我们贡献了 30%overs。

And if this even, well, turns into 10% month-over-month, then we should do a billion by the end of the year.

而哪怕这个数字掉到 10%月环比，我们年底也应该能做到 $1B。

But since we understand that, first of all, the market is volatile, everything happens very fast, and we look at that as an even greater number of opportunities.

但我们清楚，第一，市场是波动的，一切都发生得非常快，而我们把这看成是更多的机会。

That's why we're constantly thinking about new products.

所以我们一直在琢磨新产品。

And right now we're preparing a new product.

现在我们就在准备一个新产品。

It will be, uh, launched, I think, within 2 weeks.

我想它会在两周之内上线。

It's called Higgsfield Supercomputer.

它叫 Higgsfield Supercomputer。

And this is a product — maybe, whoever uses, for example, Claude Code or, say, OpenClaw, they understand that, well, the people who use this are already living in the future, they understand that agentic systems, that is, when agents autonomously do all the work, and when there's no longer that layer of people interacting with each other, and everything moves as much as possible to autonomous agents, and there a phenomenal growth in productivity happens.

这个产品——用过比如 Claude Code 或者 OpenClaw 的人大概能懂，用这些东西的人已经活在未来了，他们明白 agentic 系统，也就是让智能体自主完成全部工作，当人与人之间来回协作的那一层不再需要，一切尽可能交给自主智能体的时候，生产力就会出现惊人的跃升。

—happens, because foundational models have gotten much better, and there have been breakthroughs in working with memory, with context, and self-evolving systems have appeared, where they can evolve and learn on their own, and we can't ignore that.

——之所以会这样，是因为基础模型变强了太多，在记忆和上下文的处理上出现了突破，还出现了自我进化（self-evolving）的系统，它们能够自我演化、自我学习，这一点我们没法忽视。

That's why we're preparing a new product.

所以我们在准备一个新产品。

Higgsfield Supercomputer is something similar to OpenClaw or Claude Code, only the main differentiator is that it's focused on media generation.

Higgsfield Supercomputer 有点像 OpenClaw 或者 Claude Code，只不过最主要的差异化在于它专注于媒体生成。

And, for example, uh, well, Claude, for example, there are no media models there at all — in Claude there's no image and video generation.

拿 Claude 举例，它里面根本没有媒体模型，Claude 里没有图片和视频生成。

And literally this week we released an integration with Claude.

而就在这一周，我们上线了跟 Claude 的集成。

And at the moment we've simply had exponential growth there.

眼下我们在那边直接就是指数级增长。

And we're the biggest MCP for Claude, for generating media.

我们是 Claude 上最大的媒体生成 MCP。

That is, if you follow along on Twitter, then there, well, it trended on Twitter for 3 days that Higgsfield did an integration with Claude.

也就是说，你要是刷 Twitter 就会看到，Higgsfield 和 Claude 做了集成这件事在 Twitter 上热了 3 天。

There.

就这样。

And next, well, the logical step for us is to make a full-blown product.

接下来，对我们来说合乎逻辑的一步，就是干脆做一个完整的产品。

Translated into plain Russian, this means generating an ad video clip in one — >> Yes, yes, I wanted to interrupt you a little, to translate, yes.

翻译成大白话，就是一条指令生成一支广告片——对对，我想稍微打断你一下，翻译一下。

That is, right now frameworks like Claude Code, OpenClaw, Hermes, and other technologies there can do everything a person could do on a computer.

就是说，现在像 Claude Code、OpenClaw、Hermes 这些框架，还有别的一些技术，已经能做人在电脑上能做的一切事。

It connects applications that previously couldn't be connected into a single workflow and a specific task.

它能把以前互不相通的应用串进同一条工作流、同一个具体任务里。

That is, and this week MCP came out, MCP for Claude.

而就在这周，MCP 出来了，给 Claude 的 MCP。

That means that MCP is a technology with which, uh, different applications can talk to each other.

MCP 是这样一种技术：让不同的应用彼此对话。

That is, sitting inside Claude, their big audience can say in a single prompt: "Install MCP for Higgsfield".

也就是说，坐在 Claude 里，他们庞大的用户群只要一句提示词就能说：“Install MCP for Higgsfield”。

He installs it, and there he drops in his photo and says: "Create me a full ad campaign for this".

他装好，把自己的照片往里一丢，然后说：“给我做一整套完整的广告活动”。

Claude goes to Higgsfield by itself, picks by itself which model is needed, which one for Instagram, which one for TikTok, which one for cinematic video.

Claude 自己去 Higgsfield，自己挑该用哪个模型，哪个给 Instagram，哪个给 TikTok，哪个做电影感的视频。

It creates all of this and hands it to him.

全部生成好，交到他手上。

That is, he doesn't need to go into Higgsfield and use the interface.

也就是说，他根本不用进 Higgsfield、去用那个界面。

That is, agents now communicate over MCP without an interface and do the work turnkey.

也就是说，智能体现在通过 MCP 在没有界面的情况下互相沟通，把活儿一站式做完。

That's, in plain language, what Yerzat said.

这就是我用大白话解释 Yerzat 刚才说的东西。

Ah, yes.

啊，对。

Why is this important?

为什么这件事重要？

Because, well, any direct-to-consumer business whose, well, main sales channel is through Reels, through TikToks — for such businesses it's important to produce video in gigantic quantities.

因为任何一家 DTC 公司，主要销售渠道就是 Reels、就是 TikTok，这类公司必须海量地产出视频。

That is, for example, they have a new product coming out, like, I don't know, perfume.

