Reid Hoffman 对 AI 现状的看法：xAI、SpaceX 与 Anthropic 的监管危机

Reid Hoffman on the State of AI: xAI, SpaceX, and the Anthropic Regulatory Crisis

Reid Hoffman Critiques xAI and SpaceX AI Strategies

Reid Hoffman, co-founder of LinkedIn and investor in OpenAI and Anthropic, characterizes xAI as a "complete train wreck" and asserts that SpaceX is not an AI company. Hoffman argues that SpaceX is attempting to "buy its way into relevance" through acquisitions rather than developing organic AI capabilities.

xAI's Operational Instability

Hoffman's assessment of xAI is grounded in the company's leadership instability. By May 2026, all 11 of xAI's original co-founders had departed the company. This exodus, which began in February 2026 with the resignation of operationally central co-founder Tony Wu, has led to the company undergoing its "third restart." Additionally, Hoffman notes that xAI's Grok models have consistently lagged behind competitors from OpenAI and Anthropic in benchmark performance.

SpaceX's "Roll-up" Strategy

Following SpaceX's IPO on June 12, 2026, the company acquired the AI coding tool Cursor. Hoffman views this acquisition not as a sign of strength, but as evidence of a lack of internal capability, comparing SpaceX's approach to the "roll-up" strategy of the IAC conglomerate.

Furthermore, Hoffman dismisses SpaceX's claims of AI credentials based on its compute leasing business. While SpaceX leases AI infrastructure to companies like Anthropic, Hoffman describes this as being a "premium-priced CoreWeave," asserting that providing infrastructure does not make a company an AI company.

Regulatory Risks and the Anthropic Model Suspension

Unpredictable government intervention poses a significant risk to AI companies, as evidenced by the U.S. government's June 11, 2026, export control order forcing Anthropic to pull its Fable and Mythos models from the market.

The Fable and Mythos Incident

The suspension of foreign national access to these models was triggered by Amazon CEO Andy Jassy reporting a jailbreak in the Fable 5 model. Although Anthropic was already working to address the vulnerability and had flagged security concerns, the government's response was described by Hoffman as "autocratic willy-nilly" and lacking a principled rule of law.

Asymmetric Enforcement

Hoffman expressed particular concern over the asymmetry of the government's actions, noting that Anthropic was penalized while OpenAI was not. He warns that such unpredictable regulatory behavior creates a new category of investor risk for AI companies preparing for IPOs.

The Competitive Landscape: OpenAI vs. Anthropic

Contrary to the narrative of a zero-sum "cage match," Hoffman argues there is ample room for both OpenAI and Anthropic to succeed by occupying distinct competitive lanes.

  • Anthropic: Positioned as a leader in code, with expansion into design and legal sectors.
  • OpenAI: Functioning primarily as a consumer search front-end, with a strong but under-discussed Codex coding product.

Regarding the AI coding market, Hoffman suggests that Cursor—recently acquired by SpaceX—may have already peaked, facing mounting pressure from Claude Code and Codex since early 2026.

AI's Impact on the Workforce and Gen Z

Despite data showing that AI is displacing thousands of U.S. jobs per month—with graduate unemployment rising from 3.6% in 2019 to 5.6% in 2026—Hoffman argues that the entry-level job slump is misattributed to AI.

The "Agency Mindset"

Hoffman attributes the current employment struggles of Gen Z to global turbulence, pandemic-era over-hiring, and the difficulties of implementing remote work. He encourages graduates to adopt an "agency mindset," treating AI as a tool and companion rather than a threat. He suggests that Gen Z can differentiate themselves by becoming "AI native," helping organizations transition into AI-driven entities.

Future Ventures: Manas AI and Drug Discovery

Following his departure from the Microsoft board, Hoffman is focusing on Manas AI, a venture aimed at AI-driven drug discovery. The company focuses on generating small molecule proposals, which Hoffman describes as an "AI drug discovery factory for creating monopolies," leveraging the pharmaceutical industry's existing IP framework to establish market dominance.

