← All-In

OpenAI's $150B conversion, Meta's AR glasses, Blue-collar boom, Risk of nuclear war

第2024-09-27期2024-09-27English会员内容
收听本集00:00 / 1:35:49
阅读设置
按你的习惯阅读
正文字号
下载播客精读
把这一篇,带走慢慢读

英文 · 精读 · 当前版本

下载功能还未开放

正文语言

精读 · 按主题梳理核心观点、论证与案例。

§01The reported $150B round and the conversion terms

After a skit mourning departed OpenAI executives, the hosts turn to reports of a $150 billion valuation and a raise of roughly $6 to 7 billion, with Sam Altman reportedly getting about 7%, near $10 billion. The reporting is Bloomberg's, based on ongoing investor talks. The hosts note it is contingent on ending the nonprofit structure and removing the profit cap, and nobody on the panel says the money has been wired or the documents signed.

The planned structure is a for-profit benefit corporation. The hosts explain that it is treated like a C corporation for tax and legal purposes, but directors owe duties to a stated mission as well as shareholders. Reuters is cited saying the nonprofit would remain a minority shareholder, with the size of its stake still open, and that the 100x return cap for investors would go.

§02The bull case: capital widens the moat

Friedberg's bull case is that new capital extends OpenAI's lead in model performance and infrastructure across consumer, enterprise and API products, citing the o1 model, voice and the unreleased Sora, in a market worth multiple trillions of dollars.

On the numbers, a chart shows revenue of $28 million in 2022 and $2 billion in 2023, with a reported $3.4 billion run rate, about 50 times revenue. Sacks says he heard $4 to 6 billion, which would be roughly 30 times, or 15 times next year if growth hits 100%. He calls the valuation defensible, but asks whether any moat is durable, pointing to Meta's Llama 3.2 voice release.

§03The bear case: four pressures on OpenAI

Chamath argues that open-source models at least as good as OpenAI's could push the economic value of models toward zero. He adds that Meta and Google can embed assistants in their own front doors, that data may run out so rivals race on costly synthetic data, and that senior churn is hard to square with a company said to be on a rocket path.

It reads a little bit like a telenovela.

好对话,值得读完整。

免费试读到此结束。开通书架会员,继续阅读全文,解锁中英对照与电子书下载。