比如说，他们有个新品要上，我不知道，就说香水吧。

And they need to generate not one ad, but they need to generate several thousand variations, automatically upload it into the ad network, measure which ad performs better, and then scale it up further.

他们要生成的不是一条广告，而是好几千个版本，自动投进广告网络，测出哪条广告跑得更好，再把它放大。

So this is, like, the state of the art of the market right now.

所以这就是现在市场的最高水准。

All the to-consumer businesses in the world, uh, the biggest ones, uh, the billion-dollar companies and the small companies, they all compete on this.

全世界所有面向消费者的业务，最大的那些，十亿美元级的公司和小公司，全都在这一点上竞争。

That's why it's very important for all of them, uh, to increase the productivity, uh, of their performance creatives.

所以对他们所有人来说，提升自己效果创意素材的产出效率非常重要。

And here we released the MCP, we, I mean we released it before, before the supercomputer, and we, there was, for example, no marketing budget when we released it.

我们就是在这个节点上发布了 MCP，而且是赶在超级计算机之前发布的，发布的时候连市场预算都没有。

It just insanely blew up the internet all on its own, to the point that it became the most popular MCP right now.

它完全是靠自己把互联网炸开了锅，一路火到成了现在最受欢迎的 MCP。

Yeah, you know, I also wanted to talk about this.

对，我还正想聊聊这个。

Right now the industry leaders are saying that AI, uh, well, is cutting workers, and many say that in the future there will be a profession like agent orchestrator or model orchestrator, right?

现在业界领袖都在说，AI 正在削减岗位，很多人说未来会出现一种职业，叫智能体编排师，或者模型编排师，对吧？

Say, to solve any task, some complex app, there might be close to a dozen different models used for different tasks.

比如说，要解决任何一个任务，做一个复杂的应用，可能会用上十来个不同的模型来分管不同的活儿。

Like, image — one model is good; on the internet — another model is good.

图像这块，某个模型强；上网这块，另一个模型强。

Third, maybe thinking, right, Opus, the latest model, is great at doing that.

第三块可能是推理，最新的那个 Opus 模型做这个就特别棒。

And the people who put all these models together to get a great harness and a great, um, it's a whole, how do you say it, that's exactly where the State of the Art is right now, the art of startups.

而把这些模型全都拼装起来、做出一个出色的 harness 的人——这本身就是一整门学问，这才是当下真正的最高水准，是创业公司的艺术。

And your platform turns out to be an operating system that itself picks the right model for the task at hand and, well, sort of takes on all this — not so much the headache, uh — all the >> technical knowledge of where to use which model, where which one is more cost-effective, where to save.

这么看，你们的平台其实就是一个操作系统，它自己就能针对任务挑出合适的模型，把这些——与其说是麻烦事，不如说是所有 >> 技术上的门道：哪里该用哪个模型、哪个更划算、哪里能省钱——全都替你扛了。

Uh, so can you talk about why your product is turning into an operating system?

那你能讲讲，为什么你们的产品正在变成一个操作系统吗？

Why is it more cost-effective to do it with you than for me to orchestrate it myself across a dozen different models?

为什么在你们这儿做，会比我自己去编排十来个不同的模型更划算？

Yeah.

对。

Yeah.

对。

Well, well, everyone needs performing videos, whether it's, uh, an ad, or whether it's final content for consumption, for example some kind of mini-series or a film, then it has to be very high quality.

所有人都需要能出效果的视频，不管是广告，还是给人看的最终成品内容，比如某种迷你剧或者电影，那质量就必须非常高。

And quality is a huge deal in both cases.

这两种情况下，质量都是决定性的。

And generating video specifically as a final product is very hard.

而要把视频直接做成最终成品，是非常难的。

And at the moment, uh, you need almost, well, not dozens of models, but at minimum, for example, to generate a good ad here, you need to use seven or eight models in combination.

就目前来说，倒也不用几十个模型，但至少——比如要在这儿生成一支好广告——得把七八个模型串起来用。

And on top of that you need very deep expertise in these models.

而且还得对这些模型有非常深的理解。

To have.

得有才行。

Uh, and, well, the point of Higgsfield is that we're roughly 6 months ahead of the whole market in understanding the models, how to orchestrate them.

而 Higgsfield 的意义就在于，我们在理解这些模型、以及怎么编排它们上，大约领先整个市场 6 个月。

Uh, well, and that's how we hold the lead.

靠这个，我们守住了领先位置。

Uh, well, we just talk a lot with B2B, and we understand that, or, for example, Hollywood people come to us, we talk with the absolutely most legendary people from Hollywood, the ones who are Oscar-nominated and so on, and they come to us and say: “We want to make a feature film, we want to make series.”

我们跟 B2B 客户聊得非常多，也因此了解到——比如好莱坞的人会主动找上门，我们接触的是好莱坞最传奇的那批人，拿过奥斯卡提名的那种，他们来跟我们说：“我们想做长片，我们想做剧集。”

But talking with them, we realize that before they get to a feature film they still need about a year to, well, immerse themselves in AI video, well, in this subject, whereas Higgsfield gives them ready-made tools right away.

但聊下来我们发现，他们离拍出一部长片还差着大概一年，得先扎进 AI 视频这个领域，而 Higgsfield 直接就把现成的工具递到他们手里。

I mean, there are stitched-together agents, and if you need an ad on the level of the World Cup or the Super Bowl, then Higgsfield has agents that can make an ad on the level of the World Cup, the Super Bowl.

