Autonomous Surgery's 2026 Landscape: One Robot Got Cleared, One Got Liquidated
Updated: Aug 26
Ask a surgeon what changes when a robot takes on part of a procedure.
Most will not start with the mechanics.
They start with trust.
Surgical robots are no longer simple instrument extensions.
They are beginning to see, interpret, and, in narrow ways, decide.
That is the gap between assistance and autonomy.
The Autonomy Ladder
Autonomy is not binary.
It has six levels, from manual surgery to full independence.
Level | Description | Human role |
0 | Manual surgery | Surgeon performs the procedure |
1 | Robotic assistance | Robot assists specific actions |
2 | Shared control | Surgeon and robot jointly control movement |
3 | Supervised autonomy | Robot performs defined surgical tasks under continuous human supervision |
4 | Conditional autonomy | Robot manages broader surgical workflows within predefined conditions; human intervenes when required |
5 | Full autonomy | System performs the surgery independently |
Nearly every commercial system sold sits at Level 1 or 2.
The frontier, where research labs are working, is Level 3.
Who's Actually Building It, and Who Just Failed
Three groups compete for the same future, and 2026 has already separated the funded from the failed.
Established robotics companies hold the installed base.
Intuitive Surgical remains the incumbent, with more than 8,000 da Vinci systems installed worldwide and over 12 million procedures completed.
On July 22, 2026, Johnson & Johnson MedTech ended Intuitive's two-decade run as the only serious soft-tissue robotics player.
The FDA granted De Novo authorization for J&J's OTTAVA system, covering ten general surgery procedures including gastrectomy, gastric bypass, and appendectomy.
J&J states that OTTAVA's table-integrated arms cut OR footprint by 30 to 50 percent compared to boom-mounted systems, a figure drawn from the company's own claims support documentation rather than an independent study.
Medtronic's Hugo RAS reached a comparable milestone: its first U.S. commercial cases were performed at the start of 2026.
None of these three systems claim autonomy beyond assisted or shared control.
Research labs are where the actual autonomy work happens.
At Johns Hopkins, Axel Krieger's team built on its 2022 Smart Tissue Autonomous Robot, STAR, which completed intestinal anastomosis in pigs without human guidance.
Their successor system, SRT-H, performed eight gallbladder removals on lifelike simulated models with expert-level accuracy, using a large language model to plan tasks and a separate low-level controller to execute them.
The results were published in Science Robotics.
None of it has been tested on a live human patient.
AI and robotics companies are trying to own the intelligence layer instead of building another arm.
Moon Surgical's Maestro, an FDA-cleared robotic assistant, added edge-to-cloud data capture across its installed base, positioning the company to sell insight from procedure data rather than the robot alone.
Vicarious Surgical took the opposite path.
On July 21, 2026, its shareholders voted to dissolve the company and liquidate its assets, ending a twelve-year effort to build a single-port, autonomy-leaning surgical robot, according to the company's own SEC filing.
The lesson is not that autonomy failed.
It is that capital discipline decides who gets to keep trying.
The Seven-Layer Stack
The robot is one piece of a bigger system.
Data: video, sensor data, demonstrations, outcomes
Perception: computer vision, anatomical understanding
Intelligence: models that predict the next action
Planning: deciding what happens next
Control: turning decisions into movement
Safety: detecting failure, keeping a human in the loop
Validation: simulation, testing, clinical evidence
SRT-H's hierarchical model sits across layers 03 and 04.
Moon Surgical's cloud pipeline sits at layer 01.
Most companies own one or two layers; almost none own the full stack.
Why It's Arriving Slowly
A warehouse robot that fails drops a box.
A surgical robot that fails can injure a patient.
Every patient's anatomy is different, and the rarest, most dangerous moments are exactly where training data is thinnest.
Vicarious Surgical's dissolution shows the other constraint: even a well-funded team can run out of runway before the technology reaches the OR.
That is why the first autonomous systems will handle single tasks: positioning an instrument, tracking anatomy, one suturing sequence, while the surgeon stays responsible for the whole operation.
Task autonomy first, then procedural autonomy, then broader autonomy.
Why It Matters
Autonomous surgery will not arrive as one breakthrough.
It will emerge from robotics, AI, data, simulation, and safety engineering converging at once, funded by companies disciplined enough to survive the wait.
2026 has already shown both sides of that bet: one company cleared, one company liquidated.
The architecture is visible.
The companies that outlast this phase will define what autonomous surgery looks like once surgeons shift from controlling the robot to supervising it.
For the full breakdown of where each cleared system sits on this ladder, see the Autonomy Registry. For a claim-by-claim check of what manufacturers say versus what their FDA filings support, see the Autonomy Claims Audit.
