# Antioch 获 3200 万美元 A 轮，打造物理 AI 仿真测试基建

- 来源：Rohan Paul (@rohanpaul_ai)
- 发布时间：2026-09-08 23:37
- AIHOT 分数：45
- AIHOT 链接：https://aihot.news/items/cmtsu67ek01cvrob55esx2x40
- 原文链接：https://x.com/rohanpaul_ai/status/2097348553513509161

## AI 摘要

Antioch Robotics 完成 3200 万美元 A 轮融资，由 Greylock 领投，旨在让物理 AI 团队像跑软件测试套件一样测试新版本。其仿真基础设施将场景变为软件对象，可在 Isaac Sim 或 Isaac Lab 中编写测试并接入 CI/CD，使仿真成为物理 AI 的发布门禁。合作方包括 Amazon、Ring、NVIDIA Robotics 和 Nebius AI。

## 正文

Physical AI teams should be able to test a new release the way software teams run a test suite.

simulation must become trustworthy enough to act as a release gate for physical AI.

@antiochrobotics just raised a $32 mn Series A to narrow that gap.

The company is building simulation infrastructure where the scenario itself becomes a software object.

You can write a test locally in Isaac Sim or Isaac Lab, define parameter distributions, then fan that scenario out across cloud GPUs. More importantly, those simulations can be connected directly to CI/CD.

In robotics, the bug may depend on a weird combination of wind, lighting, geometry, sensor noise, human traffic, or contact dynamics. Turning that exact physical state into a repeatable test is much harder.

But with simulation, when a robot fails in deployment, you can reproduce the same failure in simulation, vary the conditions around it, and keep those tests for every release after that.

### 引用推文

> Antioch：Today, we’re announcing Antioch’s $32 million Series A, led by @GreylockVC with participation from @A_StarVC, @Category_VC, @BoxGroup, @IcehouseVenture, and ang...