也就是说，我们有拼装好的智能体，你要是需要世界杯或者超级碗级别的广告，Higgsfield 就有能做出世界杯、超级碗级别广告的智能体。

If you need to make original content on the level of Netflix, Higgsfield has agents that can make content on the level of Netflix.

你要是想做 Netflix 级别的原创内容，Higgsfield 也有能做出 Netflix 级别内容的智能体。

And we've accumulated that kind of expertise.

这样的专业积累，我们已经攒下来了。

There were a lot of factors why, but in the end we ended up with it.

背后有很多因素，但最终我们确实拥有了它。

And I see, so now we're putting this into an agent system >> By the way, I was recently talking with Darmen, the lawyer.

明白了，所以我们现在正把这些放进智能体系统里 >> 对了，我最近跟达尔门聊过，就是那位律师。

He's the GP of a fund that makes films, and he said he'd struck a deal with Higgsfield.

他是一家做电影的基金的 GP，他说他已经跟 Higgsfield 谈成了。

And on the basis of Higgsfield's technology they're already going to release a full-length film with Kazakh >> So you've already made your own Higgsfield Originals, your uh series under the Higgsfield brand, from start to finish with AI.

他们要基于 Higgsfield 的技术推出一部长片，带哈萨克 >> 也就是说，你们已经做出了自己的 Higgsfield Originals，Higgsfield 品牌下的剧集，从头到尾都是 AI 做的。

And if, well, if the film comes out, it'll be the first fully AI film in cinemas.

如果这部电影真上映了，那就会是第一部完全由 AI 制作、进院线的电影。

I mean, right now some films are coming out in cinemas, that is, some part of them is made with AI, but still most of it is shot in the classic format.

现在院线里也有一些电影上映，其中有一部分是用 AI 做的，但大头还是按传统方式拍的。

And uh when do you expect this, and will you be the first ones to do an end-to-end film that will come out in cinemas, right there offline?

那你们预计这会在什么时候发生，你们会是第一个把端到端的 AI 电影真正搬进线下影院的吗？

Yes, of course.

是的，当然。

So on May 16 at Cannes, uh, we're presenting the first AI film, fully generated.

5 月 16 日，我们在戛纳发布第一部完全由 AI 生成的电影。

And fully generated, you have to understand, in less than 3 weeks, somewhere around plus-minus 2 and a half weeks there, uh, a full-length AI film.

而且是完全生成出来的，你得明白，用了不到 3 周，前后大概两周半，一部 AI 长片。

And globally, if we talk about original content, then yes, you have to understand that there's Netflix, which is also worth 400 plus billion dollars, and they spend 18 billion dollars a year on content.

从全球看，说到原创内容，是的，你得明白，有 Netflix 在，它市值也有 $400B 多，而他们一年要花 $18B 在内容上。

That is, that's Netflix's budget for content.

也就是说，这是 Netflix 的内容预算。

And even if you strip away half of it for some inefficiencies, which, well, Netflix is a big corporation, then 9 billion is the minimum budget that Netflix spends on content production alone.

就算砍掉一半算在各种低效上——毕竟 Netflix 是家大公司——那 $9B 就是 Netflix 光花在内容制作上的最低预算。

At the same time Netflix makes 600 originals a year.

与此同时，Netflix 一年出 600 部原创。

And since we've now started making series and films, we've calculated the whole economics of how to do this very well.

而因为我们现在开始做剧集和电影了，我们把这件事怎么做的整个经济账算得非常清楚。

And right now, to make 1 minute of an AI film, you need to generate 215 minutes.

现在要做出 1 分钟的 AI 电影，得生成 215 分钟。

And as of today, uh, you also need around 10 people.

而到今天为止，还需要大约 10 个人。

So in terms of costs one film comes out to, uh, 300,000 credits for GPU, for the models, and 100,000 is give or take salaries, sound, and the figure comes out to around 400, well, a maximum of 500,000 dollars right now to make a feature-length film using AI.

也就是说算成本，一部电影下来，GPU 和模型要 300,000 credits，$100K 上下是工资、声音，最后数字大概是 $400K，最多 $500K，这就是现在用 AI 做一部长片的价。

And at the same time, for Netflix to shoot comparable content costs around 30 million dollars.

而与此同时，同样的内容让 Netflix 来拍，大概要花 $30M。

Even if the content performs the same, the ROI of an AI film will be simply 30 times, 30-40 times better.

就算内容表现一模一样，AI 电影的 ROI 也会直接好上 30 倍、30 到 40 倍。

And for Higgsfield, for example, to reach the valuation, oops, the rate of 600 originals a year like Netflix has, well, you can take today's cost of 500,000 dollars and multiply it by 600 - that's 300 million dollars of budget to push Netflix aside on the size of the content library.

比如说 Higgsfield 要达到像 Netflix 那样一年 600 部原创的估——哦，那个产出速率，那就拿今天 $500K 的成本乘以 600，就是 $300M 的预算，就能在内容库规模上把 Netflix 挤下去。

But model quality is growing, meaning for one minute you'll need to generate significantly less AI content.

但模型质量在往上走，也就是说做一分钟需要生成的 AI 内容会少很多。

And prices, uh, they, you need fewer people, so the price is falling.

而价格这块，需要的人更少了，所以价格在往下掉。

And by the end of the year, I think, what we're forecasting in our financial model, in order to compete with Netflix, to not fall behind on content, you need to spend roughly 100 million dollars a year.

