Robotics Predictions for 2027: 6 Specific Calls

Predictions are cheap when they are vague. Here are six specific calls for where physical AI and robotics go in 2027, the kind you can check against reality later. They are drawn from where the field's builders and investors are actually placing their bets.

1. The humanoid field starts to thin out

The money that flooded into humanoids buys runway, not results, and 2027 is when the gap between demo and deployment starts to sort winners from stories. Expect the first real consolidation: a few humanoid companies finding a repeatable job, and others quietly running out of narrative.

2. Driver-out freight moves from pilot to lane

Self-driving trucks have been "almost there" for years. This call is narrower: in 2027, expect specific freight lanes running genuinely driver-out. Not a nationwide rollout, but a real, repeatable route or two that proves the economics rather than the concept.

3. Most new robots learn in simulation first

Training in generated worlds stops being a research trick and becomes the default. By the end of 2027, expect the majority of new robot policies to be trained and stress-tested in simulation before they ever touch hardware, with the real world reserved for final validation.

4. Manipulation has its breakout in fulfillment

Navigation is largely solved; grabbing a random object is not. Warehouses, with their endless supply of varied objects and clear economics, are where manipulation gets good first. Expect visible progress in real fulfillment operations, not just lab demos.

5. One-brain-any-body models get real traction

The idea of a single foundation model that drives many different robots moves from ambition to early evidence. Expect at least one credible demonstration of a model transferring usefully across distinctly different machines, the first proof that the generalist bet is more than a slide.

6. Capital gets more selective, not smaller

The frontier-or-bubble question does not resolve, but the money changes shape. Expect fewer, larger bets on teams with real deployments, and a harder time for companies whose main asset is a good highlight reel. The dabblers thin out; the builders raise up.

The common thread is a field growing up: less spectacle, more proof. Some of these will land early and some late, which is exactly what makes 2027 worth watching.

These are the debates that fill the room at AUTONOMOUS every year. To see which calls hold up, and hear the next set, join the 2027 waitlist for San Francisco.