On September 30, 2026, TechCrunch AI reported on Destro AI, a startup whose stated competitive edge is that it does not build robots at all. The company's pitch, quoted in the piece, is blunt: one of the biggest reasons it wins against robotics companies is that it is not one. Its product is described as a coordination
TechCrunch AI published the report on September 30, 2026, at 16:00, framing it around Destro AI's differentiation strategy rather than its hardware. The line the company gave the outlet is the spine of the story: one of the biggest reasons it beats robotics companies is that it is not one. That sentence draws the company's self-defined boundary clearly. It sits between the machine and the human operator, selling coordination, not a chassis.
As described, the product targets a specific failure mode: humans and machines working from different pictures of the same job. An operator states a goal in natural language or through an existing workflow; the system decomposes that
The claim lands at a moment when hardware barriers in robotics are flattening fast. Over three years, supply chains for joint modules, dexterous hands, force sensors and batteries matured to the point where building a robot became less interesting than building one cheaply. Hardware alone is increasingly hard to defend as a moat. Capital flows reflect that: Skild AI, a pure software play, raised a $300 million Series A in July 2024 at roughly a $1.5 billion valuation, while Physical Intelligence raised about $400 million in November of the same year at r
From a technical standpoint, Destro AI is betting on the gap between task semantics and motion control. When a person says move this crate to line three, the machine faces five separate problems: object recognition, path planning, grasp pose, safety boundaries and failure recovery. Ambiguity in any one of them halts execution. A coordination layer compresses fuzzy intent into a verifiable sequence of steps and asks for human takeover before things break. There is precedent for the value of that position: in industrial automation, the durable margins sat with PLCs, SCADA and MES software, not with any single machine model.
On the business side, not building the body produces a structurally different balance sheet. Hardware makers carry inventory, spares, depreciation and field service teams. A software coordination layer has near-zero marginal cost and can be reused across brands and form factors, so one scheduling logic can serve quadrupeds, wheeled platforms and arms at once. Public markets and private investors have already voted. Skild AI and Physical Intelligence reached billion-dollar valuations in early rounds on software-first stories, which suggests selling the brain without the body is the easier pitch right now.
For the industry, if the coordination layer standardizes, hardware makers slide down the value chain, repeating the PC pattern where assemblers earned thin margins while operating systems and chips captured the profit. Liu Gong sees direct pressure on Chinese robot body manufacturers: once differentiation collapses to cost and delivery speed, price wars become close to inevitable. The narrative has a soft spot, though. Coordination quality depends on deep knowledge of a body's kinematics and sensor behavior, so a pure software team often needs hardware vendors to open low-level interfaces, which hands bargaining power back to the other side. The real test is intervention rate. Customers pay for fewer stoppages and less rework, and if escalations to humans stay frequent, the software premium does not hold.
- Industry base: the International Federation of Robotics reported roughly 541,000 industrial robots installed worldwide in 2023, with China accounting for about 51 percent of new installations.
- Comparable valuations: Skild AI raised a $300 million Series A in July 2024 at about a $1.5 billion valuation; Physical Intelligence raised roughly $400 million in November 2024 at about a $2.4 billion valuation, both on software-first models.
- Hardware-side contrast: Figure AI closed a $675 million Series B in February 2024 at about a $2.6 billion valuation,
For the industry, the next three months will most likely bring narrative catch-up: more robotics companies will lead with software and coordination language, downplay body specs, and move deployment time and intervention rate onto their homepages. The decisive variable over the coming year is who owns the standard. If independent software vendors supply the coordination layer, hardware makers will be pushed to open low-level interfaces and profit shifts toward software. If leading body manufacturers build the loop in-house, the window for standalone soft
Questions worth exploring:
- Whether the coordination layer becomes the robotics value chain's equivalent of an operating system, capturing the most pricing power
- How the interface-access standoff between software-first robot companies and hardware vendors will shape bargaining power on both sides
- Whether field intervention rate can become a common acceptance metric for robot deployment and force the industry to publish data
This item comes from TechCrunch AI (published 2026-09-30 16:00:00). Wenfeng Radar provides a Chinese/English summary and commentary on overseas AI developments; the original content is owned by its author.
Daily aggregation of overseas AI developments with in-depth analysis. Bookmark us to never miss an important signal; back to homepage.