到年底，我想，按我们财务模型的预测，要跟 Netflix 竞争、在内容上不掉队，一年大概要花 $100M。

That is, uh, Netflix spends 18 billion dollars a year on creating content.

也就是说，Netflix 一年花 $18B 来做内容。

So here there's an obvious gigantic financial, well, well, opportunity, uh, an opportunity that, well, we don't want to miss either.

所以这里摆着一个显而易见的巨大财务机会，而这个机会我们也不想错过。

And so, that is, there's this gigantic opportunity.

所以说，就是存在这么一个巨大的机会。

And since we're now the platform that, that everyone in the world uses to make this content, it's very important for us to be leaders in creating content, because we turn all of this, for those who need it in gigantic volumes, into an agentic workflow, so that it can be used like a content factory.

而既然我们现在是全世界都在用来做这类内容的平台，那我们在内容创作上当领头羊就特别重要，因为我们把这一整套都变成了 agentic workflow，给那些需要海量产出的人用，让它可以当成一个内容工厂来跑。

For those who want to, for example, make films, it's more of a prosumer interface, where teams of 10-15 people can sit and work on the work, where there are roles, there's a director, there's an editor, sound guys, and on Higgsfield they can collaboratively make a film in a short time and there, competitive markets.

对于那些想拍电影的人，那就是更偏 prosumer 的界面，10 到 15 人的团队可以坐下来一起干活，里面有分工，有导演，有剪辑，有录音师，他们可以在 Higgsfield 上协作，用很短的时间做出一部电影，面对的就是有竞争的市场。

Obviously, AI, AI video has already heavily disrupted the advertising market.

很明显，现在 AI 视频已经把广告市场狠狠颠覆了一遍。

a large amount of advertising is already made by production houses that have moved over from real production into AI, and they generate it on Higgsfield.

大量广告已经由那些从实拍制作转到 AI 的制作公司在做，而且他们就是在 Higgsfield 上生成的。

But over the course of this year what will happen is that the really big production houses, the ones that make actual films, long content, series, they too will, well, move into AI.

但今年之内就会发生：那些真正的大型制作公司，就是做电影、长内容、剧集的，他们也会转到 AI 上来。

That's simply economically inevitable.

这在经济上根本就是无法避免的。

Thank you.

谢谢。

You know, the next topic I wanted you, Murat, to open up.

下一个话题，我想请 Murat 您来展开讲讲。

So Higgsfield, uh, you already have close to 200 people working, right.

Higgsfield 这边，你们已经有将近 200 人在做事了，对吧。

around 200 or 150 >> already 200 plus >> 200 plus people and these are exclusively our fellow citizens, young people, the average age was 24 when we last >> already 25 >> already 25, I wanted, we're glad, for you to open up this to-to-to-to-to-to this topic, we work a lot with young people, we listen to a lot of young startups, so that, well, sort of, they'd hear it firsthand, how young Kazakh-Kazakhstanis compete with OpenAI, well, they don't compete anymore, right, with Sora - Sora shut down, with ByteDance, with these trillion-dollar companies.

200 左右还是 150 >> 已经 200 多了 >> 200 多人，而且清一色是我们自己的同胞、年轻人，上次我们统计的时候平均年龄是 24 >> 已经 25 了 >> 已经 25 了，我想，我们也很高兴，请您把这个话题展开讲讲，我们跟年轻人打交道很多，也听过很多年轻创业公司，就是想让大家从当事人嘴里听到，年轻的哈萨克斯坦人是怎么跟 OpenAI 竞争的——不过现在已经算不上竞争了，对吧，Sora 那边关掉了，还有跟 ByteDance，跟这些万亿美元级的公司。

So let's open up this topic.

就这个，咱们把这个话题展开讲讲。

We call this the Higgsfield effect, right, the effect on young people, right.

我们把这个叫做 Higgsfield 效应，对，对年轻人产生的效应。

And today Higgsfield, the whole team, right, they've become real heroes, heroes of the city, heroes of the country, right, and they motivate the guys so much, right, that even here in Kazakhstan, without leaving here, you can become an extra world-class specialist, right, meaning earn some insane money.

今天 Higgsfield 整个团队，他们真的成了英雄，城市的英雄，国家的英雄，而且他们对年轻人的激励大到——就在我们哈萨克斯坦，人不用出去，也能成为超一流的世界级专家，也就是能挣到那种不正常的钱。

There.

就是这样。

And before that, when the talk was that Higgsfield would be based in Almaty, at first people said that here you wouldn't even find five, uh, well, uh, employees whose level would match the tasks being set.

在这之前，当时说 Higgsfield 要落在阿拉木图，一开始还有人说，在这儿连五个人都招不到——水平能匹配上那些任务的员工。

Today more than 200 of them are already working, and I think in a year it'll probably already be 400.

今天已经有 200 多人在这儿工作了，我想，大概一年之后就会有 400 人。

There.

就是这样。

So this opens up the country as a country of talent thanks, first and foremost, to our education, probably.

也就是说，这让这个国家作为一个人才之国被看见，而这首先大概要归功于我们的教育。

We were just discussing this with Zhumabek, right, and the system — KTL, RShO, MShA, Nazarbayev Intellectual Schools, our universities, right, these talents, these guys, this whole system of sifting them out, right, I mean it's bearing exactly this kind of fruit. There.

