唐·威尔逊：GPU市场的万亿美元之问

有些市场故事，你想知道每一个细节。

There are some market stories where you want every detail.

我和乔已经基于这个做了好几期播客了。

Joe and I have made quite a few podcasts on that basis.

但有时候你只有十分钟，就想快速了解市场在动什么。

But sometimes you've only got ten minutes and just wanna know what's moving markets fast.

这就是巴克莱简报播客。

That's the Barclays Brief Podcast.

每周，巴克莱市场和研究部门的专家都会帮你快速了解最新动态，以及接下来该关注什么。

Every week, experts from Barclays Markets and Research get you up to speed on what's happening and what to watch next.

简洁、聚焦、简短。

Concise, focused, and brief.

无论你在哪里收听播客，搜索“巴克莱简报”即可。

Search Barclays Brief wherever you get your podcasts.

你好。

Hi.

我是Barry Ritholtz，邀请你加入我的商业大师播客。

I'm Barry Ritholtz inviting you to join me for the masters in business podcast.

每周，我们都带来与那些塑造市场、投资和商业的人们的对话。

Every week, we bring you conversations with the people who shape markets, investing, and business.

我与CEO、诺贝尔奖得主、市场创新者以及传奇投资者交谈。

I speak with CEOs, Nobel laureates, market innovators, and legendary investors.

无论你持有股票、债券、房地产、大宗商品，甚至加密货币，这些讨论都是你绝对不能错过的。

Whether you own stocks, bonds, real estate, commodities, even crypto, these are discussions you absolutely need to hear.

在Apple、Spotify或任何你收听的地方订阅商业大师播客。

Subscribe to the masters in business podcast on Apple, Spotify, or anywhere you listen.

Bloomberg音频工作室，播客、广播、新闻。

Bloomberg Audio Studios, podcasts, radio, news.

大家好，欢迎收听另一期Odd Lots播客。

Hello, and welcome to another episode of the Odd Lots podcast.

我是Joe Weisenthal。

I'm Joe Weisenthal.

我是特蕾西·阿勒维。

And I'm Tracy Allewy.

特蕾西，我们现在在芝加哥。

Tracy, we're here in Chicago.

这是我们连续第二年在芝加哥做现场节目了，这个城市对我们来说很棒。

Second year in a row that we've done a live show here in a great city for us.

没错。

That's right.

我们要对话的这位，去年我们也聊过，但十二个月里发生了很多变化。

And we're gonna be talking to someone who we spoke with last year, but a lot has changed in twelve months.

这位正在尝试为计算能力构建一个市场。

So this is someone who is trying to build a market for compute.

所以我们将要对话的是唐·威尔逊。

So we're gonna be speaking with Don Wilson.

他是DRW的创始人兼CEO。

He is the founder and CEO of DRW.

他确实是推动GPU这类商品期货市场建设的先锋人物。

He is really at the forefront of this effort to build futures markets for things like GPUs and so forth.

来听听我们的对话吧。

So check out the conversation.

唐，我们跟你聊，差不多正好是一年前。

We talked to you, Don, basically exactly a year ago.

上次我们在这里，聊了像GPU期货之类的话题。

Last time we were here, we talked about things like GPU futures.

但在AI领域，那感觉就像十年前的事了，因为变化实在太大了。

But in, like, AI time, it might as well have been ten years ago because so much has changed.

我们想聊聊GPU市场结构，以及所有相关的事情。

We wanna talk GPU market structure, all that stuff.

过去这几天，在AI交易圈里其实挺有意思的。

The last couple of days have actually been kinda interesting in the AI trade world.

所有人都在盯着CoreWeave。

Everyone's watching CoreWeave.

大家都在关注CoreWeave的债券。

Everyone's watching CoreWeave bonds.

在IPO之前，有一个Anthropic的投资者会议。

There was an anthropic investor meeting before the IPO.

有一些波动，好像他们的收入增长只有50%，而不是51%，或者诸如此类。

A little jitter as though maybe their, you know, maybe their revenue only grew up, whatever, 50% versus 51%.

具体数字我不清楚。

I don't know the exact numbers.

你现在怎么看？

Give us your read right now.

是不是AI领域有点降温了？

Like, is there a little slowdown happening in AI?

就是说……

Like, is there...

我们是不是在放慢节奏？

Are we shifting into a lower gear?

