Industry

Databricks 工程师称安全短板可能压垮一批推理服务商和 neocloud

Why we picked it

作者是 Databricks 的工程师,讲的是他亲身踩过的坑:租的 GPU 裸金属服务器上居然能看到别人挖矿。如果你在选 GPU 云或推理服务商,这条能提醒你把安全隔离列入验收项,别只盯着单价。

Databricks 工程师 Yuchen Jin 在 X 上提出观点:不少推理服务提供商和 neocloud(GPU 云服务商)会因安全问题倒闭。他引用自己此前创业时的实测经历:测试多家 neocloud 时安全状况糟糕,甚至在已独占预留的裸金属 GPU 服务器上能看到他人挖矿。他同时指出 Databricks 把安全作为 AI 平台全线(AI Gateway、推理、训练)的首要事项,投入大量工程资源。

Body

这段发言的点在于:企业客户把模型权重、数据和请求都交给推理服务商,一旦多租户隔离和物理机管控不到位,风险由租户承担。裸金属服务器出现他人挖矿,说明预留实例的隔离机制形同虚设。

对采购 GPU 算力的企业来说,这条推文是一个筛选信号:选服务商时不能只比价格和可用性,安全审计、租户隔离方式同样要进评估清单。对 neocloud 创业者而言,安全能力可能成为差异化门槛。

Original · Yuchen Jin

I think many inference providers and neoclouds will collapse because of security.

At my previous startup, we tested many neoclouds. Security was so bad that I could literally see other people mining crypto on bare-metal GPU servers we reserved.

At Databricks, security is a top priority across our entire AI platform: our AI Gateway, inference, and training. A tremendous amount of effort goes into getting security right.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include neocloud, Databricks, GPU cloud, AI safety.

Source attribution: curated from Yuchen Jin (@Yuchenj_UW).