This year is proving to be a momentous one for robotics, as humanoids grow increasingly sophisticated and models move from pilot projects and lab demos toward real-world deployment. In coverage from AI Business, industry experts laid out where the sector is heading and how companies can stay ahead.
The convergence of hardware and software
The central shift is the convergence of advanced foundation models with maturing hardware, which is transforming robots from single-purpose machines into adaptive, general-purpose systems. Jan Liphardt, CEO of robotics software company OpenMind and a former Stanford professor, argues that robotics itself is not new — what is fundamentally different is that hardware for general-purpose and humanoid robots is maturing quickly, meaning low-cost, reliable platforms have finally arrived.
Software has evolved past narrowly defined functions at the same time. In Liphardt’s telling, the field has realized that large language models can sit on top of physical hardware and control it, opening a clear path to humanoid robots that remember, learn, navigate, and share skills with other robots. Falling hardware costs, rising reliability, and more capable software together give robots new levels of interactive capability, expanding use cases and accessibility. The barrier, he suggests, is broken: the industry now knows how to create general-purpose robots that people can connect with emotionally — robots that can teach, help, and adapt. What is new is not the idea but its feasibility.
The software side of this convergence is visible in recent research: vision-language-action foundation models that control varied robot bodies from camera views and text instructions, such as Ant Group’s LingBot-VLA, and world models that simulate environments for embodied training, like LingBot-World.
A push into homes
One of the clearest consequences of this convergence is a renewed push into consumer environments. Where home humanoids have long been a much-discussed possibility, analysts now expect 2026 to be the year designs move from factories toward homes. George Chowdhury, an analyst at ABI Research, expects humanoid companies to focus on consumer and home applications this year. Traditional OEMs with “physical AI” products already hold strong customer bases in logistics and manufacturing — and are developing mobile manipulators as their answer to humanoids — giving them a firm grip on commercial markets. More companies entering these markets, often using teleoperation to make products viable today, should push humanoid prices down while reassuring investors that the technology is market-ready.
More testing, more clarity
Despite the potential, robotics still has much to prove as it moves from pilots to scale. Tim Ensor, general manager of Intelligence Services at Cambridge Consultants, observes that the excitement of 2025 gave way by year’s end to a recognition that substantial work remains. He expects this year to be when major business users genuinely engage with these robots — more testing, more experimentation, and more clarity about where the challenges lie in delivering real value.
After high-profile pilots such as Figure AI’s work with BMW, logistics and manufacturing will remain the primary proving grounds. The companies best placed to benefit, in this view, are those building both mechanical and AI platforms simultaneously, while companies that fail to keep pace with the rapid proliferation of the technology risk being left behind.
Risks on the path to adoption
The route to widespread adoption is not fully clear. Chowdhury warns that security, privacy, and technological immaturity remain major hurdles: teleoperation raises surveillance concerns, robotics foundation models still lack the robustness needed for broad deployment, and integration expertise is scarce. He also flags macroeconomic exposure — humanoids and smart robots are intrinsically linked to the fortunes of AI, and a bursting AI investment bubble would leave robotics innovators highly exposed.
Regulation adds a further tension between productivity and safety, Ensor notes: robots must move fast enough to justify the investment yet be safe enough to work alongside humans. Businesses will have to decide how to get robots working faster while guaranteeing safety — a trade-off inherent to the latest generation of platforms.
The social question
For Liphardt, the remaining challenge is ultimately social rather than technical. With a clear path to humanoids in homes, hospitals, schools, and workplaces — the first technology with the potential to perform human tasks in this way — the big question is what society does with it. The prospect is both exciting and unsettling. As robots become more capable, the industry is no longer asking whether the technology will work, but how quickly businesses and society can adapt.
Limitations and what to watch
- These are expert forecasts, and humanoid timelines have repeatedly slipped in the past; predictions of a consumer breakthrough deserve particular skepticism.
- Analyst remarks are paraphrased from press coverage; original phrasing is in the linked source.
- Teleoperation-backed products blur the line between autonomy and remote human labor — buyers should ask what share of a “robot’s” work is actually automated.
- Cost, reliability, and safety data from real deployments remain scarce; business cases built today rest heavily on vendor claims.
