FieldAI-powered robots operate mapless autonomy across a construction and industrial site for monitoring and inspection.
Undisclosed, August 21, 2025
FieldAI announced $405 million raised across two rounds, including a $315 million tranche, driving its valuation to about $2 billion. The company develops Field Foundation Models (FFMs) that enable robots to operate without maps, GPS, or pre-defined travel paths, cutting deployment time and cost. FFMs are hardware-agnostic and run on humanoids, autonomous vehicles and other platforms. Training mixes real customer-site data with synthetic data from thousands of simulations. Early deployments focus on monitoring, surveying and inspections. The new funding will accelerate engineering, global expansion and hiring to roughly triple headcount as FieldAI scales commercial operations.
Summary lead: FieldAI, a maker of artificial intelligence software for robots, disclosed that it has secured a total of $405 million in funding from two separate rounds. The larger of the two rounds provided $315 million and was led by a group that includes Bezos Expedition, Prysm and Temasek. The new capital lifts the company’s valuation to about $2 billion, up from roughly $500 million a year earlier.
The company says the proceeds will be used to speed engineering work and expand its presence outside its current markets. Management plans to roughly double headcount by the end of the year to support product development and international growth. Reported plans include hiring across engineering and operations to scale development of locomotion and manipulation capabilities.
FieldAI develops a software platform built around a class of AI systems the company calls Field Foundation Models or FFMs. These models are designed specifically for physical robots operating in real-world places such as construction sites and factories. The company says FFMs are made to be risk-aware, to reduce the tendency of generic models to make unsafe choices, and to cut the amount of customization needed for each new robot type.
The platform is reported to run on humanoid robots, autonomous vehicles and a range of other hardware without requiring bespoke autonomy stacks for each machine. FieldAI trains FFMs with data gathered from customer sites where the software is in use and supplements that with synthetic data created from thousands of robot simulations. Those simulations are powered by an open-source simulation tool from a major graphics and AI hardware vendor.
One of the standout claims is that FFMs can operate without pre-built maps, without GPS, and without user-defined travel paths. That capability is intended to reduce the time and cost associated with deploying robots in fast-changing, unstructured environments where maps are often unavailable or quickly out of date. The company reports hundreds of deployments across industrial sites where robots are used for tasks such as monitoring construction progress and inspecting production equipment.
FieldAI’s software includes measures that estimate confidence in navigation and decision making so operators can set thresholds that prevent robots from acting when the predicted risk of error is too high. Customers are able to deploy multiple FieldAI-powered robots in the same facility and configure them to coordinate work rather than operate independently.
The largest recent investment closed earlier this month and named several lead backers including Bezos Expedition, Prysm and Temasek. The company also counts venture arms of major chipmakers among its backers. Earlier reporting indicates the firm was founded in 2016 by a former robotics technologist with a background working on space-related projects. The company has been described as headquartered in Irvine, California, and previously had a valuation of about $500 million a year earlier.
Different outlets and documents have reported closely aligned but slightly different figures for parts of the financing, with one account listing a recent round as $314 million while other accounts list $315 million. The total reported across two rounds is consistently near $405 million.
The funding push is aimed at accelerating development of autonomy that can be deployed quickly in dynamic, hazardous and changing environments. Key technical goals include improving safety, reducing deployment time, and broadening the range of tasks robots can perform without heavy site-specific setup.
FieldAI raised a total of $405 million across two funding rounds. After the latest investments, the company’s valuation is reported at about $2 billion.
The largest recent round, reported at about $315 million, listed lead investors that include Bezos Expedition, Prysm and Temasek. The company also has backing from the venture arms of major chipmakers.
Field Foundation Models are AI systems built to run on physical robots. They are designed to be risk-aware, to work across different robot types, and to reduce the need for custom autonomy software for each device.
Yes. The company reports that its models can function without pre-made site maps, without GPS, and without user-defined travel paths, making them suitable for fast-changing environments such as construction sites.
Training uses a mix of real-world data collected where the software is deployed and synthetic data produced from thousands of robot simulations. The simulations use an open-source simulation tool from a major AI hardware vendor.
Funds are earmarked to expand engineering capacity, accelerate product development across locomotion and manipulation, and grow the company’s international reach. The company plans to roughly double headcount by year end.
Reported deployments include hundreds of industrial environments, with examples such as construction site monitoring and factory equipment inspection.
| Feature | What it means | Why it matters |
|---|---|---|
| Field Foundation Models (FFMs) | AI models designed for physical robot use | Reduce customization and aim to improve safety and reliability |
| Mapless operation | Robots can function without pre-built maps or GPS | Simplifies deployments in changing or uncharted sites |
| Synthetic training data | Thousands of simulated robot runs create labeled data | Augments site data to speed model training and cover rare scenarios |
| Multi-robot coordination | Multiple robots can be deployed and set to work together | Improves throughput and reduces the need for manual oversight |
| Funding and scale-up | $405 million raised; $315 million in the lead round; valuation near $2B | Resources to expand engineering, global reach, and hiring |
West of Wolf Pack Park (university campus), August 21, 2025 News Summary Final financing closed and…
, August 21, 2025 News Summary Apex Money Lending Group has broadened its commercial real estate…
Florida, August 21, 2025 News Summary The Florida Legislature has approved HB 7031, establishing new sales…
Columbus, Ohio, August 21, 2025 News Summary Columbus, Ohio, ranks fifth in the nation for apartment…
News Summary Westchester County is making significant strides to combat the affordable housing crisis affecting…
South Korea, August 21, 2025 News Summary A team of South Korean designers has introduced Re:shell,…