发展轨迹上有什么变化吗？

Is there anything in the trajectory that's changed?

任何增长速度这么快的新事物，都不会以恒定的速度增长。

Any new thing that's growing this fast is not going to grow at a constant rate.

它不会以恒定的速度增长……

It's not going to grow at a...

一阶导数不会是常数。

The first derivative isn't going to be constant.

二阶导数也不会是常数。

The second derivative isn't going to be constant.

所以对我来说，这只是预期中的波动。

So, to me, this is just kind of expected volatility.

我不知道，也许人们八月去度假了，所以消耗的Anthropic代币少了。

I don't know, maybe people went on vacation in August so they didn't consume as many anthropic tokens.

但你并没有从中解读出更多的东西。

But you're not reading anything more into it than that.

我没有。

I'm not.

好的。

Okay.

等等。

Wait.

那像CoreWeave这样的公司，你们做计算交换和硅数据，很大程度上是这样一个想法：如果你为计算力创造一个市场，那么云公司就可以对冲他们未来的收入流，资本成本也能降下来。

So with something like CoreWeave, a big part of what you're doing with compute exchange and silicon data is this idea that if you create a market for compute, then cloud companies can hedge their future revenue streams and the cost of capital can come down.

在这个星期，当你看到CoreWeave的债券收益率达到15%时，这说明了什么？

In a week like this, when you see the yield on a CoreWeave bond at 15%, what does that tell you?

是不是意味着，如果计算期货市场能运行起来，CoreWeave的所有问题就都能解决？

Is the suggestion that, like, well, if if compute futures were up and running, all of CoreWeave's problems will be solved?

那，

Like, what

你对此怎么看？我确实认为，这凸显了对风险管理产品的巨大需求。

do you about I mean, I do think that it's highlighting the the massive need for risk management products.

所以你想一想，如果你是做新式云计算的，你跟全球信用最好的公司签了合同，那你当然没问题，对吧？

And so, you know, think about if you're a neocloud and, you know, you sign contracts with the most creditworthy companies in the world, then you're fine, right?

但是还有很多其他客户用GPU，他们的信用并不是全球最好的。

But there are a lot of other consumers of GPUs that are not the most credit worthy companies in the world.

而如果你做新式云计算，你想把服务卖给他们。

And if you're a neocloud, you want to be able to sell to them.

所以我认为市场在传递的信号就是，如果你按需出售计算服务，或者你把服务卖给信用不太好的客户，那你的融资成本就会更高。

And so I think what the market is saying is if you're selling on demand compute or you're selling to less credit worthy entities, then your cost of capital is just going to be higher.

现在想象一下，如果GPU期货市场出现了，你作为一家新式云厂商，你可以买下GPU，部署到你的数据中心，卖出一系列的GPU期货，然后如果你想，你还可以按需出售计算服务。

Now imagine once GPU futures exist, you can, as a neocloud, buy the GPUs, deploy them in your data center, sell a strip of GPU futures, and then just sell on demand compute if you want.

或者你可以跟信用不太好的公司签六个月的合同，让他们提前支付很大一部分的合同金额。

Or sell six month contracts to less creditworthy companies that can post a big percentage of the contract value upfront.

市场会奖励那些这么做的公司，它们基本上就是大规模地进行基差交易。

And the market is going to reward companies that do that, that basically just have on a giant basis trade heavily.

他们的融资成本会低得多。

Their cost of capital is to be much lower.

特里提到，好吧，GPU期货还不存在，而且白宫那边有些风声，比如国家安全什么的，然后他就说，我不明白。

Terry mentioned, okay, the GPU futures don't exist yet, and there's some message from the White House, like national security, etcetera, and he's like, I don't get it.

你明白吗？

Do you get it?

你……

Do you...

你能告诉我们为什么这些产品还不存在吗？

Can you tell us something about why these products don't exist yet?

也许是因为监管上对上市有什么顾虑？

Maybe what the regulatory hesitancy is about the listing?

对，顺便提醒我们一下监管审批的状况如何。

Yeah, and just remind us what the state is of regulatory approval.

是的，现在有一个公开的评论期。

Yeah, so right now there's a comment period that's open.

CFTC延长了评论期。

The CFTC extended it.

如果他们不采取任何行动，期货可能在11月9日上市。

If they don't do anything, the futures can get listed November 9.

