《经济学人》AI算力成新石油！华尔街做市商悄然入场，一场万亿

共 1 集

剧集目录

集	标题	时长
P1	《经济学人》AI算力成新石油！华尔街做市商悄然入场，一场万亿	9:46

P1 · 《经济学人》AI算力成新石油！华尔街做市商悄然入场，一场万亿级市场正在重塑 [C1] p01 原声 (P1)

《经济学人》AI算力成新石油！华尔街做市商悄然入场，一场万亿级市场正在重塑 [C1] p01 原声 (P1)

发布于 2026-11-03 · 时长 9:46

字幕摘录

时间	英文	中文
0:06	Artificial intelligence is changing many things about the modern world.	人工智能正在改变关于现代世界的许多事物.
0:12	Language is no exception.	语言也不例外。
0:25	The word compute, for instance, used to be regarded as a verb.	例如,该词计算曾被视为动词.
0:29	Today even grammar pedants	今天甚至连语法小说
0:31	are begrudgingly accepted as a noun, referring to the processing power generated by advanced	被粗鲁地接受为名词,指高级的处理功率
0:37	chips.	筹码.
0:38	This new non-verbal compute is the fuel of a technological revolution.	这种新的非语言计算是技术革命的燃料.
0:44	Demand	需求
0:44	for processing UMF has surged and it is in short supply.	用于加工的UMF已经激增,而且供应不足。
0:49	About $360 billion was spent	花了大约3600亿美元

