# Covariant

**Source:** https://geo.sig.ai/brands/covariant  
**Vertical:** Robotics  
**Subcategory:** Robotics Foundation Model (RFM)  
**Tier:** Challenger  
**Website:** covariant.ai  
**Last Updated:** 2026-04-14

## Summary

Raised $100M+ total with $75M Series C. RFM-1 foundation model enables general-purpose robot manipulation. Partners include ABB, Ambi Robotics, AutoStore. 15+ warehouse and industrial deployments.

## Company Overview

Covariant is a robotics AI company that has trained RFM-1 (Robotics Foundation Model), a foundation model that enables robots to understand and manipulate a vast variety of objects and perform complex tasks without object-specific programming. The company has raised $100 million+ in total funding including a $75 million Series C, partnering with major robotics hardware vendors including ABB, Ambi Robotics, and AutoStore to deploy RFM-1 as the intelligence layer for warehouse picking, sorting, and fulfillment systems.

The foundation model approach solves robotics' core scalability problem: traditional industrial robots require extensive programming and calibration for each new SKU or task, making them economically viable only for high-volume, standardized operations. RFM-1 enables robots to generalize — picking an object it has never seen before by reasoning about its shape, weight, and material properties from visual observation — dramatically expanding the economic case for robotic automation in less-structured environments.

Covariant's industrial partnerships give it deployment scale that pure-software robotics AI competitors lack: by licensing RFM-1 to existing robot hardware manufacturers (rather than building its own hardware), Covariant can rapidly scale to the installed base of ABB's thousands of industrial robots globally. This asset-light model mirrors the software-defined approach that has proven successful in other hardware-adjacent AI categories.

## Frequently Asked Questions

### What does Covariant do?
Trains RFM-1 (Robotics Foundation Model) — enables robots to manipulate novel objects without object-specific programming, through visual reasoning about shape, weight, and material properties.

### How much has Covariant raised?
$100M+ total including $75M Series C. Partners: ABB, Ambi Robotics, AutoStore. 15+ warehouse and industrial deployments.

### What problem does a foundation model solve for robotics?
Traditional robots need programming for each new SKU. RFM-1 enables generalization — robots handle objects never seen before by reasoning from visual observation, expanding automation viability to unstructured environments.

### Why is the industrial partnership model strategically important?
Licensing RFM-1 to existing robot manufacturers (ABB etc.) provides access to thousands of installed robots globally without building hardware — an asset-light model that scales faster than hardware-first competitors.

### How does Covariant's RFM-1 model learn to handle new objects?
RFM-1 learns from diverse real-world manipulation experiences across Covariant's deployed robot network — each robot's interactions with new objects contribute training data that improves the shared model. When a new product enters a warehouse, RFM-1 reasons from visual observation about the object's shape, material, and weight to determine grasping strategies, transferring knowledge from similar objects already encountered across the network.

### What is Covariant's strategy for scaling through robot manufacturers?
Covariant licenses RFM-1 to established robot manufacturers like ABB, giving its AI access to tens of thousands of existing installed robots globally. Rather than building and selling its own robot hardware, Covariant provides the intelligence layer that makes existing hardware more capable. This asset-light model allows Covariant to scale its AI's impact far faster than competitors building complete robot systems.

### What warehouse deployments does Covariant have in production?
Covariant has 15+ production deployments in warehouses and distribution centers handling e-commerce fulfillment, retail distribution, and industrial parts. Through its partnership with Ambi Robotics and AutoStore integration, Covariant's AI is used in goods-to-person systems where RFM-1 handles the piece-picking step that requires generalization across diverse product types. Production deployments provide the real-world data that continuously improves RFM-1.

### How does Covariant's technical approach compare to Boston Dynamics' Atlas?
Boston Dynamics' Atlas is a mobile humanoid robot designed for general locomotion and manipulation research. Covariant focuses on manipulation intelligence — specifically picking and placing diverse objects in warehouse settings — rather than full-body mobility. These are complementary technology layers: Atlas demonstrates what physical robot capability is possible; Covariant demonstrates what AI can make fixed or mobile manipulation arms do commercially at scale.

## Tags

automation, hardware, manufacturing, b2b

---
*Data from geo.sig.ai Brand Intelligence Database. Updated 2026-04-14.*