Robot training data has moved from a quiet infrastructure layer into one of the busiest segments tied to embodied AI, and Axis Robotics wants to position itself at the center of that shift. In an interview with PANews, founder Chris Feng said the core issue is the scale of the data gap: while GPT-3 needed about 15 million hours of human internet data, robotics may require 100 million hours or more because physical-world tasks involve space, motion, object states, and the consequences of actions in changing environments. Feng described today’s data supply stack as a tradeoff between three main sources: real-robot teleoperation, simulation, and first-person human video. Real-robot data offers the closest match to deployment hardware but is expensive, with one hour of high-quality teleoperation data costing as much as $200 by industry estimates cited in the interview. Simulation scales better and is easier to validate, though the sim-to-real gap remains a central constraint. First-person data can expand coverage across real environments at lower cost, but its value depends heavily on future robot form factors and camera placement. Axis says it has collected more than 2.2 million robot trajectories and 28,000 cumulative data hours through browser-based teleoperation. The company also uses blockchain on Base to record relationships between tasks, verified data, and contributors, aiming to support incentive design and provenance tracking. Over the next six to 12 months, Axis plans to broaden its data products, triple its community size in half a year, and raise annual recurring revenue to between $500,000 and $1 million while trying to join supplier lists at leading robot model companies.

A CoinW Research report carried by PANews says rising expectations for Unitree Technology’s IPO are acting as a fresh catalyst for the robotics sector, shifting market attention from headline-grabbing robot demos to commercialization, supply-chain depth and scalable delivery. The report says Unitree, one of China’s best-known embodied AI companies, has built a broad product lineup spanning quadruped robots, humanoid robots, robotic arms and core components, which sets it apart from companies focused mainly on technical showcases. According to the report, Unitree filed its listing application in March 2026 and received registration approval in June 2026. CoinW argues that this has pushed the discussion beyond whether a robotics company can go public and toward whether the industry itself is ready for large-scale commercial deployment. The note also highlights Unitree’s revenue figures disclosed in its prospectus, showing revenue of RMB 159 million in 2023, RMB 392 million in 2024 and RMB 1.708 billion in 2025. The report maps potential beneficiaries across the robotics chain, from reducers, ball screws and motors to sensing, force control, controllers, lightweight materials and precision manufacturing. It also compares industry paths in the U.S., Japan, South Korea and China, and says any sustained sector re-rating will still need to be backed by orders, shipments and real-world applications.

NVIDIA has open-sourced Alpamayo 2 Super, a new inference model for autonomous driving and embodied AI, as the company sharpens its push into physical AI and robotics. The model has 34 billion parameters in total, combining a 32 billion-parameter NVIDIA Cosmos 3 Super Reasoner visual-language reasoning core with a 2 billion-parameter Diffusion Action Expert. According to the company, Alpamayo 2 Super adds Chain-of-Causation reasoning, allowing it to simulate edge cases in a virtual physical world within milliseconds and derive optimal decisions. It scored 79.2 on the LingoQA autonomous driving benchmark. CEO Jensen Huang said physical AI and robotics represent the next major wave of artificial intelligence, adding that self-driving vehicles are essentially robots with four wheels. The report also mapped out Taiwan’s role in NVIDIA’s ecosystem expansion, naming TSMC for foundry work on chips including Thor and Jetson, Foxconn and Quanta for system integration and mass assembly, and several industrial PC makers for edge control hardware. Other companies cited include Solomon in 3D AI vision sensing, HIWIN and related firms in precision motion components, Mirle in pneumatic control, and Delta Electronics in lightweight power modules. The article added that the 2026 Taiwan Automation Intelligence and Robot Show, scheduled for Aug. 19, could add to market attention on the sector.