而现在，11月9日刚好在11月3日之后。

Now, conveniently November 9 is after November 3.

所以也许华盛顿的一些人会对算力期货感到更安心，我也不确定。

And so, you know, maybe some people in Washington will be more comfortable with the compute I futures don't know.

我没办法解释他们的担忧是什么。

I can't explain what the concern is.

我听说过各种不同的说法。

I've heard all sorts of different theories.

一种说法是，有些指数不够稳健。

You know, one theory is that some of the indices are not very robust.

我想说...

I would...

好的。

Okay.

所以我对这个指数的稳健性非常好奇，因为很多GPU是通过双边承购协议销售的。

So I'm actually very curious about this, the robustness of the indices, because a lot of GPUs are sold via bilateral offtake agreements.

就像Anthropic和SpaceX有某种协议，对吧，用他们的计算能力。

So it's like Anthropic has some deal with SpaceX, right, to use their compute.

我们其实不知道那个价格是多少。

We don't really know the price of that.

那为什么有人看着GPU价格指数就应该相信它确实反映了市场的情况呢？

Why should anyone feel confident then when they look at an index of GPU prices that, like, it actually represents something resembling the state of the market?

看这儿。

Here.

听我说。

Listen.

我觉得这得买家自己小心。

I think it's buyer beware.

好好了解一下你交易期货所依据的那个指数吧。

Learn about the index that you're trading futures on.

好的。

Okay.

硅谷数据那支团队做得非常出色，他们收集了大量数据，主要关注新云服务商的按需计算。

You know, Silicon Data, that team has done an amazing job of collecting massive amounts of data from, you know, and it's focused on on demand compute from neoclouds.

新云服务商很多。

You know, are a lot of neoclouds.

超级云服务商却很少。

There are very few hyperscalers.

所以，这些指数主要聚焦于新云服务商，而按需计算方面透明度很高。

So, are indexes focused on neoclouds, And of which there are there is a lot of transparency about on demand compute.

而且，他们做得很好，打造了一个符合IASCO标准的指数。

And you know, they've done a very good job to create an index that, you know, will be IASCO compliant.

那才是真正的基准。

That's the benchmark.

其他一些供应商，我觉得他们可能根本没听说过IASCO合规这个说法。

Some of the other providers I don't think have ever heard the word, the phrase IASCO compliant.

等等，什么是IASCO？

You wait, what's an IASCO?

抱歉。

Sorry.

那么，IASCO是，我不太确定，它是国际监管机构的一个组织，我记得在2013年他们为指数制定了最佳实践准则。

So, IASCO is the, I don't know, it's the kind of group of international regulators that I think in 2013 came up with a best practices approach for indices.

我们以后可以做一期关于IASCO的节目。

We should do an IASCO episode one day.

既然你已经在期货合约上工作了一段时间，能不能跟我们讲讲你在GPU定价走势方面的发现？

Now that you've been working on the futures contract for a while, can you maybe tell us what you've learned about, I guess, the pricing trajectory for GPUs over time?

因为这是大家担心的一个点。

Because this is one of the concerns.

你可以说你整体看好算力市场，但这并不一定意味着你看好单个GPU的价值。

You can say you're bullish on the compute market overall, but it doesn't necessarily mean that you're bullish on the value of a single GPU.

GPU会随着时间贬值。

It can depreciate over time.

也许它会变得更高效，然后人们会转向新的事物。

Maybe it gets more efficient and people move on to new things.

所以这就是那个万亿美元的问题。

So that is the trillion dollar question.

这就是为什么这些我

That is why these I'm

很高兴我问了这个问题。

glad I asked it.

而这正是这些期货合约需要存在的原因，因为没有人知道B200四年后或五年后的价值会是多少。

And that is why these futures need to exist because, you know, nobody knows what the value of a B200 will be four years from now, five years from now.

所以，通过引入远期交易的期货合约，我认为最初CME的期货合约会覆盖到三年后。

And, you know, and so by introducing futures contracts that trade way out, I think initially the CME's futures contracts are going to go out three years.

随着它们流动性增强，期限会进一步延长。

And, as they become more liquid they'll go out more.

但这让人们可以对冲这种风险。

But that enables people to hedge that.