我们刚才正好跟Zhumabek聊到这个，还有这套体系——KTL、RShO、MShA、Nazarbayev Intellectual Schools、我们的大学，这些人才，这些年轻人，这一整套层层筛选他们的体系，结出的正是这样的果实。

And, well, I don't know, for me this is just so exciting, so inspiring, honestly, yeah, just — >> thank you.

而且说实话，这件事对我来说真的太让人兴奋、太鼓舞人了。>> 谢谢。

Yerzat, please open up about your culture.

Yerzat，请你展开讲讲你们的文化。

Everybody knows that you guys work a whole lot.

所有人都知道你们工作强度非常大。

Explain why you — I mean, right now everybody works a lot, not just you, any AI comp—, any AI startup.

解释一下为什么你们——现在其实所有人都在拼命工作，不只是你们，任何一家 AI 公司、任何一家 AI 创业公司都是。

The only method to win is to outwork everyone.

唯一能赢的办法就是比别人干得更多。

And also about the philosophy — that everything changes so fast and you have to be ready to experiment, to pivot somewhere.

还有关于理念的部分——一切变化得太快，你必须随时准备好去做实验、随时准备好往某个方向 pivot。

With you it's like — bang, an idea for a film comes along, a week later you've made the film, tomorrow — bang, an individual film, right, one that comes for just a single person.

在你们那儿是这样的——啪，一个电影的点子冒出来，一周后电影就做出来了；明天啪，又是一部个性化电影，只为一个人而生。

Maybe you can do that — about this culture, that — >> yes, well, big picture, when we were starting Higgsfield, it was personally very important for me to build this story in Kazakhstan.

要不就从你们这个文化讲起，就是——>> 好，从大处说，我们创办 Higgsfield 的时候，对我个人而言，把这件事放在哈萨克斯坦做非常重要。

Uh, because, well, I have friends, other entrepreneurs, startup guys, who went off to America to do startups back when we were all at the very beginning.

因为我有一些朋友、一些其他创业者、做 startup 的人，在我们大家都还处在最初阶段时，就跑去美国做创业公司了。

We had constant debates.

我们之间一直争论不休。

I always said it's a mistake, because in America the strong talent is already working at OpenAI, at Anthropic.

我一直说这是个错误，因为在美国，最强的人才已经在 OpenAI、在 Anthropic 工作了。

And, well, A-players will be very hard to hire, to put together.

所以要招到、要凑齐 A-player 会非常难。

Uh, and meanwhile in Kazakhstan there definitely is, well, underrated talent, uh, that you just, well, need to gather in one place and it should work out.

而与此同时，哈萨克斯坦确实存在被低估的人才，只要把他们聚到一个地方，就应该能成。

Well, and in the end it did work out.

结果确实成了。

And then, watching it — if we speak more philosophically, I guess — when I just observe from the outside how the team works, how it coordinates itself, uh, I started thinking about how there are different, like, different nationalities, different countries have different approaches and cultures when it comes to business.

后来我这么观察下来——如果说得再哲学一点——当我只是从旁边看着团队怎么运转、怎么自我协调时，我开始思考：不同民族、不同国家对做生意有不同的方式和文化。

For example, everybody knows about the Japanese business culture — I mean, uh, everything there runs like clockwork, like a factory.

比如说，大家都知道日本的商业文化——那边一切都像钟表一样精准运转，像一座工厂。

Uh, super-organization, a gigantic one, of a gigantic number of people, and everyone works like a clock, like one single mechanism.

超级组织，规模巨大，管着庞大数量的人，所有人都像钟表一样运转，像同一台机器。

We know the Chinese business system, which is also similar to the Japanese one, but there are slightly different distinctions there, because they also win more through volume and sheer quantity.

我们也知道中国的商业体系，它跟日本类似，但有一些细微差别，因为他们还更多是靠体量和数量取胜。

Uh, and we know the Anglo-Saxon business model, which is the most, probably the most aggressive business model, the most — uh, where, uh, you close very complex deals, uh, you take high risks, well, the classic business model.

我们还知道盎格鲁-撒克逊式的商业模式，它大概是最激进的一种商业模式——去做非常复杂的交易，去承担高风险，就是那种经典的商业模式。

And it seems to me that right now we're seeing, like, a Kazakh business culture, and it's probably similar.

而我觉得我们现在看到的是一种哈萨克式的商业文化，它大概也有相似之处。

And obviously, I mean, the Anglo-Saxons, they conquered the world through sea expeditions, and the Japanese, uh, through their culture.

很明显，盎格鲁-撒克逊人是通过海上远征征服世界的，日本人则是通过他们自己的文化。

And it seems to me that the Kazakh business model, which I just see, well, I watch how 200 people work in that kind of offensive mode — that probably our nomadic heritage plays a part here, in the sense that, uh, fast coordination, fast, uh, very fast — I see how teams, without much management, when some external threat happens in the market, when some event happens in the market and you have to react fast, I see how our team internally self-organizes very quickly.

而我看到的这种哈萨克商业模式——我就是在旁边看着 200 个人以那种进攻模式工作——我觉得大概是我们的游牧传统在这里起了作用：快速协调，很快，非常快——我看到各个小组在几乎没有管理介入的情况下，一旦市场上出现外部威胁、出现什么突发事件、需要快速反应时，我们内部的团队会非常快地自组织起来。

I mean, there's literally none of that where I have to, well, explain to people that a change has happened, that we need to get together, that we need to build some kind of process.

