字号·· | 护眼
CNBC

AI界最著名的“末日论者”之一将因Anthropic的IPO获利数十亿美元One of AI's most notorious 'doomers' stands to make billions in Anthropic's IPO

点「原文对照」整页切到原文,或双击某段只看那段的原文。

爱沙尼亚科技亿万富翁扬·塔林因主张人工智能可能终结人类而闻名,预计将成为Anthropic即将到来的首次公开募股(IPO)最大赢家之一。

An Estonian tech billionaire famous for his stance that AI could end humanity is expected to be one of the biggest winners in Anthropic's upcoming IPO.

塔林联合创立了Skype,后来将其出售给eBay。五年前,他领投了Anthropic的首轮融资,即1.24亿美元的A轮融资——那时该公司的Claude聊天机器人尚未上市,也未在华尔街和硅谷掀起风暴。

Jaan Tallinn co-founded Skype and later sold it to eBay. He led Anthropic's first funding round, a $124 million Series A, five years ago — well before the company's Claude chatbot was on the market and took Wall Street and Silicon Valley by storm.

Anthropic定于周四在纳斯达克上市,股票代码为ANT,这可能成为有史以来规模最大的IPO,也是人工智能实验室的首次公开亮相。

Anthropic is set to list on the Nasdaq Thursday under the ticker symbol ANT in what could be the largest IPO of all time, and the first public debut for an AI lab.

塔林的确切持股未予披露。但考虑到Anthropic预期的首发估值约为1.4万亿美元,他初始投资的回报可能高达数百亿美元。其实际回报将取决于后续融资轮次的稀释程度,以及他是否在过程中出售过股份。

Tallinn's exact holdings haven't been disclosed. But given Anthropic's expected debut valuation of roughly $1.4 trillion, the return on his initial investment could be well into the tens of billions. His exact return would depend on the level of dilution in subsequent funding rounds and whether he sold any shares along the way.

Anthropic最早的支持者通过关注人工智能生存风险的慈善圈子与公司建立联系。他们有时被称为“厄运论者”。

Anthropic's earliest backers connected with the company through philanthropic circles worried about AI's existential risk. They're sometimes described as "doomers."

2020年,塔林对CNBC表示,人类无法存活到本世纪末的概率为六分之一,引用了牛津大学教授托比·奥德的一本书。塔林还描述了一个失控的人工智能场景:一个模型被留在夜间运行,“第二天早上世界已不复存在”。尽管继续在科技领域工作,塔林仍联合创立了剑桥生存风险研究中心和未来生命研究所,后者因最早公开呼吁暂停人工智能发展而知名。

In 2020, Tallinn told CNBC that there's a one-in-six chance that humans won't survive this century, referencing a book from Oxford professor Toby Ord. Tallinn also described a runaway AI scenario where a model is left working overnight and "the world is no longer there in the morning." While he continued to work in tech, Tallinn co-founded the Cambridge Centre for the Study of Existential Risk and the Future of Life Institute, known for one of the first public calls to pause AI.

尽管包括埃隆·马斯克和苹果联合创始人史蒂夫·沃兹尼亚克在内的知名人士签署,该研究所2023年的《暂停巨型人工智能实验》公开信对行业影响甚微。

Despite high-profile signatories, including Elon Musk and Apple co-founder Steve Wozniak, the institute's "Pause Giant AI Experiments" letter in 2023 had little effect on the industry.

据知情人士透露,塔林对人工智能的看法受到研究员埃利泽·尤德考斯基的影响,后者合著了2025年出版的《如果有人构建它,所有人都会死》一书,探讨缺乏适当管控的人工智能带来的灾难性风险。

Tallinn's view on AI was influenced by researcher Eliezer Yudkowsky, who co-authored the 2025 book "If Anyone Builds It, Everyone Dies," about the catastrophic risk of AI without proper checks, according to a person familiar with the matter.

塔林(Tallinn)的估计净资产超过100亿美元,他后来与Anthropic首席执行官达里奥·阿莫代伊(Dario Amodei)有了交集。阿莫代伊曾与几位后来的联合创始人一起从OpenAI出走,理由是认为萨姆·奥尔特曼的公司没有足够认真地对待AI安全。

Tallinn, who has an estimated net worth of omore than $10 billion, later crossed paths with Anthropic CEO Dario Amodei, who had defected from OpenAI with his eventual co-founders, citing views that Sam Altman's company wasn't taking AI safety seriously enough.

据认识两人的知情人士说,阿莫代伊和塔林在有效利他主义(effective altruism)社群中有过交集。有效利他主义有时简称EA,是一种主张利用数据和证据最大化慈善影响力的哲学。虽然它没有官方成员,但据多位人士说,其硅谷人脉网络最终为Anthropic提供了早期风险投资资金中的很大一部分。

Amodei and Tallinn overlapped in the effective altruism community, according to people who know both of them. Effective altruism, sometimes known as EA, is a philosophy that aims to use data and evidence to maximize philanthropic impact. While there's no official membership, its Silicon Valley network ended up generating much of Anthropic's early venture capital money, according to multiple people.