差不多一年前的十一月，H100的按需每小时价格一直在跌、跌、跌。

Back in November of, you know, almost a year ago, the price of H100s, the hourly on demand price had just been declining, declining, declining.

当时大家都在说，为什么需要GPU期货？

And everybody was saying, why do you need GPU futures?

这东西只会往下跌。

These things just go down.

结果就在感恩节前后，价格突然反转，开始往上涨了。

And then, you know, right around Thanksgiving, they turned around and kind of went up into the right.

发生了什么？

And what happened?

是Anthropic推出了最新模型，性能大幅提升，突然之间推理需求暴增。

You know, it was Anthropic's newest model came out which was a massive step up in And all of a sudden there was a ton of demand for inference.

最近我们还看到，B200、B2300和H100的价格又迎来了一波上涨。

And, you know, more recently we've seen, looking at B200, B2300 and H100 prices, another surge upward in price.

顺便说一句，与此同时，硅数据代币指数价格却暴跌了。

And by the way, at the same time that the silicon data token index price has plummeted.

这很有意思。

It's interesting.

所以说……

And and so...

这些东西波动很大。

I mean, these things are volatile.

它们可能会涨。

They can go up.

也可能会跌。

They can go down.

所以对冲工具是非常必要的。

And and a hedging instrument is badly needed.

有些市场故事你确实想了解每一个细节。

There are some market stories where you want every detail.

乔和我就是基于这一点做了不少期播客。

Joe and I have made quite a few podcasts on that basis.

但有时候你只有十分钟，就想快速了解什么在带动市场变动。

But sometimes you've only got ten minutes and just wanna know what's moving markets fast.

这就是巴克莱简报播客。

That's the Barclays Brief Podcast.

每周，巴克莱市场和研究团队的专家都会让你快速了解当前动态以及接下来该关注什么。

Every week, experts from Barclays Markets and Research get you up to speed on what's happening and what to watch next.

简明、聚焦、简短。

Concise, focused, and brief.

在你收听播客的任何平台搜索“巴克莱简报”即可。

Search Barclays Brief wherever you get your podcasts.

深入了解那些塑造当今复杂经济的创新者、颠覆者以及科技驱动的趋势。

Gain insight on the innovators, disruptors, and tech driven trends shaping today's complex economy.

我是卡罗尔·马瑟。

I'm Carol Masser.

我是蒂姆·斯特内维克。

And I'm Tim Stenevek.

用彭博商业周刊每日播客结束你的工作日吧。

Wrap up your workday with the Bloomberg Businessweek Daily Podcast.

我们带你深入探讨那些塑造你世界的重大故事，从人工智能的演变到全球商业重心的转移。

We bring you deeper dives into the stories shaping your world from the evolution of AI to the shifting priorities of global business.

还有硅谷的权势人物和最新的科技趋势。

Plus Silicon Valley power players and the latest tech trends.

白天错过的精彩对话，这里都能补上。

Catch up on the conversations you miss during the day.

在苹果播客、Spotify或你常用的任何平台订阅彭博商业周刊每日播客。

Subscribe to the Bloomberg Businessweek daily podcast on Apple, Spotify, or anywhere you listen.

比如打个比方，假设GPU随时间贬值，尽管需求增加，但总体价值还是会下降，这在很多大宗商品市场里都见过。

Just as a hypothetical, let's say GPUs depreciate over time, and there's more demand, but in general, the value goes down, which is something we've seen in a lot of commodity markets.

那样的话，有个能让你对冲未来GPU收入流的市场，是不是只会让大家借更多钱、建更多产能，而不是真正降低风险？

In that case, does having a market where you're able to hedge your future GPU revenue streams, does that just enable people to, I guess, borrow more and overbuild versus genuinely reduce risk.

你只不过让大家在一场看似对冲的交易里，围绕同样的风险去借钱，但实际上根本并没有朝对你有利的方向走。

You're just allowing people to borrow against the same risk ostensibly with a hedge, but the underlying isn't actually moving in your favor.

这说得通吗？

Does that make sense?

说不通。

No.

不对。

No.

这里的关键是，期货合约要想存在，必须有愿意买的买家和愿意卖的卖家。

I mean, I mean, here, the reason it doesn't make sense is that for a futures contract to exist, there has to be a willing buyer and a willing seller.