根本不存在那种我还得去跟大家解释“情况变了，我们得聚一下，得搭个流程”的事。

The self-organization happens even faster than that.

自组织发生得比这还要快。

The fact that people coordinate themselves into these, like, squads, of 10, of 20 people, attack the problem, solve it fast, capture the market.

就是人们会自己协调成一个个小分队，10 人、20 人一组，扑上去攻这个问题，快速解决，抢占市场。

And here, it seems to me, this is what's probably built into our cultural code and into our DNA — this fast self-organization, fast reaction, some smart seizures of the market.

我觉得这大概就是刻在我们文化基因和 DNA 里的东西——快速自组织、快速反应，还有相当有章法的市场攻占。

This, well, I'd like this kind of Kazakh business culture to be cultivated.

我希望这样一种哈萨克商业文化能够被培育起来。

And going forward, uh, I really hope that the people who come out of Higgsfield will go on to build new companies, new startups.

接下来，我非常希望从 Higgsfield 出去的人能继续去做新的公司、新的创业项目。

And this playbook will keep being cultivated, keep working, and, well, a large number of successful companies out of Kazakhstan on the global market. >> Thank you, Yerzat.

而这套 playbook 会继续被传承下去、继续奏效，然后会有大量来自哈萨克斯坦的成功公司出现在全球市场上。>> 谢谢你，Yerzat。

So, right now speed of distribu— that's the most important thing in startups.

现在分发的速度是创业公司里最关键的东西。

We see that companies become unicorns, decacorns in less than one year.

我们看到，成为独角兽、十角兽用不到一年时间。

Yes, we see it with Anthropic.

比如我们看 Anthropic。

Whereas before some top company, like Google, Apple, would do a presentation once every six months, right — Anthropic over the past month, 20 features in 30 days, right?

以前那些顶级公司，比如 Google、Apple，半年才开一次发布会，而 Anthropic 过去一个月里，30 天上了 20 个功能，对吧？

You, you — I don't know, you probably ship even more features per month.

你们——我不知道，你们每个月发的功能可能还更多。

What do you think, uh, is this going to be the standard now, that unicorns and decacorns appear within a year?

你觉得，这会成为现在的标准吗——独角兽、十角兽一年内就出现？

Before they took at least 10 years, and now the standard has become one year.

以前至少要 10 年，现在标准变成了一年。

What do you think, will it stay this way or is it a temporary phenomenon?

你觉得这种状况会一直持续下去，还是只是暂时现象？

Uh, well, of course, right now there's this kind of tectonic shift going on, because AI is starting to heavily reformat a large number of markets, and that's a giant opportunity.

当然，现在正在发生一场板块级的剧变，因为 AI 开始大幅重塑大量市场，这是个巨大的机会。

Plus the markets are already digitized, I mean, before that there was the internet revolution, the software revolution, the SaaS revolution — it had already transformed, uh, a large number of markets into digitalization.

另外，市场本来就已经数字化了，在这之前有互联网革命、软件革命、SaaS 革命，它已经把大量市场改造成了数字化形态。

And the ready-made field, like, already exists, and AI just accelerates everything, and AI is starting to transform everything.

现成的地盘已经摆在那儿了，AI 只是把一切加速，AI 开始改造一切。

And this period — well, why we work so much, why our competitors there, or the giants like OpenAI and Anthropic, work so much, is because everyone understands that this interval is going to be very short.

而这个阶段——我们为什么这么拼，我们的对手、或者 OpenAI、Anthropic 这样的巨头为什么这么拼，是因为所有人都清楚，这段时间会非常短。

I mean, uh, this is only going to be happening for a few years, then the market consolidates, the market gets monopolized and, well, the same thing happened in the internet market, back during the Web2 revolution, when social networks appeared, when YouTube appeared — I mean there was that, yeah, Wild West, then a sharp consolidation and everything kind of froze, and the monopolists, well, came to control everything, and a new configuration appeared, sort of.

也就是说，这种局面只会持续几年，之后市场就会整合、就会被垄断，就跟当年互联网市场一样：Web2 革命时期，社交网络出现、YouTube 出现，那时候就是一片 Wild West，然后是急剧整合，一切像被冻住了，垄断者开始掌控一切，新的格局就此定型。

That's why, well, we clearly understand that the window is very short, inside the company everyone understands this, and that's why we also understand that we have to make it in time, we have to make it faster than the American teams, than the Chinese teams.

所以我们非常清楚，这个窗口很短，公司内部人人都明白这一点，所以我们也明白，我们必须赶得上，必须比美国团队、比中国团队更快。

And because of that, well, we're maximally, well, afraid of missing this chance, and we try to do the maximum that's possible.

正因为这样，我们特别怕错过这个机会，所以拼尽全力去做到极限。

Murat, my next question for you is a classic one.

Murat，我下一个问题要问您，是个经典问题。

About AI — it's everywhere now — about the AI hype, about the AI bubble.

关于 AI——它现在无处不在——关于 AI 炒作，关于 AI 泡沫。

Uh, what do you think about it?

您怎么看这件事？

There you go, the classic question.

就是这么个经典问题。

The competition in the sector is very crazy, but I want to say that the volume of funding is also crazy.

这个赛道的竞争非常疯狂，但我想说，融资规模同样疯狂。

And the volume of venture investment in the first quarter of 2026, compared with the first quarter of 2025, has increased threefold.

2026 年第一季度的风险投资规模，与 2025 年第一季度相比，增长到了三倍。

There has never been such a dynamic of growth before this.