翻译后的正文（简体中文）

Reid Hoffman 对 AI 现状的看法：xAI、SpaceX 与 Anthropic 的监管危机

Reid Hoffman 批评 xAI 与 SpaceX 的 AI 战略

Reid Hoffman，LinkedIn 联合创始人、OpenAI 与 Anthropic 投资者，将 xAI 称为“彻底的灾难”，并断言 SpaceX 并非 AI 公司。Hoffman 认为 SpaceX 正在通过收购来“买入相关性”，而不是发展有机的 AI 能力。

xAI 的运营不稳定性

Hoffman 对 xAI 的评估基于其领导层的不稳定。到 2026 年 5 月，xAI 原始的 11 位联合创始人全部离职。这场离职潮始于 2026 年 2 月核心运营联合创始人 Tony Wu 的辞职，导致公司进入其“第三次重启”。此外，Hoffman 还指出，xAI 的 Grok 模型在基准性能上始终落后于 OpenAI 与 Anthropic 的竞争对手。

SpaceX 的“收割”策略

在 SpaceX 于 2026 年 6 月 12 日完成 IPO 后，公司收购了 AI 编码工具 Cursor。Hoffman 认为此举并非实力的体现，而是内部能力不足的证据，并将 SpaceX 的做法比作 IAC 集团的“收割”策略。

此外，Hoffman 对 SpaceX 基于计算租赁业务的 AI 资质宣称持否定态度。虽然 SpaceX 向 Anthropic 等公司租赁 AI 基础设施，但 Hoffman 形容这相当于“高价的 CoreWeave”，并指出提供基础设施并不等同于成为 AI 公司。

监管风险与 Anthropic 模型停摆

不可预测的政府干预对 AI 公司构成重大风险，这一点从美国政府于 2026 年 6 月 11 日发布的出口管制令中可见一斑，该令迫使 Anthropic 将其 Fable 与 Mythos 模型下架。

Fable 与 Mythos 事件

对这些模型的外国用户访问被暂停，是因为亚马逊 CEO Andy Jassy 报告了 Fable 5 模型的越狱漏洞。虽然 Anthropic 已在着手修复该漏洞并标记了安全风险，政府的回应被 Hoffman 形容为“专横随意”，缺乏原则性的法治。

不对称执法

Hoffman 对政府行为的不对称性表示担忧，指出 Anthropic 受罚而 OpenAI 并未受到影响。他警告说，这种不可预测的监管行为为准备 IPO 的 AI 公司创造了一类新的投资者风险。

竞争格局：OpenAI vs. Anthropic

与零和“笼中斗”叙事相反，Hoffman 认为 OpenAI 与 Anthropic 完全可以通过占据不同的竞争赛道共存并取得成功。

  • Anthropic： 以代码领域领袖定位，并向设计和法律领域扩展。
  • OpenAI： 主要作为面向消费者的搜索前端，拥有强大但鲜为人知的 Codex 编码产品。

谈及 AI 编码市场，Hoffman 认为最近被 SpaceX 收购的 Cursor 可能已经达到顶峰，自 2026 年初起便面临 Claude Code 与 Codex 的持续压力。

AI 对劳动力与 Z 世代的影响

尽管数据显示 AI 正在每月取代数千个美国岗位——毕业生失业率从 2019 年的 3.6% 上升至 2026 年的 5.6%——Hoffman 认为入门级岗位的衰退被错误地归咎于 AI。

“主体性思维”

Hoffman 将 Z 世代当前的就业困境归因于全球动荡、疫情期间的过度招聘以及远程工作实施的困难。他鼓励毕业生采用“主体性思维”，将 AI 视为工具和伙伴，而非威胁。他建议 Z 世代通过成为“AI 原住民”，帮助组织向 AI 驱动的实体转型，从而实现差异化。

未来项目：Manas AI 与药物发现

在离开 Microsoft 董事会后，Hoffman 将重点放在 Manas AI 上，这是一家致力于 AI 驱动药物发现的创业公司。该公司专注于生成小分子提案，Hoffman 将其描述为“用于创建垄断的 AI 药物发现工厂”，利用制药行业现有的知识产权框架来建立市场主导地位。

翻译后的摘要（简体中文）

投资者 Reid Hoffman 批评 xAI 为“灾难”，并称 SpaceX 的 AI 战略为“买入相关性”，同时警告美国政府对 AI 模型的不可预测干预。

翻译后的标题（简体中文）

Reid Hoffman 对 AI 现状的看法：xAI、SpaceX 与 Anthropic 的监管危机

Sources