Part I: What Will Probably Happen in 2026

As the new year kicks off—with well wishes, outsized resolutions, and the customary illusionary promise to finally get in shape—here at Robots, actually! we’re doing what we do best: taking a hard look at the shapes of robotics as an industry.

2026 will be a year of a few uncomfortable realizations. Some hype will deflate. Some trends will strengthen. And one chronic blind spot will remain stubbornly misunderstood and an expensive mistake to make.

Here is what we think we will see in 2026. In our next post, we will also share what we think should happen.

#5 The humanoid bubble burst is coming

Humanoid robots made for some jaw-dropping headlines last year: sleek demos, viral videos, influencer collaborations, say-what?! timelines, the whole works. Bold claims about “general-purpose labor” have cemented, sometimes unhelpfully so, bipedals and humanoid robots at the center of public imagination and popular discussions. (And we will let you pick your own take if these claims were foolishly bold or naïvely bold or VC-pressurized bold).

But in 2026, robots will have to face reality – with all its messiness, contradictions, paradoxes, unpredictability, pot holes and loopholes included. Humanoid companies will have to put up or pay up for the cost of the hype finally self-correcting (or running out of steam – again, have your pick).

This isn’t to say humanoids are going away. Far from it. Companies like Tesla, Figure AI, and Agility Robotics will continue to break ground with extraordinary advances. What will start phasing out, however, is the idea that humanoids are near-term answer for large-scale deployment. (You are not dead quite yet, dear purpose-built manipulator community. Not for some time, for those who pay real attention)

The reality is that humanoids are still

  • expensive to build and maintain
  • clunky in the real world
  • people tend to not want them and feel uncomfortable around them (minor detail)

In 2026, investors and customers alike will start focusing on the trickier points:What is the value of legs and the ROI on that investment?

#4 Dual use will proliferate and drones will dominate as a form

If there is one robotic form factor whose dominance in 2026 is almost guaranteed, for all the wrong reasons, it’s the drone. And it will not be the hobby drones you buy your kid on Amazon, or the kinds that populate last-mile delivery fantasies, but the ones driven by 19th century politics conducted by 21st century means and overseen by, well, we can say it because as anthropologists we know it for a fact, a species whose biological evolution is slow (we are not in the 19th century; not yet. Just ask your prefrontal cortex next time you have to fight your partner for the last piece of chocolate, let alone covet a continent).

Ongoing conflicts around the world have made two things excruciatingly clear. Dual-use will go from strength to strength. The robotics community, most of which was until very recently almost uniformly opposed to any kind of cooperation with the military or developing for military uses, will have to do some serious soul searching and strategy ballet in facing the trade-offs of plentiful funds in exchange for conflicted and convoluted ethics. And second, fairly inexpensive and rapidly iterated aerial systems offer an asymmetric advantage that traditional platforms struggle to counter. As a result:

  • Defense budgets will continue flowing toward drones
  • Dual use will proliferate
  • Autonomy, navigation, and perception will advance rapidly
  • Swarming and redundancy will become core design principles

How will this play out in civilian applications? How will governance models catch up to this reality? Sitting as we are in Germany, notorious for its paperwork, which is so strict where drones in (even remotely) populated environments are concerned that entire businesses exist to help you fill it out (yes, that bad), our mind boggles. We´d say bring out the popcorn, but this is actually not the kind of movie we want to see – or live in (even if clearly, we will have to).

#3 Physical AI will start catching up to AGI as the real and necessary goal

For the past few years, Artificial General Intelligence has been the salt of every AI conversation. You simply could not ignore the topic. But in 2026, the center of gravity will gain an additional focal point.

The hard problem won’t be thinking anymore—it will be doing (that is a hard problem also for humans, by the way; and let´s face it, a good number of us have a problem with erm… both. So who are we to judge).

What matters is not whether a model can reason abstractly, but whether it can:

  • Pick up unfamiliar objects
  • Navigate unpredictable environments
  • Recover from failure
  • Learn new tasks with minimal retraining and in real time

A new era is at the door: that of systems where perception, control, learning, and embodiment are inseparable. Vision–language–action models, imitation learning, and simulation-to-real transfer will increasingly matter as much, if not more, than benchmark scores.

Intelligence without a body is not much of intelligence at all. This will soon be accounted for in machines.

And let´s not forget: world models are coming.

#2 Climate robotics will continue to be the underdog nobody pays attention to, but should

Here´s a safe but sad one: climate robotics will continue to struggle, overcome by the general-purpose humanoid craze. However, the acceleration of climate change and environmental degradation will increasingly demand the deployment of robots at scale:

  • Agriculture will increasingly be a viable multiplier sector
  • Climate will spur forth even more the drone trends that defense alone does not
  • It will not be about flashy hype but about robust and reliable machines

Given the current craze around humanoids, it is likely that despite climate being the logical next frontier where talent and capital should concentrate, as it stands, the talent currently working on much-needed climate solutions continues to work without much capital.

While it´s debatable how intelligent machines are, that´s a good example of just how unintelligent humans can be.

#1 Most everyone will continue to get the key ingredient—culture—wrong

And finally, the least technical—and most important—prediction.

In 2026, most large-scale robot deployments will still struggle not because of hardware, software, or AI, but because of culture.

Robots are not deployed in neutral environments. They enter workplaces, institutions, communities, and societies. Each of these are governed not only by standards, regulations, and policies. More diffusely they enter a realm of

  • Existing norms and power structures
  • Cultural differences
  • Incentives that don’t align with automation
  • Workers who were not consulted
  • Managers who misunderstand what robots are good at
  • Ideologies and emotions, shaped by Hollywood blockbusters (and a good number of B-movies)

Deploying robots instead or alongside humans requires having built with culture and society in mind, and having protocols in place, for the robot and for the company, of what happens when the robot leaves the sci-fi of labs and enters the chaotic terrain of everyday life.

The organizations that succeed will be the ones that:

  • Design workflows around robots, not against them
  • Invest in human–robot collaboration, not replacement narratives
  • Treat deployment as change management, not installation

And the robotic companies that ace the race to the viable long-term product life (one that does not wear out when the novelty quickly fades) are the ones which think about culture from day 1.

And as we all know, for successful companies, it is day 1 every day.