36Kr Reports on Investment Trends in Embodied AI

On 2025-07-15, 36Kr reported on investment trends related to Embodied AI.

Embodied AI integrates robots, computing resources and real-world data, so where capital flows can shift the center of gravity in the business. If, as 36Kr reported, the range of investors is expanding, this could affect not only research and development but also businesses supporting mass production and data collection.

The focus of this story is how to interpret the flow of funding into Embodied AI. The original article discusses the interest of investors, companies and state capital, but this article does not examine its contents in detail and instead provides context based on publicly available information. The key point is that Embodied AI is not just software; it brings together robots, control systems, computing resources and real-world data. Investment therefore needs to be assessed not only by model accuracy but also by the connections among mass production, component supply, data collection and deployment in the field. This development is part of that broader trend, and the fact that sources of capital are expanding beyond venture capital to include large companies and state capital may indicate that the field is moving from research and development toward industrial deployment. However, it is necessary to distinguish whether the original article points to investment enthusiasm or to the actual construction of supply chains. This cannot be determined without checking the original article. From a competitive standpoint, Embodied AI is unlikely to be a contest over algorithms alone. Advantages can be divided among those with computing infrastructure, those able to mass-produce robots and those able to continuously collect data from real-world settings; a one-off demonstration is unlikely to create much differentiation. Costs are also likely to be judged not only by model usage fees and robot prices but by total costs, including maintenance, retraining and component replacement after deployment. Public information shows that companies are not all adopting the same approaches or prioritizing the same applications. Three points warrant attention going forward. First, whether the technical approach to Embodied AI envisioned by the original article centers on robots or computing infrastructure. Second, whether capital is flowing into research or mass production. Third, which applications companies are targeting for monetization. See the original article for details.

For Japanese companies, it is likely to become more important to decide where to focus their strengths: robots, components, control systems or deployment in the field. In particular, if Embodied AI requires mass production and real-world data, a key question will be how to connect it with Japan’s existing manufacturing base.