所以，如果你的假设成立，那得是他们能以某个设定的价格——可能偏高——卖出期货合约，而且市场上对这种合约有无限的需求。

So so your hypothesis would work if they could just go and sell futures contracts at some price that was made up, maybe too high, and there was kind of infinite demand for that.

好吧，但另一方必须有理性的买家。

Okay, well, But there has to be a rational buyer on the other side.

所以我认为，这些远期合约的价格将会是……

And so I think what we'll see is that those forward contracts are going to be...

实际上会以更低的价格成交。

Are in fact going to be at lower prices.

因为远期市场上有太多想要卖出的对冲需求了。

Just because there's so much hedging demand to sell in the forward space.

而且需要有投资者愿意通过购买这些期货合约来获取回报。

And there will need to be, you know, investors who want to generate a return by buying those futures contracts.

什么是……

What is the...

如果往后看几年，要让这个市场变得丰富，需要怎样的云和推理提供商的生态系统呢？

If we cast out a few years, what is the sort of ecosystem of clouds and inference providers that needed to exist for this to be a rich market.

因为，比如说，如果AI的未来只是……

Because, like, if the if the future of AI is just...

假设AI的未来只是Anthropic和OpenAI，而云的未来只是Meta、Alphabet和亚马逊，那就不需要期货市场了。

Let's say the future of AI were just Anthropic and OpenAI, and the future of cloud were just Meta, Alphabet, and Amazon, there wouldn't need to be a futures market.

那就只需要双边协议、承购协议就行了。那么，除了OpenAI和Anthropic之外，我们需要多少长尾推理提供商，才能让这个市场变得有用，而不是仅仅靠双边协议就能解决呢？

It would just be all bilateral agreements, off take agreements, So how much, like, how fat of a tail of non open AI and anthropic inference do we need such that this is a useful market that cannot just be solved by bilateral agreements?

嗯，我觉得我们已经到那一步了。

Well, I mean, I think we're already there.

你看，开源模型的使用量正在快速增长。

Mean, look at the growing use of open source models.

增长非常迅猛。

I mean, it's exploded.

现在你都能看到一些估值几十亿美元的新公司，专门把开源模型部署在 GPU 上，然后出售推理服务。

And you're seeing companies with many billion dollar valuations that have just been created to put open source models on GPUs and sell the tokens.

比如 Fireworks 和 Together。

You know, like Fireworks and Together.

所以我们已经拥有一个非常丰富的生态系统了。

And and so we already have a really rich ecosystem.

那它现在足够丰富了吗？

Is it rich enough yet?

绝对够了。

For sure.

好的。

Okay.

有意思。

Interesting.

我们要不要聊聊永续合约？

Should we talk about perpetuals?

等等。

Wait.

我能问一下吗？对。

Can I ask Yeah?

请说。

Go for

我能问一下吗？

Can I ask?

关于GPU期货，怀疑者常说的一点是，GPU或算力不够同质化，不适合做期货市场。

On GPU futures, so one of the other things that people say, the skeptics, is they say GPU GPU or compute is not homogenous enough for futures market.

这里面有几个不同的方面。

And there's a few different things.

他们说，比如有不同的软件架构，这可能会大幅改变价值。

They say, like, there is different software arrays, and that could significantly change the value.

还有一点是，在AI推理这部分，有很多其他玩家，比如AMD、Cerberus、Etched，还有那些Tranium，微软的那个什么，以及TPU等等。

There's also the fact that there's, like, at least in the inference part of AI, there's a lot of other players who want AMD and Cerberus and Etched and all these other Tranium and whatever the Microsoft one was and TPUs, etcetera.

这个市场实在是太多样了，真的很难用一个指数来涵盖。

And this is just way too diverse of a market to really be captured in an index.

你有多大的信心，要么市场会足够同质化来支持期货，要么大家对于H100指数和类似产品之间的基差有足够的了解，从而让这个市场能够服务于多样化的供应商群体？

How confident are you that either the market is gonna be homogenous enough for futures or that there will be enough sophistication about the sort of basis between the h 100 index versus something similar such that this market could serve a diverse provider base?

首先，光是在H100或者B200上的投入金额就已经足够大，足以支撑一个单独的期货市场了。

I mean, first of all, the the amount of money going into, you know, just h one hundreds or just b two hundreds is large enough to support a single futures market.

