Enigma Raises $71M to Bring Physical AI to Industrial Robotics

Two 26-Year-Old Hackers Just Raised $71M to Fix the Worst Part of Robots—Talking to Them The physical AI startup Enigma thinks the real barrier to ubiquitous robots isn’t intelligence. It’s that nobody has built the volume knob yet. SAN FRANCISCO — If you have ever watched a $150,000 industrial arm sit idle because an operator couldn’t figure out the teach pendant, you already understand why Enigma just landed $71 million in seed funding. The company, co-founded by two 26-year-olds who met as teenage hacking competitors and later served in Israel’s Unit 8200, emerged from stealth on July 27 with a thesis so simple it almost sounds like a joke: robots aren’t failing because they’re dumb. They’re failing because talking to them feels like editing a config file to change the volume. Index Ventures and Ribbit Capital co-led the round, joined by Conviction Partners and a cohort of individual investors from OpenAI, Anthropic, DeepMind, xAI, Cognition, and Wiz. The deal ranks among the largest seed financings ever in the physical AI sector. Not bad for a 25-person company with no product, no revenue, and a launch strategy built around letting strangers on the internet sword-fight with robots inside a hangar. “Once in a generation, a technology shift reshapes not just software, but the structure of entire industries,” CEO Jonathan Jacobi told reporters. “We believe AI’s next chapter is moving beyond chatbots and screens into systems that can understand, adapt to, and operate in the physical world.” That is the standard pitch. What makes Enigma different is where Jacobi and CTO Gal Niv chose to start. The Unusual Bet: Interfaces First, Brains Second Most robotics companies lead with capability. How dexterous is the hand? How fast does the model adapt to a new task? Enigma leads with a question that sounds more like a product design critique: “What’s the ultimate experience?” Shardul Shah, partner at Index Ventures, framed the bet bluntly. “Someone who’s an insider may start with the capability of teleoperation or dexterity, but Enigma is starting from a very different place,” he said. The founders are outsiders—not roboticists—and investors are treating that as an asset. Shah, historically known for cybersecurity bets like Wiz and Duo Security, sees the same pattern repeating in physical AI. The bottleneck isn’t what the machine can do. It’s whether a human can make it do the thing without wanting to throw the pendant out a window. Jacobi’s favorite analogy involves washing dishes. “If you had to do your dishes and spent 15 minutes explaining to a robot where to put everything, everyone reaches the point of ‘Forget it, I’ll just do it myself.’ Right now, everyone is at that point—even with the most capable models.” The observation is not theoretical. Factory floors are littered with smart operator panels nobody uses. In one automotive welding line, a team spent three days debugging an interface crash caused by a .NET Framework version mismatch. In another case, a trajectory planned perfectly in simulation failed on the real arm because of a coordinate reference buried in the teach pendant’s settings. Industrial robotics has an interface problem, and Enigma is betting it is also the adoption problem. Robots.online: A Public Laboratory with Sword Fights and Chemistry Sets To study how humans want to interact with machines, Enigma built something deliberately provocative: robots.online, a web portal that gives anyone real-time control over more than 100 proprietary robots housed in hangars in Israel and California. The robots draw with paintbrushes, engage in mock sword combat, and mix flasks of colored liquid in simple chemistry experiments. Users issue commands via text, voice, video demonstrations, or tap-drag-and-drop gestures. The experiment is live, and at press time, no usage data had been released. But the design intention is transparent. Enigma is trying to figure out which modality—or combination—produces the most reliable behavior, and which generates the most useful training data for the underlying foundation models. A Hacker News comment captured the dual sentiment swirling around the launch: “100 robots in hangars doing sword fights and chemistry? This sounds like a gimmick. But the data collection angle is actually smart—every interaction is training data for their foundation model.” Another put it more succinctly: “If they release an API for this, the ROS community will lose its mind.” The experiment also operates as a data engine at a moment when physical AI is starved for real-world interaction data. Industry estimates suggest usable embodied intelligence might require at least 10 million hours of multimodal data. Traditional collection methods—mock production lines, sensor-gloved humans, teleoperation recordings—are slow and expensive. Enigma flipped the model: open the robots to the world, let curiosity drive data generation, and refine the interface in parallel. Not everyone is convinced. On a robotics-focused subreddit, skepticism mixed with fascination. “The fact that they built 100+ robots from scratch in 11 months is either incredibly impressive or deeply concerning about quality. Probably both.” The Founders: Unit 8200, a Teenage Prodigy, and a Blank Slate Jacobi and Niv’s backstory reads like a compressed origin myth. Both 26. Both met as teenagers competing in hacking contests. Both served in Unit 8200, Israel’s signals intelligence unit that has spawned cybersecurity unicorns at a rate that defies probability. Jacobi began studying computer science and math at 13. At 17, he became Microsoft’s youngest-ever employee, recruited personally by Assaf Rappaport, who would later found Wiz and sell it to Google for $32 billion. Jacobi subsequently worked at a startup acquired by Wiz. Niv was Unit 8200’s youngest-ever operations commander. Neither has a robotics background. That blank slate is the point. The robotics industry has spent decades optimizing control theory and hardware precision. Enigma’s bet is that the next decade belongs to people who understand interfaces, data pipelines, and the strange psychology of human-machine interaction. The team includes AI lab alumni, math Olympiad winners, and multiple PhD dropouts—a composition that looks less like a traditional robotics