展开字幕全文（135 条）

序号	英文	中文
1	Artificial intelligence is changing many things about the modern world.	人工智能正在改变关于现代世界的许多事物.
2	Language is no exception.	语言也不例外。
3	The word compute, for instance, used to be regarded as a verb.	例如,该词计算曾被视为动词.
4	Today even grammar pedants	今天甚至连语法小说
5	are begrudgingly accepted as a noun, referring to the processing power generated by advanced	被粗鲁地接受为名词,指高级的处理功率
6	chips.	筹码.
7	This new non-verbal compute is the fuel of a technological revolution.	这种新的非语言计算是技术革命的燃料.
8	Demand	需求
9	for processing UMF has surged and it is in short supply.	用于加工的UMF已经激增,而且供应不足。
10	About $360 billion was spent	花了大约3600亿美元
11	on AI compute last year around the world, according to Boston Consulting Group, about	据波士顿咨询集团说,
12	than on copper.	比铜上。
13	By 2030 BCG expects $2.3 trillion in transactions, roughly the size of the global	到2030年,卡介苗预计交易额为2.3万亿美元,大致相当于全球交易额。
14	trade in crude oil today.	今天的原油贸易。
15	For all its impressive size, however, the compute market lacks sophistication.	然而,计算市场的规模令人印象深刻,但缺乏精密度。
16	The hyperscalers are still a somewhat arcane piece of IT jargon for a handful of giant	超标器仍然是信息技术术语中一个有点尖端的部分对于少数巨型
17	cloud computing providers, including Amazon, Google and Microsoft, and frontier AI labs such	云计算供应商,包括亚马逊、谷歌和微软以及前沿AI实验室
18	as Anthropic and OpenAI make deals for tens or even hundreds of billions of dollars of compute.	作为Anthropic和OpenAI的交易数百甚至数千亿美元计算。
19	But such arrangements are unavailable to thousands of smaller firms.	但数以千计的小公司没有这种安排。
20	Buyers and sellers also have few	买卖双方也很少
21	of the derivatives that in established markets allow them to hedge their exposures and speculate.	在既定市场上允许其套期保值和投机的衍生工具。
22	The compute supply chain then looks ripe for marketisation.	计算出供应链之后,市场化的时机已经成熟。
23	In the most simple form, chipmakers	最简单的是,芯片制造者
24	like Nvidia and AMD sell advanced AI semiconductors to equipment manufacturers such as Dell and Hewlett	如Nvidia和AMD将高级AI半导体出售给戴尔和休利特等设备制造商
25	Packard Enterprises,将硅包装成架子和服务器,出售给数据中心
26	operators.	操作员。
27	Besides the hyperscalers, those include a few large neo-clouds,	除了超标外还包括几个大新云
28	one more linguistic innovation denoting independent providers like CoreWeave and Nebius,	又一个语言创新表示独立供应商,如CoreWeave和Nebius,
29	and hundreds of smaller, less well-known firms.	以及数百家规模较小、不太知名的公司。
30	The operators in turn rent out access to compute	接线员转而租出去计算
31	to customers for a fixed period, often several years.	给顾客的固定期限,往往是几年。
32	Customers range from the hyperscalers,	顾客从超规模的,
33	which supplement their in-house compute with rentals, to start-ups running open source AI models.	用租赁来补充他们内部的计算, 用于启动企业运行开源AI模型。
34	So far so simple in theory.	目前为止,理论上很简单。
35	In practice, things quickly get complicated.	实际上,事情很快就变得复杂了。
36	That is because compute	那是因为计算
37	is heterogeneous.	异质的。
38	Different chips are sold in different servers, at different locations	不同的芯片在不同服务器上销售,地点不同
39	for different time periods.	不同时间段。
40	Matching buyers and sellers is finicky.	买家和卖家的配对很刺激
41	Every business that needs	每个需要的生意
42	processing power has a compute guy, or several, buying second-hand capacity through an opaque	处理电源有一个或几个计算员,通过不透明方式购买二手电源
43	network of middlemen.	中间人网.
44	Many of these informal brokers, some shadier than others, tout their	许多这些非正式经纪人,有些比其他经纪人更阴暗,说他们
45	通过WhatsApp和Telegram等信息应用提供.
46	The whole thing looks less like	整件事看起来不太像
47	trading a high-tech commodity and more like peddling illegal drugs.	交易一种高科技的商品更像是卖毒品
48	In a market that is so	在这样一个市场里
49	and immature, providers and purchasers alike prefer the safety of long-term contracts.	供应商和买方都倾向于长期合同的安全。
50	On paper, these guarantee capacity to users and revenues for themselves.	从纸面上看,这些保障能力是用户的,也是他们自己的收入。
51	In reality,	事实上,
52	given how fast the technology and the market of AI are evolving, it is hard to tell how much deals	鉴于人工智能的技术和市场发展的速度有多快,很难说出有多少交易
53	are worth.	这是值得的。
54	Today's buyers and sellers may not be around in a few years to honour their commitments.	今天的买家和卖家在几年内可能无法兑现承诺.
55	The surging yields on core weaves bonds highlight the uncertainty.	核心交织债券的增产凸显了不确定性。
56	Who will make a killing?	谁会杀人?
57	And who may not make it at all?	谁会活不下去?
58	With lenders struggling to price risk,	随着贷款人挣扎于价格风险,
59	many smaller operators buying and renting compute face double-digit interest rates.	许多较小的经营者购买和租赁计算,面临两位数的利率。
60	Fortunately for everyone involved, proper markets are being built around the compute supply chain.	幸运的是,在计算供应链周围正在建立适当的市场。
61	Some budding bosses, such as Compute Exchange, list GPUs, the increasingly familiar shorthand	一些萌芽的老板,如计算交换, 列出GPU, 越来越熟悉的短手
62	for graphics processing units, AI's silicon workhorse by location, rental duration, price	对于图形处理单元,AI按地点、租赁期限、价格分类的硅马
63	and availability.	和可用性。
64	Titans of global derivatives trading, including CME Group and the Intercontinental	包括CME集团和洲际
65	Exchange, hope to launch their own compute futures.	交换,希望推出自己的计算未来.
66	They have teamed up with startups	他们和初创企业结为伙伴
67	pioneering compute indices.	率先计算指数。
68	Pegged to the rental price of popular AI chips,	以流行AI芯片的租金为代价
69	they await regulatory approval.	他们等待监管批准。