这样的增长势头此前从未出现过。

On top of that, 80% of this money goes directly into big projects.

而且这些钱里有 80% 直接流向了大项目。

There you go.

就是这样。

I mean, that volume of funding is very large and the competition is very high.

也就是说，融资体量非常大，竞争也非常激烈。

And all of this is happening because AI is really, well, its adoption is a tectonic shift in all industries.

这一切之所以发生，是因为 AI 的落地确实是所有行业的一次板块级剧变。

In all of them.

所有行业。

It's, we probably — it's even hard for us to assess the consequences of all this, but nevertheless, and probably not all AI projects will survive, that's for sure.

这一切的后果，我们大概现在还很难评估，但不管怎样，恐怕并不是所有 AI 项目都能活下来，这一点是肯定的。

But it seems to me one should worry more not even about AI Native companies, but about ordinary companies, especially SaaS companies, because every one of them will be in a zone of systemic risk.

但我觉得，更该担心的甚至不是 AI Native 公司，而是普通公司，尤其是 SaaS 公司，因为它们每一家都会处在系统性风险区里。

That's all the companies that put out, uh, software, right, because it can now be replicated very easily.

也就是所有做软件的公司，因为软件现在非常容易被复制。

All the companies that provide Software as a Service, the same thing.

所有提供 Software as a Service 的公司，也是一样。

So, I mean, in that sense, yeah, I mean, well, there's no need to worry too much about AI itself, no need, because AI itself is going to be just fine.

所以在这个意义上，其实不必太替 AI 本身担心，不需要，因为 AI 本身不会有事。

You need to worry about the incumbents, the classic services, when they'll be, like, well, in a very heavy disruption, and apparently the ones who survive will be those who master exactly these new tools as fast as possible.

该担心的是在位者、那些传统服务，它们会遭遇非常剧烈的颠覆；而能活下来的，看起来是那些尽可能快地掌握这些新工具的人。

Yerzat, thank you so much for your time.

Yerzat，非常感谢你抽出时间。

Uh, to finish up, uh, I wanted to ask this question.

最后，我想问这样一个问题。

Uh, what do you think in general about not just video AI, but AGI in general, artificial super intelligence?

你整体上怎么看——不只是视频 AI，而是 AGI、artificial super intelligence？

And where, where are we heading?

我们到底在走向哪里？

You're standing right on the front line, so for ordinary, uh, for all of us, for the people living here, what should we expect over 5 years, maybe 10 years, maybe 2 years, right?

你就站在最前沿，那么对普通人、对我们所有生活在这里的人来说，未来 5 年、也许 10 年、也许 2 年，该期待些什么？

Like, how will it directly affect our life, all of us ordinary people?

它会怎样直接影响我们所有普通人的生活？

Yeah, of course.

好，当然。

Well, AI is first and foremost an increase in productivity.

AI 首先是生产力的提升。

And classical economic theory says that when productivity rises under capitalism, it quickly happens that, well, inequality in society accelerates.

而古典经济学理论认为，在资本主义体制下，生产力提升会让社会不平等迅速加剧。

And that's probably the biggest concern for developed economies — that inequality is going to increase.

这大概就是发达经济体最大的担忧——不平等会不断扩大。

For us, as Kazakhstan, it's an even riskier situation, because we're already, well, in that Gini distribution not, well, not at the very top.

对我们哈萨克斯坦而言，情况风险更高，因为我们本来在基尼系数分布上就已经不算靠上。

And since we're, like, part of the capitalist economy, fairly, well, we're in a vulnerable position, uh, globally.

而且既然我们身处资本主义经济之中，我们在全球范围内的处境是相当脆弱的。

That's why we also understand the importance of, like, building AI companies.

所以我们也明白，做出 AI 公司这件事有多重要。

We have a lot of people in the company who understand this, a lot of patriots working there, and for them it's kind of like a mission, because, well, you can't repeal the laws of economics.

公司里有很多人明白这一点，有不少怀着家国情怀的人在做事，对他们来说这更像是一种使命，因为经济规律是废除不了的。

Uh, we'll be facing a large number of new challenges, like, as a society they'll be standing in front of us, because, well, that's just how capitalism is built.

作为一个社会，我们将会面对大量新的挑战，因为资本主义本来就是这么运转的。

And, uh, further on, yeah, really, maybe some of that stuff that's fashionable to discuss in, obviously, in rich countries — UBI, you know, Universal Basic Income — that there will be some redistribution systems, like a welfare state, uh, something like that will, well, be happening in late stage capitalism economies, like in the US, like in Europe.

再往后，确实，可能会出现那些在富裕国家里很时髦的话题——UBI，也就是全民基本收入——会有某种再分配机制，比如福利国家，类似的东西大概会在晚期资本主义经济体里出现，比如美国、比如欧洲。

And Kazakhstan, of course, is in a pretty vulnerable position here.

而哈萨克斯坦在这里当然处在相当脆弱的位置。

Uh, and we, as Higgsfield, are trying to do the maximum of what we can do in these realities.

而我们 Higgsfield，是在这样的现实条件下尽力去做我们能做到的极限。

Uh, and globally, yeah, that's probably the biggest problem.

从全球来看，这大概是最大的问题。

Uh, but at the same time, of course, the number of new opportunities that are appearing is also enormous.

但与此同时，新涌现出来的机会当然也是巨大的。

I mean, for young people who want to, uh, create something, some new value, uh, in the market, then AI is a giant opportunity, because, as we said earlier, there's a transformation happening, a revolution, uh, of industries, uh, of markets, and it's a giant opportunity.