lab and more like an early-stage AI startup that wandered into hardware by accident. The Technical Stack, in Brief Enigma is building three things simultaneously: foundation models for physical AI, a robot-agnostic software layer that works across different manufacturers’ hardware, and the human-robot interfaces being tested on robots.online. The models supposedly require a fraction of the training data that traditional approaches demand when adapting to new machines. If true, that addresses one of the main cost barriers in AI-powered robotics deployment. The robot-agnostic middleware play is worth watching. Right now, every robot manufacturer has its own proprietary stack. An integrator who wants to deploy arms from two vendors is effectively managing two separate software ecosystems. A hardware-independent intelligence layer—something like an Android for robot arms—could shift value from hardware margins to the software and data layer. Several Chinese firms, along with Google and Tencent, are exploring similar territory, though Enigma’s explicit focus on interface design as the wedge is distinct. The Market Context: $40.7 Billion in Funding and a 47% CAGR Enigma enters a market that is, by any measure, frothy. Global robotics funding hit $40.7 billion in 2025, a 74% year-over-year jump and roughly 9% of all venture dollars, per CB Insights. The physical AI market specifically—encompassing foundation models, embodied intelligence, and the software layers that connect them—was valued at around $890 million in 2025 and is projected to reach $15.2 billion by 2032, a CAGR of roughly 47%, according to MarketsandMarkets. The competitive field is already crowded. NVIDIA’s GR00T framework uses natural language prompts for robot configuration. Meta’s V-JEPA claims adaptability to new robots with under 100 hours of footage. Skild AI raised $1.4 billion in January 2026 at a valuation north of $14 billion to build a “universal robot brain.” Physical Intelligence’s π₀ model is already being used by NVIDIA in healthcare simulations and has been licensed by Chinese module maker Fibocom. NEURA Robotics closed up to $1.4 billion in June 2026 at a valuation around $7 billion, with backing from Tether, NVIDIA, Amazon, and Qualcomm. Against that backdrop, a $71 million seed round is both ambitious and relatively modest. Skild alone has raised roughly $2 billion cumulatively. Enigma’s round reflects what TNW described as “investor confidence in the team rather than the roadmap.” The business model is, in Jacobi’s own word, deliberately undefined. Privacy, Safety, and the Uncomfortable Questions Opening physical robots to public control introduces risks that are less hypothetical than one might hope. At GeekCon 2025, white-hat hackers demonstrated a full remote takeover of a commercial robot in under three minutes, propagating malicious code to an unconnected device and using it to attack a simulated human. Security researchers at Baidu have validated remote hijack vectors on multiple popular robot models—reverse-engineering device keys, compromising MQTT networks, and exploiting near-field communication links. Enigma’s robots.online experiment also sits at the intersection of data privacy debates that have roiled the AI world. Research has shown that publicly available LLMs can re-identify about a quarter of anonymized interview participants. Robots equipped with cameras and microphones, collecting interaction data in real time, magnify that risk. The company has not yet published a detailed privacy framework for the experiment, leaving open questions about how interaction data is stored, who can access it, and what downstream uses are permitted. On Reddit, one commenter raised a practical concern that cuts deeper than the theoretical risks: “The hardware cost alone for 100+ custom robots must be significant. If the interface research doesn’t yield something scalable, this just becomes an expensive art project.” That is, in essence, the wager. $71 million to find out whether the volume knob for robotics is text, voice, video, gestures, or something nobody has thought of yet. The Rappaport Connection and the Unit 8200 Network The relationship between Jacobi and Wiz founder Assaf Rappaport is not incidental. Rappaport recruited Jacobi to Microsoft when Jacobi was 17, and Jacobi later worked at a startup Wiz acquired. Investors from Wiz are listed among Enigma’s backers, though Rappaport’s personal involvement as a direct angel has not been confirmed. What is clear is that Enigma represents a migration pattern: Unit 8200 alumni, historically concentrated in cybersecurity, are now moving into physical AI. The skills that make a capable security researcher—systems thinking, adversarial analysis, comfort with complex hardware-software stacks—map surprisingly well onto the robotics interface problem. What Happens Next The robots.online experiment is now live. The company is hiring engineers in Tel Aviv and California and scaling GPU infrastructure for model training. Partnerships in healthcare, logistics, and entertainment have been mentioned but not detailed. Jacobi has declined to share specific use cases, which is either strategic opacity or genuine uncertainty. Given the company’s age—less than a year—both explanations are plausible. Index’s Shardul Shah, when pressed on the investment, returned to the outsider thesis. The robotics industry, he suggested, has been solving the wrong problem with extraordinary precision. If the interface problem is solved, the intelligence already available might suddenly become useful. If not, Enigma will have built a fascinating hangar full of sword-fighting robots and generated a lot of data trying. The experiment will answer one question eventually: when you give the internet a fleet of physical robots and ask it to teach you how to talk to them, what do you learn? For $71 million, the answer had better be more interesting than “text works fine.” One Hacker News commenter put it with characteristic bluntness: “The volume knob analogy is perfect. Current robot interfaces are like adjusting volume by editing a config file and rebooting. We need continuous, intuitive control.” The whole bet—$71 million, 100 robots, two 26-year-old Unit 8200 veterans, and a deliberately undefined business model—hinges on whether Enigma can build the knob.