World Labs’ acquisition of SceniX is being framed by CEO Fei-Fei Li and SceniX co-founder Yunzhu Li as a move to solve one of robotics’ hardest bottlenecks: the shortage of training and evaluation data. In an a16z interview, the two said robots are the first major proving ground for “spatial intelligence,” a category World Labs sees as the next frontier for AI. Their joint plan centers on a real-to-sim-to-real stack that maps physical environments into aligned digital worlds, where robots can be trained and tested more safely, more quickly and at larger scale. Li said World Labs’ Marble model can already turn images and text into geometrically consistent worlds, while SceniX brings robotics, simulation and evaluation expertise. Yunzhu Li argued that the company is not trying to bet on a single robot body or a single model architecture. Instead, it is building infrastructure that can support different hardware types, multimodal policy models and simulation workflows. Both executives said simulation is not a substitute for real-world data, but a necessary partner to it, particularly for counterfactual reasoning, reliability testing and iteration speed. They also said near-term deployment is more likely in semi-structured settings such as warehouses, restaurants and hotels than in fully unstructured home environments.

Agibot has removed Luo Jianlan from the partner team list on its official website, adding weight to online speculation that the company’s former chief scientist may have left. Luo had previously been listed as partner, senior vice president and chief scientist, and appeared on the company’s first publicly disclosed partner roster in September 2025 alongside Deng Taihua, Peng Zhihui and Yao Maoqing. Other public-facing changes have surfaced as well: Luo’s personal webpage now only identifies him as an assistant professor at the Shanghai Institute of Advanced Intelligence, while his X profile no longer includes Agibot-related titles. The timing has drawn attention because Zhiyuan Innovation said on July 24, 2026 that it had started the process for a Hong Kong listing. Agibot and Luo had not publicly announced any personnel change as of publication. Luo joined Agibot in April 2025 as chief scientist, led the creation of the company’s Embodied Intelligence Research Center, and later became a partner and senior vice president. His past work spans robot learning, real-world reinforcement learning and embodied intelligence, with prior research roles at UC Berkeley, Google X, BAIR, and projects including SERL and HIL-SERL.

Axis Robotics said on July 27 that it had raised a $12 million seed round led by Hack VC, with participation from Nomad Capital, Pi Network Ventures and 10K Ventures. The company positions itself as infrastructure for physical AI rather than a model or hardware builder, using browser-based teleoperation to let users collect robot interaction trajectories without owning robots. Its system pairs remote simulation tasks with a Web3-based attribution layer on Base. Each approved trajectory receives a Data ID tied to a contributor wallet, submission time, quality score and task metadata, while the full data stays off-chain. Axis says this structure is meant to turn contributor output into verifiable records that can later support reward distribution, licensing or governance. The incentive stack now includes POINTS, possible fiat revenue sharing from enterprise task packages, and a token that has not yet launched. The company has disclosed community experiments, academic benchmarks and a list of partners, but key business details such as revenue-sharing ratios, contract values and repayment data remain undisclosed. That leaves a central question unresolved: whether enterprise demand can become recurring enough to support the broader incentive loop.

Google DeepMind on July 30 introduced Gemini Robotics 2, a new intelligence layer for next-generation robots built around three separate models: Gemini Robotics 2, Gemini Robotics ER 2, and Gemini Robotics On-Device 2. The company said the system gives robots whole-body control, finer manipulation, and the ability to work together across multiple machines. DeepMind also said the platform can help robots adapt to unpredictable environments and transfer learned skills across different robot bodies. In the company’s demonstration, Apptronik’s Apollo 2 humanoid robot carried out a spoken instruction by walking to a table, picking up a watering can, and placing it into a green box on a lower shelf. The release also highlighted dexterous actions such as knot tying and sealing zipper bags with a 22-degree-of-freedom five-finger hand, along with complex packaging tasks using a standard two-finger parallel gripper. Gemini Robotics ER 2 is now available in Google AI Studio and has entered private preview on the Gemini Enterprise Agent Platform, while the VLA and on-device models remain limited to early partners.

Unitree Robotics' IPO on China's STAR Market reached the 'submitted for registration' stage in just 73 days, the fastest in 2026. The company raised $618M at a $6.18B valuation, with Meituan, Tencent, and Alibaba as investors. It shipped over 5,500 humanoid robots in 2025, leading globally.