Frontier signal stream

Figure disclosed Index, a Figure-exclusive data-collection pipeline it had been running in stealth for four months to build what it calls the world's largest and most diverse physical dataset. Disclosed metrics: over 264,000 app downloads across 100+ countries, over 44,000 weekly active users, over 16 million videos uploaded by its network of "Creators", uploads running at 30 minutes of video every second (which Figure characterises as 4.9 years of human work uploaded per day), and $15m already paid out to contributors. Figure states it is "on a path to 100x" and is "committed to spend over $1B the next 12 months on data and compute". The argument is that generalisation is a data problem and that the data needed to train a general-purpose robot does not exist on the internet — it has to be sampled from the real world — so Figure is buying that distribution directly rather than scraping it. Why it matters: a stated $1bn+ 12-month data-and-compute commitment is a capital-scale disclosure from a private company, and it reframes Figure's moat as a proprietary data pipeline rather than hardware alone. All figures are company-stated and unaudited.

Figure AI unveils Index, a global dataset collected from real-world environments via an app with 264,000 downloads across 100+ countries, to train general purpose robots.

Boston Dynamics' Atlas performed football-inspired goal celebrations and delivered the ceremonial match ball to the referee at halftime of Brazil's round-of-16 match against Norway. Hyundai Motor Group described it as the first time its robotics had been integrated into a live match environment.

Hyundai Motor deployed Boston Dynamics' Atlas in a halftime activation during the FIFA World Cup 2026 Round of 16 match at New York/New Jersey Stadium on 5 July 2026, with the robot delivering the ceremonial match ball. Hyundai is the tournament's Official Robotics Partner and describes it as the first robotics-powered halftime activation and the first integration of a humanoid robot into a live World Cup match environment.

Apptronik announced the opening of a newly expanded Robot Park in Austin, Texas — a nearly 90,000-square-foot flagship data-collection and training facility for humanoid robots — and simultaneously unveiled Apollo 2, the current version of its humanoid platform, in both bipedal and wheeled-base configurations. Fleets of Apollo 2 are described as already operational across Robot Park and at customer and partner sites, with similar data-collection workflows deployed at research partner Google DeepMind and at customers Mercedes-Benz and GXO. The data collected feeds Gemini Robotics, Google DeepMind's foundational robotics models, under an existing research partnership; Apptronik frames Robot Park, Apollo 2 and that partnership as a single continuous learning loop. Two forward-looking details matter for the thesis: the wheeled-base configuration is deliberately designed to conform to existing industrial mobile-robot safety standards so it can drop into current customer operations, and the company states that everything learned via Apollo 2 is feeding development of Apollo 3, named as its next-generation commercial product. CEO Jeff Cardenas positions this as a move away from demos toward daily on-the-job work. Note that Apollo 2 is not new hardware — the release states it has been 'the workhorse behind Robot Park for more than a year already' — so this is a formal platform unveiling and capacity expansion rather than a first reveal. No financial terms, fleet counts or facility cost were disclosed.

Boston Dynamics announced plans to transform a 323,000 square foot facility at 1601 Trapelo Road in Waltham, Massachusetts into a robotics and AI centre expanding advanced manufacturing and AI capability, with construction expected to begin before the end of 2026. Regional press reported the investment at about $100m and roughly 1,250 new jobs.

Collaborative Robotics announced the second-generation Proxie mobile robot, adding autotasking — autonomous identification of which work to perform — along with two-armed mobile manipulation and greater payload capacity. The company positioned it as production-tested physical AI for real operations rather than a research platform.

RLWRLD introduced RLDX-1, a foundation model enabling humanoid robots to perform contact-rich tasks like grasping and pouring, with benchmark results presented at Dexterity Night in SF.

Sanctuary AI announced an updated strategy to deploy its Physical AI on existing and next-generation industrial robotic systems rather than solely on its own humanoid hardware. It reported a 99.5%+ task success rate at a 2.54-second cycle time on a complex wire-plugging production task, validated against a global Tier 1 automotive supplier's live production benchmarks. This is a repositioning from humanoid hardware developer toward a software layer running on third-party industrial hardware, and it arrived one day after a change of chief executive.

At GTC Taipei NVIDIA announced the Isaac GR00T Reference Humanoid Robot, its first open humanoid reference design. It pairs a Unitree H2 chassis standing nearly six feet and weighing 150 pounds with 31 degrees of freedom, dual Sharpa Wave tactile five-finger hands adding 22 degrees of freedom for 75 in total, and NVIDIA Jetson AGX Thor T5000 onboard compute delivering 2,070 FP4 teraflops with a 14-core Arm CPU and 128GB of unified memory. Ai2, ETH Zurich, Stanford Robotics Center and UC San Diego's Advanced Robotics and Controls Laboratory will use the design, with availability from Unitree Robotics in late 2026.