RoboForce
Last reviewed
May 9, 2026
Sources
10 citations
Review status
Source-backed
Revision
v6 · 2,638 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
May 9, 2026
Sources
10 citations
Review status
Source-backed
Revision
v6 · 2,638 words
Add missing citations, update stale details, or suggest a clearer explanation.
| RoboForce | |
|---|---|
| General information | |
| Full name | RoboForce, Inc. |
| Founded | 2023 |
| Founders | Leo Ma (CEO), Calvin Zhou (Co-Founder, VP Engineering) |
| Headquarters | Milpitas, California, United States |
| Industry | Robotics, Artificial intelligence |
| Products | TITAN mobile manipulator |
| Total funding | $67 million |
| Orders | 11,000+ (via letters of intent) |
| Website | roboforce.ai |
RoboForce is an American robotics company based in Milpitas, California, that develops AI-powered mobile manipulator robots for demanding industrial environments. Founded in 2023 by Leo Ma and Calvin Zhou, the company builds what it calls "Physical AI-powered Robo-Labor" designed to take on tedious, heavy, and hazardous tasks that humans should not have to perform. Its flagship product is TITAN, a dual-armed mobile manipulator designed for outdoor industrial work. RoboForce was spotlighted by NVIDIA CEO Jensen Huang at the GTC keynote in October 2025 for its breakthroughs in Physical AI.[1][2]
The company describes its mission as "Solving Human Needs, Beyond Emulating Human Forms," a deliberate philosophical contrast with humanoid robot developers that prioritize human-shaped form factors. Instead of pursuing bipedal humanoids, RoboForce builds purpose-built mobile manipulators optimized for the real-world geometry of industrial worksites such as solar farms, container yards, and mines. By March 2026, the company had raised a total of $67 million across three rounds and reported more than 11,000 robot orders through non-binding letters of intent, with certified deployment partners spanning more than 12 countries.[2][3]
RoboForce was founded in 2023 in Silicon Valley by a group of robotics and autonomy veterans recruited from Carnegie Mellon University, the University of Michigan, Amazon Robotics, Google, Waymo, Cruise, Tesla Robotics, ABB, and Apple. Subsequent reporting on the company's $52 million round added Tsinghua University and Ubtech to the list of institutions and employers represented on the engineering team.[1][3]
The two named co-founders bring complementary backgrounds in industrial automation and autonomous vehicle engineering:
The broader engineering organization includes contributors from CMU Robotics, Michigan Robotics, Amazon Robotics, Tesla Robotics, Cruise, Waymo, Google, Apple, Tsinghua University, ABB, and Ubtech, giving the team unusually deep representation across both academic robotics research and large-scale autonomy deployment.[1][3]
RoboForce raised capital across three publicly disclosed rounds between January 2025 and March 2026. The cumulative total reached $67 million.
| Date | Amount | Round | Lead and notable investors |
|---|---|---|---|
| January 2025 | $10 million | Early-stage / Series A | Mix of new and existing investors |
| May 2025 | $5 million | Extension | New and existing investors (announced alongside TITAN unveiling) |
| March 2026 | $52 million | Oversubscribed round | YZi Labs (lead), Jerry Yang, Myron Scholes, Gary Rieschel, Carnegie Mellon University |
| Total | $67 million |
The January 2025 round of $10 million was used to scale early prototypes and recruit AI and robotics talent. The May 2025 extension of $5 million coincided with the public introduction of the TITAN mobile manipulator and supported expansion of the company's Silicon Valley headquarters with advanced development and testing facilities.[1][5]
The March 2026 round of $52 million was led by YZi Labs, a $10 billion fund. Notable individual investors include Jerry Yang (co-founder and former CEO of Yahoo), Myron Scholes (Nobel Laureate in Economic Sciences), and Gary Rieschel (founding partner of Qiming Venture Partners and previously a co-founder at SoftBank Venture Capital). Carnegie Mellon University also participated, reflecting RoboForce's continued ties to the institution where Leo Ma earned his graduate degree. The round was reported as oversubscribed and was announced on March 16, 2026.[2][3][6]
As part of the deal, Ella Zhang, Managing Partner and Head of YZi Labs, joined RoboForce's board of directors. YZi Labs publicly identified four reasons it led the round: founder execution (citing Ma's track record deploying autonomous systems across hundreds of industrial sites), engineering depth from the cross-disciplinary team, market demand created by labor shortages in solar, logistics, and mining, and a technology inflection in which the convergence of edge computing, simulation, and reasoning models has finally bridged what the firm calls the "demo-to-real" gap.[6]
In its investment note, YZi Labs cited supporting market data, including a figure that 53 gigawatts of U.S. solar projects were delayed in 2024 due to worker shortages. Jing Xiong, AI Investment Director at YZi Labs, summarized the rationale by stating: "These robots are doing the work humans were never meant to endure."[6]