这些人士说,塔林早期的押注是,资助一家注重安全的AI公司是控制风险的最佳方式之一。塔林还投资了DeepMind,后者后来被谷歌收购。

Tallinn's early bet was that bankrolling a safety-conscious AI company was one of the best ways to keep a lid on the risk, these people said. Tallinn also invested in DeepMind, which was later acquired by Google.

一位要求匿名的人士说,在塔林投资时,并没有人预期Anthropic“会是一项盈利的事业”。这位因坦率谈论此事而要求匿名的人士说,当时风险投资者尚未全力投入AI,而且那还是“把AI写进融资演示文稿之前的时代”。塔林拒绝就本文接受采访。

One person, who asked not to be named, said that at the time of Tallinn's investment, there was no expectation that Anthropic "would be a profitable undertaking." Venture investors hadn't yet gone all-in on AI, and it was "before the era of putting AI in your pitch deck," said the person, who asked not to be named in order to speak candidly on the matter. Tallinn declined to be interviewed for this story.

Anthropic在2021年的首轮融资中还包括Facebook联合创始人达斯汀·莫斯科维茨(Dustin Moskovitz),他创立了Open Philanthropy(现名为Coefficient Giving),该机构深深植根于有效利他主义。这家慈善机构由霍尔登·卡诺夫斯基(Holden Karnofsky)共同创立,他是Anthropic联合创始人丹妮拉·阿莫代伊(Daniela Amodei)的丈夫。卡诺夫斯基还创办了GiveWell,后者也与有效利他主义密切相关。

Anthropic's first round in 2021 also included Facebook co-founder Dustin Moskovitz, who founded Open Philanthropy — now known as Coefficient Giving — which is deeply rooted in effective altruism. The charity was co-founded by Holden Karnofsky, husband of Anthropic co-founder Daniela Amodei. Karnofsky also started GiveWell, also largely tied to effective altruism.

FTX前首席执行官萨姆·班克曼-弗里德是另一位知名的有效利他主义者,他早年曾与塔林一同投资Anthropic。班克曼-弗里德通过其现已倒闭的对冲基金阿拉米达研究参与了2022年的B轮融资。这位前首席执行官于2023年被判欺诈罪成立,判处25年监禁。他的股份在破产拍卖中被出售,价格仅为公司当前估值的一小部分。

Former FTX CEO Sam Bankman-Fried, another known effective altruist, backed Anthropic in its early days alongside Tallinn. Bankman-Fried participated in the Series B in 2022 through his now-defunct hedge fund Alameda Research. The former CEO was convicted of fraud in 2023 and sentenced to 25 years in prison. His stake was sold in bankruptcy auction at a fraction of the company's current valuation.

据Pitchbook数据,塔林的早期投资最初使Anthropic的估值达到4.61亿美元。此后其价值大幅膨胀,最近在私募市场以9650亿美元的估值融资。该实验室已从科技巨头亚马逊、Alphabet、微软和英伟达处筹集资金,并计划按IPO价格向一批科技公司出售价值260亿美元的股份。

Tallinn's early investment initially put Anthropic's price tag at $461 million, according to Pitchbook. It has ballooned in value since, and last raised money at a $965 billion valuation in private markets. The lab has raised money from tech giants Amazon, Alphabet, Microsoft and Nvidia, and plans to sell $26 billion worth of shares to a group of tech companies at the IPO price.

Anthropic的早期投资者和一些安全界人士并未对此次IPO普遍欢呼。据一位接近该群体的人士透露,许多人在私下担心,AI安全将让位于Anthropic的商业成功和更广泛的股东基础。这位人士说,AI安全界的许多人“反对此次IPO”。

Anthropic's early investors and some in the safety community aren't overwhelmingly cheering the IPO. Many worry behind the scenes that AI safety will take a backseat to Anthropic's commercial success and wider shareholder base, according to a person close to the group. Many in the AI safety community "are opposed to the IPO," this person said.

其他人则指出Anthropic独特的公司结构,该结构通过一股超级投票权股份赋予创始人广泛的自由裁量权和对公司的控制权。联合创始人将是“创始人有限责任公司”的唯一成员,该公司将持有50.1%的投票权。但Anthropic同时也是一家公益公司,这意味着除了对股东的义务之外,它还有双重使命。它必须服务于其“公司宗旨”的最佳利益,即促进“负责任地开发和维护先进AI,以造福人类长远利益”。

Others point to Anthropic's unique corporate structure that gives the founders broad discretion and control of the company through a single super-voting share. The co-founders will be sole members of "Founder LLC," which will hold 50.1% of the voting power. But Anthropic is also built as a public benefit corporation, meaning it has a dual mandate beyond just a duty to shareholders. It has to serve the best interests of its "company purpose" to facilitate the "responsible development and maintenance of advanced AI for the long-term benefit of humanity."

阅读媒体原文 ↗