另外，这些其他产品之间是高度相关的。

Now, in addition to that, these other things are very correlated.

所以人们就可以在这些产品和那个指数之间做基差交易。

And so, you know, people will be able to trade a basis on these things versus whatever it is.

而且，显然这些可能都是基于标准的默认配置之类的。

And, know, maybe obviously, you know, these are going to be off of kind of the baseline standard configuration whatever.

所以也许有人打算进行训练，或者建立一个大型集群。

And so maybe somebody's planning on doing a training run and maybe a giant cluster.

当然，那会带有溢价。

Well, that's going to bear a premium, of course.

但如果推理需求大幅上升，你猜怎么着？

But, you know, if the demand for inference goes up massively, then guess what?

训练成本也会跟着涨。

That training run's also going to be more expensive.

所以这其实就是不同产品之间的价差。

So it's just a basis between the things.

这是其中一点。

That's one thing.

另一点是，将来——而且可能已经快了——你可以在输出上，也就是在token本身上面交易期货合约。

The other thing is that eventually, and we may be getting close, you'll be able to trade futures contracts on the output, on the tokens themselves.

哦，token期货啊。

Oh, token futures.

代币期货。

Token futures.

而且，你知道，当然，代币种类很多，大家也会提出同样的质疑。

And so, know, again, there's, of course, a diverse number of tokens and people will put up the same objections.

但我认为随着时间推移，这是有可能实现的。

But I think that over time that's going to be possible.

顺便说一句，这会带来一个非常有趣的可能性，那就是……

And by the way, that will introduce a really interesting possibility, which is...

现在想想，如果你去租GPU，按需付费，然后加载一个开源模型，通过OpenRouter之类的平台开始卖代币。

Now think about if you go and rent GPUs, you pay the on demand price, And then you load up an open source model and you start selling tokens through OpenRouter or something.

那么，这样做能盈利吗？

Well, can you do that profitably or not?

嗯，你能赚到的那个利润率，就像炼油厂一样，对吧？

Well, that profit margin that you can generate is akin to what an oil refinery does, right?

或者说，就像大豆压榨厂那样。

And or what a soybean crush plant does.

所以这就会形成一个裂解价差。

So going to have a crack spread.

在石油行业里，我们称之为裂解价差。

So in oil we call it a crack spread.

在大豆行业里，我们称之为压榨价差。

In soybeans we call it a crush spread.

所以你是第一个听到这个的。

So you heard it here first.

在DRW，我们把这个叫做挤压价差。

At DRW we call this the crunch spread.

那太棒了。

That's great.

等等，这是

Wait, is this

代币期货听起来像是你们正在积极研究的东西，对吧？

token futures are something you're actively working on, it sounds like?

是的。

Yes.

好的。

Okay.

那你觉得现在处于什么阶段呢？

And what sort of stage would you say you're I

其实现在还只是想法阶段，但说实话，硅谷数据的代币指数已经提供了一些相当有趣的数据，而且被广泛讨论。

mean, it's, you know, it's the thought stage, but but but but, I mean, here, to to be fair, the the, you know, the Silicon Data token index, you know, is providing some pretty interesting data and is widely discussed.

所以这是个开始。

So, you know, that's a start.

那我们先回到GPU期货这个话题。

And so just going back to GPU futures for a second.

所以每次有新芯片推出，就需要单独的合约和新的流动性池。

So every time a new chip comes out, that needs to be a separate contract and a new pool of liquidity.

对吗？

Is that right?

你是这样想的吗？

Is that how you're thinking of it?

对。

Correct.

好。

Okay.

那你打算怎么……

And how do you...

这看起来挺有挑战性的。

I mean, it seems kind of challenging.

你得跟上技术更新的周期，这确实……

You have to, like, keep up with the technological cycle, which is

更新很快，但也没那么快，对吧？

It's very rapid but not that rapid, right?

而且你可能不会为每一个芯片都推出期货合约。

And you're probably not going to launch a futures contract for every single one.

你们只会为主要的那几种推出期货合约。

You're going to launch futures contracts for the ones that are in the major ones.

而且，你知道……

And, you know...

但这没关系。

But that's okay.

有各种各样的……

I mean, there are different...

想想电力合约有多少，针对每个小的节点，我的意思是，有成千上万种。

You know, think about how many electricity contracts there are for every little node and, I I mean, you know, thousands of them.

