Ali Agha's FieldAI says it crossed $100M in revenue and contracts

Why it matters

FieldAI's $100 million revenue-plus-contracts claim is a test of whether robotics investors are backing a deployable software layer, not just another demo cycle.

Ali Agha's FieldAI says it has crossed $100 million in combined revenue and customer contracts, a commercial marker for a 2023-founded robotics software company trying to prove that the real money in robots may sit above the hardware layer.

The figure, reported by Business Insider on June 26, comes with an important qualifier: FieldAI is not disclosing recognized revenue, ARR, contract duration or the split between deployed work and booked commitments. Agha told Business Insider the total comes from 30 customers across Europe, Asia and the U.S., and said FieldAI has seen "very, very fast growth in the last several months."

That is still a useful signal in a robotics market full of prototypes, pilots and demo videos. FieldAI is not trying to sell a humanoid, a robot dog, a drone or an industrial rover. It is selling the software layer that lets those machines operate in messy physical environments, from mines and construction sites to factories, data centers and defense settings. FieldAI calls the system a "universal general-purpose brain" for robots and markets one autonomy stack across robots, tasks and environments.

Before founding FieldAI in 2023, Agha spent seven years at NASA's Jet Propulsion Laboratory working on autonomous systems for difficult terrain, including Mars exploration, according to Business Insider.

That background matters because FieldAI is selling autonomy for places where robots cannot depend on clean maps, predictable warehouse aisles or scripted tasks. The founder's argument is that robotics faces a data trap: a robot needs real-world data to learn, but a company needs working autonomy before it can safely deploy robots to collect that data. "You cannot deploy because you need data to have a model that works, and you need a brain to deploy to collect the data in the real world," Agha told Business Insider.

The metric is commercial proof, but not a revenue number

FieldAI's $100 million milestone should be read as a sales and pipeline metric, not as proof of $100 million in annual revenue. Business Insider describes the figure as "revenue and customer contracts." That wording matters in AI and robotics, where a booked commitment can be very different from recognized revenue and where deployment schedules, hardware availability and safety signoffs can determine when money actually converts.

FieldAI has also not named the 30 customers behind the figure. Business Insider says customers include construction firms, data center operators, defense companies and others. In August 2025, TechCrunch reported that FieldAI had raised $405 million at a $2 billion valuation; at the time, FieldAI declined to disclose customer names.

The lack of customer names does not erase the commercial signal. It does limit how far the number can be taken. FieldAI is telling the market that customers are paying or committing to pay for autonomy software in the field. FieldAI is not giving the market enough detail to judge revenue quality, contract concentration or renewal durability.

That distinction is especially important because FieldAI has already been priced like one of the companies that could define the category. Last year, Business Insider said FieldAI raised $405 million from investors including Bezos Expeditions, Laurene Powell Jobs' Emerson Collective, Khosla Ventures, and the venture arms of Nvidia (NVentures), BHP (BHP Ventures) and Intel. The round valued FieldAI at $2 billion, according to both Business Insider and TechCrunch.

The presence of Jeff Bezos (@JeffBezos)'s investment office is part of a broader capital pattern around AI that can act in the physical world. RuntimeWire noted earlier this month that Bezos's Prometheus is another large bet on AI for physical engineering, aimed at designing and manufacturing complex products. FieldAI sits on the deployment side of that same thesis: robots do not become labor unless software can make them useful outside the lab.

A software company inside a hardware race

FieldAI's strategic choice is to avoid building a single robot body. Its mission is to build a single robot brain that can generalize across robot types and environments, and its models use physics and probability to make risk-aware decisions in uncertain settings, per Business Insider and the company's technology page.

The industrial use cases are deliberately unglamorous. In construction, Business Insider reports, FieldAI-equipped robots can walk sites, take photos and update digital copies of projects. Agha said robots are well suited to that work because they can operate 24/7 and produce more detailed notes than human site walks.

That industrial-first approach is also a data strategy. Every deployed robot can collect more operational data, which FieldAI can use to improve the software and support more complex tasks. Business Insider describes FieldAI's early deployments as starting with basic work, then expanding as the software learned from the field.

Kanu Gulati, a partner at Khosla Ventures and a FieldAI investor, told Business Insider she first met FieldAI when Agha brought an early prototype to the firm's Menlo Park office. The robot navigated from the backyard to the garage, built a 3D map and found its way down a steep flight of stairs. Gulati said customer conversations in construction, mining and nuclear sites helped convince Khosla Ventures that the market could be large.

The competition is no longer theoretical

FieldAI is selling into a category that became one of venture capital's fastest-escalating robotics trades over the past year. Genesis AI emerged in July 2025 with a $105 million seed round co-led by Eclipse and Khosla Ventures to build general-purpose AI models for robots. Genesis AI is pursuing synthetic data through a proprietary physics engine, according to TechCrunch.

Skild AI reached a reported valuation above $14 billion after a $1.4 billion Series C led by SoftBank in January 2026, according to TechCrunch, which cited Bloomberg. Skild AI, founded in 2023, also builds general-purpose robotic software and foundation models that can be retrofitted to different robots and tasks.

Physical Intelligence, another robotics AI company, raised $600 million led by Alphabet's CapitalG at a $5.6 billion valuation in November 2025, according to Axios. Its work is more closely associated with vision-language-action models and robot manipulation.

The point is not that these companies are interchangeable. Their approaches differ sharply: real-world deployments, synthetic data, manipulation models, simulation tooling, humanoid tie-ins. The point is that investors have moved from funding robot bodies to funding robot cognition. In that context, FieldAI's $100 million revenue-plus-contracts claim is a bid to separate itself from foundation-model peers whose value is still mostly inferred from technical promise.

Agility Robotics offers the hardware-side counterexample. On June 24, 2026, TechCrunch reported that Agility Robotics plans to go public through a SPAC merger valuing it at roughly $2.5 billion. Business Insider also cites Agility as a company with real-world deployments, including Amazon warehouses and GXO logistics facilities.

FieldAI is making a different claim: the winning company does not need to own the robot body if it owns the autonomy layer. That is a cleaner software business if it works. It is also a hard technical promise, because general-purpose autonomy has to survive contact with dust, stairs, weather, unfinished buildings, site-specific safety rules and the ordinary chaos of industrial work.

What FieldAI still has to prove

The open question is whether FieldAI's early commercial traction compounds into a durable platform or remains a set of high-value industrial deployments that need heavy customization. The distinction will determine whether FieldAI deserves to be valued like a software platform or like a field-services-heavy robotics integrator.

FieldAI's current disclosure leaves several important facts outside the frame. FieldAI has not disclosed recognized revenue, ARR, gross margin, contract duration, named customers, headcount or the concentration of the 30 accounts behind the $100 million figure. FieldAI has not said how much of the contract base is already deployed versus scheduled for rollout.

Those gaps are normal for a private deep-tech company. They are also where the investment case lives. A robot brain that works across many bodies and sites can become an infrastructure layer for the physical economy. A robot brain that needs extensive site-by-site tuning can still become a valuable company, but not the same kind of company.

For now, FieldAI has the ingredients investors usually look for in a category-defining robotics company: a founder with field autonomy credentials, a team built from NASA JPL, DeepMind, Waymo, Tesla, Nvidia, Boston Dynamics and Amazon alumni, large industrial use cases, strategic capital and evidence that customers are moving beyond experiments. The new milestone does not prove the whole thesis. It does show that Agha has moved FieldAI's story from lab autonomy to customer budgets, which is where the robotics market gets real.