On October 28, 2025, NVIDIA CEO Jensen Huang spotlighted RoboForce during his GTC keynote presentation in Washington, D.C. The keynote, focused on the frontier of AI and robotics, highlighted TITAN among a select group of companies advancing what NVIDIA calls Physical AI. In his remarks, Huang framed the broader trend by stating: "AI is transforming the world's factories into intelligent thinking machines, the engines of a new industrial revolution."[7]
Responding to the recognition, Leo Ma said: "We're honored to be recognized by Jensen and the NVIDIA team for our contribution to the global movement of Physical AI. Our company's mission is to elevate humans beyond dull, dirty, and dangerous work."[7]
The public association with NVIDIA's keynote significantly raised RoboForce's profile and was cited by YZi Labs as a contributing factor in the oversubscribed funding round that followed five months later.[6]
At the core of RoboForce's product is RF-Net, a proprietary 3D foundation model developed in-house. RF-Net is designed to give a robot advanced spatial understanding of its environment and to translate that understanding into manipulation actions with millimeter-level precision. The model serves as the perception, planning, and control backbone of the TITAN mobile manipulator and is iteratively improved through what the company calls a "Physical AI data flywheel": deployed robots in the field generate operational telemetry and task performance data, which is fed back into training pipelines to refine motion control, object handling, and task sequencing.[2][6]
The company's training approach combines real-world operational data from deployed units with large volumes of synthetic data produced in simulation, allowing the model to encounter many more task variations during training than would be feasible in pure physical testing. This is a central reason RoboForce describes its system as being able to operate autonomously in "complex, unstructured environments" rather than only in tightly choreographed factory cells.[7][8]
RoboForce is one of NVIDIA's named Physical AI partners and has integrated the full breadth of NVIDIA's robotics stack into its development pipeline. The company has publicly described its use of the following NVIDIA components:[3][6][8]
| NVIDIA component | Role at RoboForce |
|---|---|
| Jetson Thor | Edge computing and onboard inference for TITAN |
| Isaac Sim | High-fidelity 3D simulation for training and policy evaluation |
| Isaac Lab | Open robot learning framework for foundation model training |
| Cosmos | Synthetic data generation at scale |
| OSMO | Cloud-to-edge orchestration for fleet operations |
Jetson Thor provides the on-robot compute capable of running RF-Net inference in real time aboard each TITAN unit. Isaac Sim and Isaac Lab handle simulation-based training, while Cosmos generates large volumes of synthetic data covering edge cases that would be impractical to collect in the field. OSMO ties the system together by orchestrating model rollouts and data collection from cloud back-end to deployed edge units. Together, the stack supports the company's data flywheel and is positioned by both RoboForce and YZi Labs as a key reason TITAN can move from pilot to commercial deployment in less time than legacy industrial robots.[6][8]
RoboForce uses the term "Domain Intelligence" to describe its approach of co-designing hardware and AI together rather than building a general-purpose body and bolting software onto it. In practice this means the TITAN platform offers different base configurations (wheeled and tracked) suited to specific environments, while the manipulator arms, sensors, and software policies are tuned around the five primitive operations that dominate the company's target tasks. The result, according to RoboForce, is a robot that is both flexible enough to learn new tasks and reliable enough to run an eight-hour outdoor shift without supervision.[1][5]
TITAN is RoboForce's flagship mobile manipulator robot, designed for production-ready deployment in demanding outdoor environments. It was publicly introduced in May 2025 and showcased during NVIDIA's GTC keynote in October 2025.[1][7]
| Specification | Value |
|---|---|
| Configuration | Dual-armed mobile manipulator |
| Base options | Wheeled or tracked |
| Runtime | 8 hours of production operation |
| Dual-arm payload | 40 kg (88 lb) |
| Arm reach | 1,100 mm (43.3 in) |
| Precision | 1 mm (millimeter-level) |
| AI model | RF-Net (3D foundation model) |
| Onboard compute | NVIDIA Jetson Thor |
| Primitive capabilities | Pick, place, press, twist, connect |
TITAN is powered by RF-Net and delivers five primitive capabilities (pick, place, press, twist, and connect) that RoboForce identifies as the building blocks for the majority of industrial manual labor. The 1 mm precision figure is positioned for tasks such as fastening, electrical connection, and panel alignment, while the 40 kg dual-arm payload allows TITAN to handle objects such as solar panels and shipping components without specialized lifting equipment. The eight-hour runtime is intended to match a standard outdoor work shift.[1][5][9]
