SpaceX AI Compute: The Next Big Thing? #shorts

长期来看，人工智能算力领域的发展预计将与SpaceX公司紧密相关，这涵盖了其整个企业，而不仅仅是其人工智能部门，未来在轨道计算方面也蕴藏着巨大潜力。 核心论点是，SpaceX的人工智能部门将尽可能快地、积极地建设最大量的人工智能算力。他们将保留自身内部模型所需的算力，然后将多余的算力租赁给其他公司。这种租赁收入将反过来资助其人工智能算力基础设施的进一步扩张，从而形成一个持续增长并占据主导地位的良性循环。 Gavin在最近一次Hyperchange的对话中强调了一个关键点，即这一策略可能使SpaceX AI的营收最快在明年“完全超出市场普遍预期”。这种预计的营收激增是基于他们持续扩张人工智能算力、保留一部分供内部使用，并租赁出大量剩余算力的策略。重要的是，这一潜在的收入来源目前“尚未被大多数投资者纳入模型考量中”。

The long-term AI compute story is predicted to be strongly associated with SpaceX, encompassing the entire enterprise and not just its AI division, with future potential in orbital compute. The core thesis is that SpaceX AI will aggressively build as much AI compute capacity as rapidly as possible. They will retain the necessary compute for their own internal models and then lease out the surplus to other companies. This revenue from leasing will then fund further expansion of their AI compute infrastructure, creating a continuous cycle of growth and dominance. A key point highlighted by Gavin in a recent Hyperchange conversation is that this strategy could enable SpaceX AI to "absolutely obliterate consensus expectations" for revenue as soon as next year. This projected revenue surge is based on their continuous expansion of AI compute, retaining a portion for internal use, and leasing out the significant surplus. Importantly, this potential revenue stream is currently "not being modeled by most investors."