70	Rather than wait for exchange traded derivatives to be approved,	而不是等待外汇交易衍生工具获得批准,
71	some compute market pioneers are creating over-the-counter instruments of the sort	一些计算出来的市场开拓者正在制造一种场外的仪器
72	common in credit and foreign exchange markets.	常见于信贷和外汇市场。
73	In March, a startup called Aneira Labs arranged	3月,一个叫做Aneira Labs的启动机构安排
74	a six-figure pilot trade for June with Falcon X, a cryptocurrency broker,	6月与密码货币经纪人Falcon X进行6位数的试点交易,
75	基于Nvidia流行的H100 GPU的未来租价.
76	Luxor, another startup,	鲁克索,又一个开始
77	offers operators swap-like contracts that guarantee fixed cash flow per server per hour.	提供运营商互换式合同,保证每个服务器每小时固定现金流量.
78	If the spot price for computers below the price agreed in the swap,	如果计算机的现货价格低于交换中商定的价格,
79	the operator is paid the difference.	接线员领取差额。
80	If the spot price is higher, it is the operator that pays.	如果现货价格较高,则由经营者支付。
81	This payment explains Ethan Vera, one of the firm's co-founders, may, for example, be made	这笔钱可以解释公司联合创始人之一的伊森·薇拉,比如说,可能要支付
82	through a revenue sharing arrangement.	通过收入分享安排。
83	Luxor's bespoke derivatives are modest for now in the	鲁克索的口号衍生物现在很温和
84	tens of millions of dollars, but the potential is vast.	几千万美元,但潜力很大。
85	As American exchanges await the thumbs	当美国交易所等待拇指
86	up from regulators, upstarts are getting off the ground elsewhere.	从监管者开始,上线开始在别处落地。
87	From its exchange in Bermuda,	从百慕大的交易所,
88	Architect Financial Technologies, founded in 2023, offers futures in compute linked to the value of	建筑师金融技术公司成立于2023年,提供与价值挂钩的期货
89	chips in America.	美国的芯片。
90	The firm has also recently launched a platform where model makers bid to	该公司最近还推出了一个平台,让模型制作人参与其中。
91	provide inference, the useful output produced by an AI model in response to a prompt, rather than	提供推论,说明AI模型针对一个即时的而不是
92	underlying compute capacity.	基本计算能力。
93	公司创始人兼老板Brett Harrison认为
94	inference may prove easier to standardise than compute.	推论可能比计算更容易标准化。
95	Given the uncertainty surrounding the	鉴于目前情况不明朗,
96	all-new market, it is attracting risk-takers.	全新市场,它吸引了冒险者。
97	Many, including Mr Harrison, spent time in crypto.	包括Harrison先生在内的许多人在密码室呆了很长时间。
98	Once the most compute-intensive enterprise out there, which attracted financial engineers,	曾经是最计算密集型的企业吸引了金融工程师
99	trying to turn processing power into a tradable asset.	试图将处理能力转化为可交易资产。
100	Harrison先生是美国FTX公司的总裁
101	the American arm of the crypto exchange founded by Sam Bankman Freed, resigning two months before	Sam Bankman Freed创建的密码交易所的美国分支两个月前辞职
102	the parent firm's bankruptcy amid fraud in November 2022.	2022年11月母公司在欺诈中破产.
103	Luxor once primarily served Bitcoin miners.	卢克索曾经主要服务于比特币矿工.
104	One of Aneera's co-founders, Vishwa Naik, was a crypto entrepreneur.	阿内拉的联合创始人之一的维什瓦·奈克(Vishwa Naik)是一位密码企业家.
105	Corweave and some other	科瑞夫和别的
106	neoclouds also began life as crypto mining businesses.	新云也开始了隐秘采矿业的生活。
107	As opportunities grow, established	随着机会的增加,建立起来
108	financial companies are taking more of an interest.	金融公司越来越感兴趣。
109	DRW和Jump Trading,两个大型市场制造商,
110	have both invested in Silicon Data, the compute-focused research firm with which CME hopes	两者都投资于硅数据公司,这是计算重点研究公司,CME希望与该公司合作
111	to launch its futures.	以启动它的未来。
112	Each market trader has also assembled its own team to trade the new	每个市场贸易商也聚集了自己的团队来进行新的贸易
113	commodity.	商品。
114	这个月,Jump Trading公司收购了EtoMarkets公司。
115	founders will lead its efforts to break into the new industry.	创始人将领导其突破新产业的努力.
116	David Olson, Jump's investment chief,	大卫·奥尔森,跳跃的投资主管,
117	points out that the market will grow more rapidly with the right institutions,	指出市场将随着合适的机构而更快增长,
118	such as clearing houses that limit the risk of blow-ups by ensuring that contracts are honoured	例如通过确保履行合同来限制井喷风险的信息交换所
119	by both sides of a trade.	由交易双方。
120	We're a long way from that, he acknowledges, echoing many in this	他承认,我们离这个还有很长的路要走,
121	industry who think it may take years or even decades for the market to mature.	他们认为市场要成熟可能需要几年甚至几十年的时间。
122	But, he adds,	但是,他补充说,
123	we want to be part of that discussion.	我们想参与讨论。
124	His desire is unsurprising.	其欲无惊.
125	After all, with surging demand	毕竟,随着需求激增
126	for what remains a scarce commodity and with so many businesses around the world exposed to it,	因为它仍然是一种稀缺的商品世界上有这么多的商家都接触到它
127	the price is enormous.	代价是巨大的。
128	Mr Naik of Aneera, who in his pre-crypto days was an oil trader at	Aneera的Naik先生,他在临终前是一位石油贸易商
129	Reliance Industries, an Indian conglomerate, likens today's moment in the compute market	依赖工业,一个印度企业集团, 类似今天在计算市场的时刻
130	to the last days of the Seven Sisters, a cartel of Western giants that ruled the global petroleum	统治全球石油的西方巨头集团
131	trade between the late 1920s and the 1970s.	20世纪20年代末至70年代的贸易。
132	Their chokehold was broken by nationalisation	国家化打破了他们的窒息状态
133	in many exporting countries and the rise of the open oil market just about everywhere.	在许多出口国,开放石油市场的崛起几乎无处不在。
134	As governments covetously eye the AI labs and ventures like Aneera attempt to create a market	当政府贪婪地注视着AI实验室以及像Aneera这样的企业试图创造市场
135	from scratch, you do not need to squint too hard to see a similar future for compute.	从头到尾,你不需要太难看以看到类似的未来来计算。

该视频共有字幕 135 条。解锁更多字幕为会员功能，请移动到价格

下载安装