The main body factory is up, the large model factory is down, Ant Lingbo VLA2.0 grabs the robot's "brain entrance"

Intelligent Things Author | Jiang Yu Editor | Mo Ying At present, the embodied intelligence industry is engaged in a silent war over profit margins. In the past two years, capital has poured intensively into the robot body and core components, trying to replicate the supply chain cost reduction logic of the smartphone era. According to incomplete statistics from Zhidongxi, at least 9 of the 26 unicorn companies in the industry currently valued at tens of billions of dollars focus on hardware and ontology. On the surface, Ontology Factory is continuing to drive down robot costs and compete for higher hardware profit margins through large-scale mass production, supply chain integration and self-research of core components. However, when each robot was sent to factories, supermarkets, pharmacies and experimental scenes, another more realistic problem emerged: due to the lack of a unified and reusable capability base, different ontologies, different tasks, and different scenarios still require a large number of software engineers to go to the site to perform inefficient "special adjustments." The profits saved by reducing hardware costs may be swallowed up by the high cost of non-standard software engineering in a blink of an eye. This is also the key contradiction currently faced by the embodied intelligence industry: Ontology does not want to just be a hardware foundry without a "brain". General large model giants are eager to find an entrance into the physical world. However, the emerging vertical embodied brain players sandwiched in the middle must prove that they are not outsourcing tools delivered in one time, but a basic layer that can accumulate data, models and ecology. On July 8, Ant Lingbo open sourced the new generation embodied base model LingBot-VLA 2.00, which natively supports 20 mainstream configurations and provides full-body control. In addition to open source benefiting the embodied industry and developer ecology, this is a very strategic position: Ant Lingbo is the first to make itself a standardized capability layer common to the industry, allowing different ontologies, different scenarios, and different tasks to access the same embodied brain, thus gaining an ecological position among hardware factories, large model companies, and integrators that the industry chain cannot bypass. 1. Robot hardware is reducing costs, but algorithm deployment is still stuck in the "manual era." In the boom of embodied intelligence, the cost reduction curve on the hardware side has taken the lead. Taking humanoid robots as an example, head manufacturers are using motors, reducers, sensors, structural parts, etc.