Anthropic Gets in Bed With SpaceX as the AI Race Turns Weird
Anthropic Gets in Bed With SpaceX as the AI Race Turns Weird
Anthropic 与 SpaceX 联手，AI 竞赛走向“怪异”
Anthropic and Elon Musk’s SpaceX said on Wednesday that the two entities have signed an agreement for Anthropic to use computing resources from xAI’s data center in Memphis, Tennessee. It’s the latest tie up in an industry that is scrambling to find enough computers to run complex AI software. SpaceX and xAI were previously separate companies, but the two merged earlier this year. The combined entity, also owned by Musk, is called SpaceXAI.
Anthropic 和埃隆·马斯克（Elon Musk）旗下的 SpaceX 周三宣布，双方已签署协议，Anthropic 将使用位于田纳西州孟菲斯市 xAI 数据中心的计算资源。这是该行业为寻找足够的算力来运行复杂 AI 软件而展开激烈竞争背景下的最新合作。SpaceX 和 xAI 此前是两家独立公司，但已于今年早些时候合并。合并后的实体同样由马斯克拥有，名为 SpaceXAI。
Anthropic executives made the announcement on stage at the company’s annual developer conference in San Francisco. SpaceXAI also put out a blog post sharing more details about the deal, which will see Anthropic draw power from xAI’s Colossus 1 supercomputer.
Anthropic 高管在旧金山举行的公司年度开发者大会上宣布了这一消息。SpaceXAI 也发布了一篇博客文章，分享了关于该交易的更多细节，Anthropic 将借此调用 xAI 的 Colossus 1 超级计算机的算力。
The partnership comes at a pivotal time for SpaceXAI, which is seeking to go public as soon as next month. A relationship with a leading AI lab could bolster SpaceX’s credibility as it pitches investors on the potential gold mine in establishing more data centers, including in space.
此次合作对于正寻求最早于下个月上市的 SpaceXAI 来说是一个关键时刻。与一家领先的 AI 实验室建立关系，可以增强 SpaceX 在向投资者推介其建立更多数据中心（包括在太空建立数据中心）这一潜在“金矿”时的可信度。
Notably, Space XAI said in its blog post that Anthropic has “expressed interest” in partnering to develop “orbital AI compute capacity”—essentially, data centers in space. Having Anthropic as a potential customer could help SpaceXAI boost investor confidence that there will be buyers for its super expensive supercomputing project.
值得注意的是，SpaceXAI 在博客中表示，Anthropic 已“表达了兴趣”，希望合作开发“轨道 AI 计算能力”——本质上就是太空数据中心。拥有 Anthropic 这样一个潜在客户，可以帮助 SpaceXAI 提升投资者的信心，证明其极其昂贵的超级计算项目会有买家。
Earlier this year, Musk criticized Anthropic’s AI models in a post on X, calling the company’s policies “misanthropic” and “evil,” and alleging without evidence that the AI models showed racial and sexual biases.
今年早些时候，马斯克曾在 X 上发帖批评 Anthropic 的 AI 模型，称该公司的政策是“反人类”和“邪恶的”，并指控其 AI 模型存在种族和性别偏见，但并未提供证据。
Now, he’s significantly changed his tune, writing on X that he “spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed.”
现在，他的态度发生了显著转变，他在 X 上写道：“上周我花了很多时间与 Anthropic 团队的高层成员交流，了解他们如何确保 Claude 对人类有益，并对此印象深刻。”
Musk’s SpaceXAI first broke ground on Colossus 1 on a former Electrolux site in 2024. The company claims that it is one of the largest and fastest AI supercomputers in the world. It includes roughly 220,000 Nvidia GPUs, ranging from Nvidia’s H100 and H200 to its newer GB200 chips. The company likes to boast that it built the supercomputer in 122 days.
马斯克的 SpaceXAI 于 2024 年在原伊莱克斯（Electrolux）工厂旧址上动工建设 Colossus 1。该公司声称这是世界上最大、最快的 AI 超级计算机之一。它包含约 22 万个英伟达 GPU，涵盖了从 H100、H200 到较新的 GB200 芯片。该公司乐于吹嘘他们仅用了 122 天就建成了这台超级计算机。
Emissions from gas turbines at the data center have prompted numerous complaints from local residents. Last month, environmental protesters demonstrated at a nearby SpaceX investor gathering ahead of the IPO.
数据中心燃气轮机的排放引起了当地居民的多次投诉。上个月，在 IPO 前夕，环保抗议者在附近的一场 SpaceX 投资者聚会上进行了示威。
Anthropic plans to use its newfound computing power to “directly improve capacity for Claude Pro and Claude Max subscribers,” both companies said. The deal will give Anthropic access to more than 300 megawatts of new capacity, or approximately 220,000 Nvidia GPUs. In recent months, software developers have increasingly complained about rate limits and service disruptions to Claude Code, due to the physical limits of available computing resources.
两家公司表示，Anthropic 计划利用新获得的计算能力“直接提升 Claude Pro 和 Claude Max 订阅用户的容量”。该协议将使 Anthropic 获得超过 300 兆瓦的新增容量，即大约 22 万个英伟达 GPU。近几个月来，软件开发者越来越多地抱怨 Claude Code 的速率限制和服务中断问题，原因正是现有计算资源的物理极限。
As more and more people use services like OpenAI’s Codex and Anthropic’s Claude Code—sometimes running coding programs and agentic tasks for hours on end—the services are being bogged down. According to Anthropic, the average developer is now spending at least 20 hours per week running Claude Code.
随着越来越多的人使用 OpenAI 的 Codex 和 Anthropic 的 Claude Code 等服务——有时甚至连续数小时运行编程程序和代理任务——这些服务正变得不堪重负。据 Anthropic 称，普通开发者现在每周至少花费 20 小时运行 Claude Code。
Yesterday, The Information reported that Anthropic had committed to spend $200 billion on Google’s AI cloud services and TPU chips. Anthropic has also used Amazon’s cloud computing services and AI chips since 2023. Last month, Anthropic committed “more than $100 billion over the next ten years to [Amazon] technologies.” According to the Information report, contracts with Anthropic and OpenAI now account for “more than half of the $2 trillion in backlogs at major cloud providers” like Amazon, Microsoft, and Google.
昨天，《The Information》报道称，Anthropic 已承诺在谷歌的 AI 云服务和 TPU 芯片上投入 2000 亿美元。自 2023 年以来，Anthropic 也一直在使用亚马逊的云计算服务和 AI 芯片。上个月，Anthropic 承诺“在未来十年内向 [亚马逊] 技术投入超过 1000 亿美元”。据《The Information》报道，与 Anthropic 和 OpenAI 的合同目前占亚马逊、微软和谷歌等主要云服务提供商“2 万亿美元积压订单的一半以上”。
Paresh Dave and Zoe Schiffer contributed to this report.
Paresh Dave 和 Zoe Schiffer 对本报道亦有贡献。