也就是说，对那些想在市场上创造出某种新价值的年轻人来说，AI 是一个巨大的机会，因为正如我们前面说的，产业、市场正在发生一场转型、一场革命，这就是巨大的机会。

And obviously, capitalism is built in such a way that, uh, it's very bad when everything becomes static, and gatekeeping shows up, and young people have nowhere to apply themselves, they can't build new businesses, and everything gradually gets corrupted, gets bureaucratized.

而且很明显，资本主义的运转方式决定了：当一切都变得固化、出现了守门人、年轻人无处施展、做不出新的生意、一切逐渐腐化和官僚化的时候，那是非常糟糕的。

Uh, and AI, of course, disrupts all of that.

而 AI 当然把这一切都颠覆掉了。

Well, and we see this, like, in the American economy.

这一点我们在美国经济上就能看到。

the whole American stock market is growing purely off AI, off Nvidia, data centers, investment into infrastructure.

整个美国股市的上涨纯粹靠 AI，靠 Nvidia、数据中心、对基础设施的投资。

We see that Americans see that the S&P is going up, their, well, and the psychology of the average American, he sees that the S&P is going up.

我们看到美国人看到 S&P 在涨，普通美国人的心理就是这样：他看到 S&P 在涨。

It's like, well, in Kazakhstan a person sees that his house, the value of his house is going up, and he's willing to spend more, because in his head he thinks he has more money.

这就好比在哈萨克斯坦，一个人看到自己的房子、房子的价值在涨，他就愿意多花钱，因为他脑子里觉得自己的钱变多了。

Well, his net worth is bigger, because the price of his house went up.

他的身家变多了，因为他房子的价格涨了。

Same thing with the average American — since they have high penetration, uh, in that people hold a lot of stocks, then on average America, the average American, his wealth has grown on average, because the S&P went up, the top-10 stocks went up a lot, went up because of AI.

普通美国人也是一样——因为他们的持股渗透率很高，很多人手里握着大量股票，所以平均而言，普通美国人的财富涨了，因为 S&P 涨了，前十大权重股涨了很多，而这些都是靠 AI 涨起来的。

And because of that Americans consume a lot, spend a lot of money, because in their heads they, well, got richer.

正因为如此，美国人消费得很多、花钱很多，因为在他们自己脑子里，他们变富了。

And this system has already been growing pretty steadily for several years.

而这套系统已经相当稳定地增长了好几年。

And here AI has to constantly, well, keep supporting it.

而在这里，AI 必须持续不断地为它提供支撑。

That's why America is ready to invest even more money into infrastructure, because, well, without that, with Trump's policies and so on, well, some bad changes could happen in the economy.

所以美国愿意往基础设施里投更多的钱，因为没有这些，再加上特朗普的政策之类的，经济上可能会出现很糟糕的变化。

And so, well, it's an existential question for late stage capitalist societies — to invest even more into AI, as much as they possibly can.

所以说，对晚期资本主义社会来说，把更多的钱投进 AI、投到能投的极限，是一个生死攸关的问题。

Uh, and accordingly we'll see a rise in productivity, and accordingly, unfortunately, we'll see a rise in inequality.

相应地，我们会看到生产率上升，也相应地，很遗憾，会看到不平等加剧。

And then, well, I don't know, a social explosion, or the institutions will have to prepare some new measures in advance to smooth all of this out.

再往后，我也说不好，要么是社会层面的爆发，要么就是各类机构必须提前准备好一些新的措施，把这一切抹平。

Well, a pretty interesting time is waiting for all of us over the next several years.

总之，接下来这几年对我们所有人来说都会是相当有意思的一段时间。

I'd also like to add that right now we're running consultations with the owners and the heads of large quasi-governmental and private structures.

我还想补充一点，我们现在正在和大型准国有机构以及私营机构的所有者、负责人做咨询。

And the message that's being discussed and that's becoming more and more obvious is that regardless of which industry the holding company or some sector company works in, right, uh, you nevertheless already need to have your own in-house competence in this area.

现在被反复讨论、而且越来越清晰的一个信息是：不管这家控股公司或者某家行业公司身处哪个行业，都必须拥有自己在这个领域的专业能力。

This is becoming really critically important, because in 2-3 years — specialists don't appear in a single day.

这一点正变得极其关键，因为两三年——专家不是一天之内冒出来的。

And we're discussing exactly that: you need to build your own garden, right, your own sandbox.

我们讨论的正是这件事：要建自己的花园，也就是自己的沙盒。

It's not that much money to spend on it from a big corporation's point of view, but it will let you, at least on a 2-year horizon, already train up the people, right, so that, well, on your own, uh, how to put it, you understand where to move by yourself.

从大公司的角度看，花在这上面的钱并不算多，但至少能在两年的时间跨度里把人先培养出来，这样才能靠自己弄明白该往哪个方向走。

not just to be, like, either an AI optimist or an AI pessimist, right, but you literally need to build your own units, prepare them, train them.

不是只做一个 AI 乐观派或者 AI 悲观派，而是要实打实地组建自己的部门、把人准备好、把人培养出来。

So, like, well, that's the message that's actively going around right now.

总之，现在活跃流传的就是这样一个信息。

Thank you very much, Yerzat, Murat.

非常感谢，Yerzat、Murat。

Thanks, everyone.

谢谢大家。

Thank you very much.

非常感谢。