据报 Anthropic 签署 5170 亿美元算力合同，紧追 OpenAI 基础设施版图

My Analysis of the Text

Okay, so I'm looking at this text and immediately assessing the core information. Anthropic, a key player, has made some significant compute contract deals – a whopping $517 billion in eleven months! That's impressive, a real show of force. However, and this is crucial, it's still behind OpenAI's ambitious $750 billion plan extending to 2030. I need to keep that relative scale in mind.

There's a note on strategy, and that catches my attention. The CEO, Dario Amodei, had warned against excessively rapid investment back in early 2026. This is essential context; it suggests a measured approach, perhaps a cautious strategy that's now being re-evaluated.

Now, we see that Anthropic is pushing hard to catch up. The phrase "racing to catch up" indicates a shift, a more aggressive stance, something to note for tracking future market changes. Simultaneously, Sam Altman from OpenAI is raising a red flag about "unsustainable silliness" within the compute buildout, specifically calling out these newer cloud providers. The competition is heating up, and potential risks and unsustainable practices need to be considered.

Finally, the article source: the article from The Decoder confirms this deal, but the title's use of phrases like "reportedly" and "warning rivals about reckless risk" adds an extra layer of nuance. The market will be watching this.

Anthropic 在 11 个月内签署了价值高达 5170 亿美元的算力合同，但仍落后于 OpenAI 截至 2030 年的 7500 亿美元计划。2026 年初，CEO 达里奥·阿莫代伊（Dario Amodei）曾警告不要投资过快。如今 Anthropic 正在加速追赶，而萨姆·奥尔特曼（Sam Altman）则警告算力建设中存在“不可持续的愚蠢行为”，尤其是来自新兴云服务商的这种现象。文章《据报道，在达里奥·阿莫代伊警告对手防范鲁莽风险后，Anthropic 签署了 5170 亿美元的算力协议》首发于 The Decoder。