The wheeled and tracked base options allow TITAN to operate on improved surfaces such as data center floors and warehouse aisles as well as unimproved terrain such as solar farms, mining sites, and ports.[1][5]
RoboForce targets six primary industries with TITAN, all of which share characteristics of outdoor or rugged environments, persistent labor shortages, and tasks that are physically taxing or dangerous for humans:[2][3]
| Industry | Applications |
|---|---|
| Utility-scale solar | Outdoor construction, panel installation, inspection, and maintenance |
| Data centers | Equipment handling, AI infrastructure operations, and facility maintenance |
| Mining | Material handling and operations in hazardous environments |
| Shipping | Container yard, port, and cargo operations |
| Manufacturing | Assembly, repetitive heavy-duty workflows, and material transport |
| Logistics | Warehouse operations, fulfillment, and yard logistics |
In earlier company materials and interviews, Leo Ma also identified space operations as a long-term target sector, alongside the six core industries above.[1][9]
When TITAN was first announced in May 2025, RoboForce shared a customer-facing value proposition that compared its robot to traditional industrial labor on three axes:[9]
| Metric | RoboForce claim |
|---|---|
| Cost savings | Approximately 66% (described as $300,000 of value for a $100,000 investment) |
| Deployment speed | About 90% faster (4 weeks of robot deployment versus 52 weeks of human hiring) |
| Workforce uncertainty | About 90% reduction in workforce management uncertainty |
These figures are RoboForce's own claims as presented at the TITAN launch and are not independently audited.[9]
By the time of the March 2026 funding announcement, RoboForce reported it had secured more than 11,000 robot orders through non-binding letters of intent and had certified early partners in more than 12 countries. The company stated that initial pilot deployments were already active in the energy sector, with expansion underway into data centers, shipping, logistics, mining, and manufacturing. RoboForce described 2026 as the year it would convert pilot programs (described internally as "robot internships" in solar) into broader commercial deployments.[2][3][7]
RoboForce positions its category as "Robo-Labor" rather than humanoid robotics. The distinction is intentional: while many high-profile competitors in the humanoid robots space build bipedal, human-shaped systems, RoboForce argues that real industrial environments rarely reward human form. Solar farms, container yards, and mining sites favor wheeled or tracked bases that can carry heavier batteries and payloads, with arms that reach farther and fail more gracefully than bipedal balancing systems. This framing is captured in the company's slogan "Solving Human Needs, Beyond Emulating Human Forms."[5][8]
The message has been echoed by Leo Ma, who has framed the company's core thesis around "work that is dull, dirty, and dangerous" and the idea that "Robo-Labor is essential" for those tasks while humans should move toward higher-value roles. This language has been picked up in coverage of the $52 million round and the GTC keynote.[3][6][7]
The company and its investors regularly cite labor scarcity as the central market driver. YZi Labs has highlighted persistent labor gaps in renewable energy, data centers, shipping, mining, and manufacturing, and pointed to the figure that 53 GW of U.S. solar projects were delayed in 2024 because of worker shortages. RoboForce frames TITAN as an answer to that bottleneck rather than a replacement for skilled trades, with deployments concentrated in tasks that are difficult to staff at scale.[6]
The stated near-term commercial strategy is to convert the company's large book of letters of intent into recurring revenue through staged pilot programs. RoboForce describes this as moving from "robot internships" in early customer environments to multi-unit production deployments. Capital from the March 2026 round is being directed at three pillars to support that conversion: continued advancement of the robot foundation model and the AI data flywheel, manufacturing scale-up and global supply chain expansion, and commercial deployment driving revenue growth.[3][6]
Key public milestones for the company include:
| Date | Event |
|---|---|
| 2023 | RoboForce founded in Silicon Valley by Leo Ma and Calvin Zhou |
| January 2025 | $10 million early-stage funding round announced |
| May 2025 | TITAN mobile manipulator publicly introduced; $5 million extension funding announced; total funding reaches $15 million |
| October 28, 2025 | RoboForce featured in NVIDIA CEO Jensen Huang's GTC keynote in Washington, D.C. |
| March 16, 2026 | $52 million oversubscribed round announced; total funding reaches $67 million; Ella Zhang of YZi Labs joins board |