Covariant
Warehouse and logistics operators seeking AI-powered robotic automation for order fulfillment and material handling.
Overview
Covariant is an American artificial intelligence and robotics technology company founded in 2017 (originally as Embodied Intelligence) by Pieter Abbeel, Peter Chen, Rocky Duan, and Tianhao Zhang. Headquartered in Emeryville, California, the company builds AI software that enables existing robot hardware to perform a wider range of warehouse and logistics tasks. Its core product, the Covariant Brain, uses a hybrid of imitation learning and reinforcement learning to power robotic arms for tasks such as picking, sorting, and depalletizing. The company has raised a total of $222 million in funding from investors including Index Ventures, Radical Ventures, and Amplify Partners. In August 2024, Amazon entered into a non-exclusive license agreement with Covariant, hired 25% of its workforce, and brought its founders on board, a deal later reported to be worth $380 million plus a $20 million final licensing payment.
Funding: Total funding of $222 million, including a $7 million seed round, $20 million Series A, $40 million Series B, $80 million Series C, and $75 million raised in 2023. Last funding valuation was $625 million.
What it does
Covariant develops the Covariant Brain, AI-driven software that powers robotic arms in warehouses and logistics centers. The software enables robots to handle a wide range of manipulation tasks—including goods-to-person picking, kitting, depalletization, item induction, and order sortation—by learning from real-world data collected from robot arms deployed in warehouses globally. The system uses deep learning and reinforcement learning to allow robots to pick and sort items from bins at rates exceeding human performance. In 2024, Covariant introduced RFM-1 (Robotics Foundation Model 1), a robotics foundation model trained on text, images, videos, robot actions, and numerical sensor readings, giving robots a human-like ability to reason and understand their environment.
Key features
- Covariant Brain AI software for robotic manipulation
- Goods-to-person picking
- Depalletization
- Item induction
- Order sortation
- RFM-1 robotics foundation model trained on multimodal data
- Hybrid imitation learning and reinforcement learning
Use cases
- Robotic order picking in warehouses
- Automated kitting and assembly preparation
- Depalletizing goods for distribution
- Item induction into automated systems
- Order sortation for logistics providers
Pricing
Covariant publishes no pricing information. There is no plan grid, no per-pick or per-robot-hour rate card, and no self-serve checkout.
Pricing is gathered from the company's public pages and may change. Check the vendor's site before buying.
Pros and cons
Strengths
- Proven deployments with major partners like ABB, Knapp, Radial, and Otto Group
- Backed by substantial funding ($222M total) from notable investors
- Technology enables robots to pick and sort items faster than humans
- Foundation model (RFM-1) trained on real-world robotics data
- Strong academic pedigree from UC Berkeley and OpenAI research
Limitations
- No public pricing; enterprise-only with no self-serve option
- Core technology and key personnel were licensed/hired by Amazon in 2024, raising questions about the company's independent future
- Whistleblower complaint alleges the Amazon deal undervalued the company and diluted investor/employee equity
- Limited to warehouse and logistics applications; not a general-purpose robotics platform
What sets it apart
- Covariant Brain is designed to work with existing robot hardware, enabling a wider range of tasks without custom engineering
- RFM-1 is a robotics foundation model trained on real-world data from warehouse robots, giving robots human-like reasoning abilities
- Hybrid approach combining imitation learning and reinforcement learning
Ecosystem
Integrations
Notable customers
Frequently asked questions
What does Covariant's AI software do?
Covariant's AI software, the Covariant Brain, powers robotic arms to perform warehouse tasks such as picking, kitting, depalletization, item induction, and order sortation. It uses deep learning and reinforcement learning to enable robots to handle a wide range of objects.
Who are Covariant's founders?
Covariant was founded in 2017 by Pieter Abbeel, Peter Chen, Rocky Duan, and Tianhao Zhang. Chen is CEO, Abbeel is president and chief scientist, and Duan is CTO.
How much funding has Covariant raised?
Covariant has raised a total of $222 million in funding, including a $7 million seed round, a $20 million Series A, a $40 million Series B, an $80 million Series C, and a $75 million round in 2023.
What is the RFM-1 model?
RFM-1 (Robotics Foundation Model 1) is a robotics foundation model launched in March 2024. It is trained on text, images, videos, robot actions, and numerical sensor readings from warehouse robots, giving robots a human-like ability to reason and understand their environment.
What happened with Amazon in 2024?
In August 2024, Amazon announced a non-exclusive license to Covariant's technology, hired 25% of Covariant's workforce, and brought the founders on board. The deal was later reported to be worth $380 million plus a $20 million final licensing payment, and has been described as a 'reverse acquihire' in a whistleblower complaint.
Sources
This profile was compiled from the company's own pages and public web research.
Last researched August 11, 2026.
Manage this listing
Own this company? Update your listing information.