石油合约也有很多。

There's a lot of oil contracts too.

只是……

Just...

石油也有很多……

There's a lot of oil....

通常我们主要讨论其中的两个。

Talk about two of them typically.

对。

Yep.

出于好奇心，其实这跟期货的话题有点分开，但我想，作为一个经营多家企业的技术专家，你……

Just out out of curiosity, and this is actually sort of separate from the futures conversation, but I guess, like, as a technologist with multiple businesses, etcetera, are you...

英伟达在这个领域的统治地位有多可动摇？

How saleable is NVIDIA's dominance in the bit?

你怎么看？

Like, what's your read on that?

因为一直有人问，其他芯片制造商是否真的能蚕食英伟达的地位。

Because this is a constant question whether any of these other chipmakers can actually eat into NVIDIA's position.

你对此有什么看法？

What's your take there?

好的。

Sure.

我认为这个市场会继续增长，而且我不确定……

I I think that the space is gonna continue to grow, and I can't...

嗯……

It...

英伟达不太可能维持住他们现有的市场份额，但我相当确信英伟达……

It's unlikely that NVIDIA will be able to maintain the market share that they have maintained, but I'm pretty confident that NVIDIA's...

英伟达会继续作为一家公司持续增长。

That NVIDIA will continue to continue to grow as a company.

获取关于推动科技行业迈向新领域的关键人物和公司的最新资讯。

Get essential news on the people and companies pushing the tech sector sector to new frontiers.

你好，

Hi.

我是Ed Ludlow。

I'm Ed Ludlow.

欢迎收听Bloomberg Tech，这是一个专注于科技、创新和商业未来的每日播客。

Join me for Bloomberg Tech, a daily podcast focused exclusively on technology, innovation, and the future of business.

每个工作日，我们都为你带来硅谷顶级公司的最新见解，以及科技界重要决策者的访谈。

Every weekday, we bring you the latest insights on Silicon Valley's top companies and conversations with tech's biggest decision makers.

在下班通勤路上收听《Bloomberg Tech》，领先掌握新闻动态。

Listen to Bloomberg Tech on your commute home and stay ahead of the news cycle.

今天就在Apple、Spotify或任何你收听的地方订阅吧。

Subscribe today on Apple, Spotify, or anywhere you listen.

去年我们对话时，我们聊的是DRW在AI方面的总体使用情况，我记得你说当时人们在用AI代理来辩论投资理论或风险分配之类的，但你们那时还没有用AI来实际部署风险。

You know, last year when we spoke, we were talking about AI use in general at DRW, and I think you said people were using agents to sort of debate investment theories or risk allocation and things like that, but you weren't using AI to actually deploy risk at that time.

Joe提到，在AI领域，事情变化非常非常快，十二个月后。

Joe mentioned that things change very, very fast in the world of AI twelve months later.

这种情况变了吗？

Has that changed?

现在AI为你们做些什么？

What's AI doing for you now?

没有。

No.

我是说，确实有更多的采用了。

I mean I mean, there's definitely more adoption.

前线采用肯定更多了。

There's more frontline frontline adoption for sure.

你们的token预算增加了吗？

Has your token budget gone up?

token预算增加了。

The token budget goes up.

是的，增加了。

Yes, it goes up.

token的使用量就像指数曲线一样增长。

The token use is like this exponential chart.

这挺有意思的。

It's pretty interesting.

你们有没有...

Do you have...

还有一个问题。

There's another question.

比如在你的各个业务中，你们有没有找到最佳的内部令牌分配方式？

Like across your businesses, have you figured out optimal internal token allocation?

我们问过很多高管，至少到现在，他们都说还在摸索中。

We ask this to a lot of executives, and at least up until recent, they're like, we're just still trying to figure this out.

但你肯定有办法让每个人都……

But you must have everyone...

当然，每个人都想要更多。

Of course, everyone wants more.

我猜每个人都总是想要最前沿的令牌成本。

I assume everyone always wants the Frontier token cost.

你们已经解决了内部令牌分配的问题吗？

Have you solved the question of, like, internal token allocation yet?

嗯，这里。

Well, here.

我觉得最重要的是要仔细考虑如何处理推理请求。

I I mean, I think the most important thing is to be thoughtful about how you route the the inference requests.

