矽谷投资人正愈来愈多地同时押注互相竞争的 AI 新创，打破了长期以来创投圈的禁忌；在 OpenAI 与 Anthropic 追求史上最大规模的 2 轮融资之际，这种做法更为明显。预计 Anthropic 最新一轮融资将募得超过 $20 billion，而 OpenAI 正寻求创纪录的 $100 billion，将两家公司推向一个日益重叠的「资金雄厚金主」池。这种动态反映出自至少 2021 年以来，AI 的资本密集度（晶片、资料中心、人才）扩张之快；当年 Sequoia 首次投资 OpenAI，之后也在 Anthropic 周边有所参与。

目前有数家大型机构正同时站在两边：Sequoia 与 Altimeter 预计会出现在 Anthropic 这轮融资中；据报导，Altimeter 即使曾称 OpenAI 是其最大押注，仍将投资超过 $200 million。包含 Amazon、Nvidia、Microsoft 在内的多家大型科技公司都曾讨论同时投资两者；此外，外界报导 Amazon 正洽谈投资 OpenAI，金额最高可达 $50 billion。Blackstone 已是 OpenAI 的投资人，并正洽谈将其对 Anthropic 的持股提高到 $1 billion。来自阿布达比的 MGX（与 OpenAI 及 Stargate 资料中心专案相关）也在洽谈向 Anthropic 投资数亿美元，等同于在包含 xAI 在内的 3 家领先模型公司之间分散下注。其他重叠还包括 JPMorgan 的成长股权部门与 Iconiq（透过对 Statsig 的收购而取得对 OpenAI 的曝险）；而随著规范松动，Andreessen Horowitz、SV Angel 等机构也已投资多家顶尖实验室。

被点名的主要风险是资讯外泄与结构性利益冲突，特别是当投资人持股大到足以取得资讯权利时；据称 OpenAI 只会在投资金额超过某个门槛时分享特定财务资料。投资机构会尝试用不同合伙人与「无形隔离墙」来降低风险；部分新创也会考虑 NDA 或其他法律保障，但创办人仍可能反对，因为投资竞争对手可能损害关系。并非所有投资人都参与：Khosla Ventures 与 Thrive Capital 已避开 Anthropic；其公开的「serial monogamist」说法与新兴的多重下注策略形成对比。Altimeter 的创办人则在 2024 的贴文中主张，公司是在市场上「win or lose on the field」。随著 AI 产品界线变得模糊（消费者 vs 开发者），冲突可能更难被忽视，这也呼应了相邻产业的模式：当估值与持股规模庞大时，例如 Harvey 估值 $8 billion、Abridge 估值 $5.3 billion，且曾有报导指出同一家公司内部的基金协调问题。

Silicon Valley investors are increasingly backing competing AI startups, breaking a long-standing venture taboo as OpenAI and Anthropic pursue 2 of the largest funding rounds in history. Anthropic’s latest financing is expected to raise more than $20 billion, while OpenAI is seeking a record $100 billion, pushing both companies toward an overlapping pool of deep-pocketed backers. The dynamic reflects how rapidly AI’s capital intensity (chips, data centers, talent) has expanded since at least 2021, when Sequoia first backed OpenAI and later also became involved around Anthropic.

Several major firms and institutions are now positioned on both sides: Sequoia and Altimeter are expected in Anthropic’s round, with Altimeter reportedly investing more than $200 million even after calling OpenAI its biggest bet. Big Tech names including Amazon, Nvidia, and Microsoft have discussed investing in both; Amazon has been reported in talks to invest up to $50 billion in OpenAI. Blackstone, already an OpenAI investor, is in talks to increase its Anthropic stake to $1 billion. Abu Dhabi’s MGX, tied to OpenAI and the Stargate data center project, is also in talks to invest hundreds of millions in Anthropic, effectively placing bets across 3 leading model makers including xAI. Other overlaps include JPMorgan’s growth equity arm and Iconiq (which gained exposure to OpenAI via the Statsig acquisition), while firms like Andreessen Horowitz and SV Angel have backed multiple top labs as norms loosen.

The main risk cited is information leakage and structural conflicts, especially when investors hold large enough stakes to obtain information rights; OpenAI reportedly shares certain financial data only above a threshold investment amount. Firms try to mitigate this with separate partners and “invisible walls,” and some startups consider NDAs or other legal safeguards, but founders may still object because backing a competitor can damage relationships. Not all investors participate: Khosla Ventures and Thrive Capital have avoided Anthropic, with public “serial monogamist” rhetoric contrasting the new multi-bet approach; Altimeter’s founder argued in a 2024 post that companies “win or lose on the field.” As AI products blur market boundaries (consumer vs developer), conflicts may become harder to dismiss, mirroring patterns in adjacent sectors where valuations and stakes are large, such as Harvey at $8 billion and Abridge at $5.3 billion, and where internal fund coordination issues have been reported within a single firm.