而且并不是每个推理请求都需要用到最前沿的模型。

And And not every inference request needs to go to a frontier model.

所以我们正在优化的就是通过最佳路由推理请求来最小化令牌消耗。

And so that's the thing that we're trying to optimize for is minimize token spend by optimally routing inference requests.

现在我们聊聊永续合约吧？

Now shall we ask about perpetuals?

好的。

Okay.

永续合约。

Perps.

开始吧。

Go.

特瑞是不是……

Is Terry...

不。

No.

特里怎么了？为什么特里错了？

What's Terry Why is Terry wrong?

为什么特里对永续合约的看法是错的？

Why Terry wrong about perps?

我的意思是，听我说。

I I mean, listen.

这里。

Here.

我我我……

I I I...

永续合约，在我看来，顺便说一句，特里就在听众里。

Perps, in my view, are Terry's in the audience, by the way.

但你也是他的客户，所以我知道。

But you're also his client, so I know.

我和特里都很喜欢聊这个话题。

And Terry and I enjoy talking about this.

我觉得PERPs在经济上是有用的。

My view is that PERPs serve a useful economic purpose.

它们就是期货合约。

They are futures contracts.

我给你举几个例子吧。

And I mean, I'll give you a couple examples.

很多人喜欢买原油ETF。

A lot of people like to buy the crude oil ETF.

原油ETF通常，曾经有一段时间，它总是持有最近月份的期货合约，而且它还会提前公布具体在哪一天从那个期货合约换到下一个。

And the crude oil ETF generally, at one point it was always in the front month future and it broadcast to the world exactly what day it would roll from that futures to the next futures contract.

我不太确定。

I don't know.

这里面可能有点滑点，有点效率不高。

There was probably a little bit of slippage in that, a little bit of inefficiency.

现在他们有点像……

Now they've kind of like...

现在他们是在一个时间窗口内操作，持有前三个合约，还有他们那些……

Now they do it over a window and they hold the, you know, first three contracts and whatever they've...

但依然存在大量滑点。

You know, but there's still a lot of slippage.

现在想象一下，有一种永续合约挂钩，比如说，布伦特指数。

Now imagine a perp that is tied to, let's say, the Brent index.

对吧？

Okay?

然后有一只ETF就直接买这种永续合约。

And and an ETF that just buys that that perp.

我觉得那样对ETF来说会高效得多。

I mean, I think that that would be a much more efficient experience for the ETF.

当然，这并不意味着对石油生产商来说那是个有效的对冲工具。

And, of course, that doesn't mean that that's a useful hedging instrument for an oil producer.

不是的。

It's not.

石油生产商需要对冲的是远期曲线。

An oil producer needs to be hedged way out the curve.

很好。

Great.

你可以同时拥有月度、季度期货和永续合约。

You can have monthly and quarterly futures and perks side by side.

它们实际上相辅相成，这是件好事。

They actually augment one another and that's a good thing.

再举一个例子。

Here's another example.

假设你是一个散户投资者，持有一只难借的股票，所有对冲基金都想做空，他们付年化100%的利率来做空这只股票。

Suppose you are a retail investor and you own a hard to borrow stock and, you know, all the hedge funds want to short, they're paying 100% annualized to short this stock.

如果你在散户账户里持有那只股票，你很可能...

You know, if you have that stock in your retail account, you probably are...

你的券商可能不会跟你说，嘿，顺便说一下，因为你持有这支难借的股票，我们每天给你300个基点。

Your broker is probably not saying to you, hey, by the way, we're going give you 300 basis points a day because you're holding this hard to borrow stock.

好吧，那就是年化1000%了，但每天30个基点。

Okay, that would be 1000% annualized, but 30 But basis points a

随便吧。

whatever.

所以你没有获得持有那支难借股票的好处。但如果你持有的是PERK，它机制上会…

And so you're not getting the benefit of owning that hard to borrow Now, if instead you own the perk It's mechanically...

机制上，它会直接传递给你。

Then mechanically, it's going to pass through.

所以这真的是一个让金融市场民主化的机会。

And and so it's a it's a real opportunity to democratize the financial markets.

所以民主…

So democrat...

民主化这个词，我一直不太喜欢，因为它听起来好听，但我不知道它到底是什么意思。

This word democratize, I've never really liked it because it just sounds good, but I don't know what it means.