whirrwork
Insights

The future of robotic work, 2026 to 2035: what's real, what's pilot, what's hype

A job-by-job sort of robot work into real, pilot and hype, with the evidence for each, and Whirrwork's clearly labelled view of where things go by 2035.

By The Whirrwork team8 min read
  • future of work
  • robotics trends
  • forecasts
  • robotaxi
  • drone delivery
  • humanoids
  • crop spraying drones

The short version

  • Real means people pay for it, at scale, today. Crop-spraying drones, warehouse robots, factory arms, drone light shows, sidewalk delivery robots and (in a handful of cities) robotaxis are real.
  • Pilot means it's paid but small, or it works for one customer under close supervision. Driverless trucks, humanoids in factories and US drone delivery at scale are here.
  • Hype means the demo is ahead of the product. General-purpose home humanoids are the clearest case right now.
  • Our view: by 2035 the biggest change won't be a robot that does everything. It'll be a lot of boring, single-job robots that quietly got cheap enough to rent by the hour.

We run a directory of 308 robot jobs, so we spend a lot of time asking a simple question: can a normal buyer actually hire this today, and for how much? That question cuts through a surprising amount of noise. Here's the sort.

How we sorted the jobs

We used three tests, in this order:

  1. Is anyone paying for it, repeatedly, without a press release? Paid rides, paid acres, paid shows.
  2. Is there a published number? Volume (rides, deliveries, acres) from the operator or a regulator, and ideally a price.
  3. Is there a legal path to scale? A permit, a rule or an exemption process that more than one company can use.

A job needs all three to be "real". Two out of three is "pilot". One or zero, however good the video, is "hype".

The sort, job by job

JobStatus in 2026The evidence
Crop spraying by droneRealUS drone-treated area rose to 16.4 million acres in 2025, up 58.7%, while the average custom rate fell from $21 to $13 per acre (ASDC 2025 survey)
Warehouse robotsRealAmazon passed 1 million robots across 300+ facilities in July 2025 (Amazon)
Factory robot armsReal542,076 industrial robots installed worldwide in 2024; 4.66 million in operation (IFR World Robotics 2025)
RobotaxisReal, in some citiesWaymo reported 500,000 paid rides a week across 10 US cities in March 2026, from 50,000 in May 2024 (TechCrunch)
Sidewalk delivery robotsReal, nicheStarship passed 10 million deliveries with 3,000+ robots in 8 countries (Starship, April 2026)
Drone deliveryScalingZipline reports 3 million+ deliveries and 140 million+ miles flown (Zipline, October 2026)
Drone light showsRealPublic bids and rate cards exist, e.g. a 2026 proposal to a US city at $185–$195 per drone all-in (City of Mountain House)
Autonomous floor scrubbingRealPublished subscriptions from $599/month and 36-month RaaS from $829/month (Rendara, RobotLAB)
Driverless trucksPilot, moving fastAurora reports 500,000+ driverless miles since commercial launch in Texas (Aurora, October 2026)
Humanoids in factoriesPilotFigure's humanoid ran 10-hour shifts at BMW for months, logging 1,250+ hours and loading 90,000+ parts, then the model was retired (Figure)
Humanoids in homesHype (for now)1X sells NEO at $20,000 or $499/month, with a remote human "Expert" for tasks it doesn't know (1X)
US drone delivery beyond line of sight, for everyoneWaiting on rulesThe FAA's Part 108 BVLOS rule was still not final in October 2026 (Drone Authority)

A few notes on the table, because the details are where the honesty lives.

Robotaxis are real, and still small. 500,000 rides a week is a genuine business. It's also, as TechCrunch pointed out in the same piece, a sliver next to Uber's 13.5 billion trips in 2025 (a figure that includes deliveries). Waymo's fleet was about 3,067 vehicles per its December 2025 NHTSA filing (NHTSA recall report). Real, growing, and nowhere near "everywhere".

Drone spraying is the quiet winner. No one makes viral videos of a spray drone doing 40 acres before lunch, which is exactly why it's working. The number of FAA-approved Part 137 operators rose 58.3% to 1,710 in a year, per the same ASDC survey. And the price went down, which is what a real market does.

Humanoids in factories are a pilot, a good one. Figure published its own numbers, including that the robot's forearm was its top hardware failure point. That kind of disclosure is what a pilot should produce. But one robot model at one plant, then retired, is a pilot by definition.

Home humanoids are a pre-order. We're not saying NEO is fake. 1X is admirably upfront that a remote human can supervise or drive the robot for tasks it hasn't learned. That's useful. It's also not autonomy, and buyers should know exactly who is looking through the cameras. (We cover this in our privacy piece.)

What moves a job from pilot to real

Our view, after cataloguing hundreds of jobs: four things have to line up, and the robot itself is only one of them.

  1. A boring, repeated task. Spraying a field, moving a tote, scrubbing a floor. Variety is expensive for robots. Repetition is cheap.
  2. A legal path more than one company can use. Ag drones got one: Part 137 certification, with a streamlined application route for drones (FAA). Robotaxis got city-by-city permits. BVLOS delivery in the US is still waiting for Part 108.
  3. A price someone will publish. When operators start publishing per-acre or per-month prices, it means the cost is predictable enough to quote without a site visit. That's a maturity signal. Only about 12% of the 651 providers in our directory publish any price at all (78 of 651, from our own dataset).
  4. Failures that are survivable and reportable. Real industries have incident reports, recalls and fixes. Waymo's December 2025 school-bus recall covered all 3,067 vehicles and was fixed by software update (NHTSA). That's what a maturing system looks like: problems found, disclosed and patched.

A word about forecasts

Forecasts are not evidence. They're opinions with spreadsheets.

A useful example: in February 2024 Goldman Sachs projected a $38 billion humanoid market by 2035, about 1.4 million units (Goldman Sachs). In August 2026 it raised that to 6.48 million units and $138 billion (Investing.com). Nothing wrong with updating a forecast. But if a number can move almost 5x in two and a half years, don't sign a five-year contract because of it.

We use forecasts for one thing: knowing which suppliers will be well funded enough to still exist when your contract ends.

Whirrwork's view: ten calls for 2035

These are our opinions, not facts. Each comes with what would change our mind.

  1. Drone spraying becomes the default for small and awkward fields in most big farming countries. We'd be wrong if: drift rules or pesticide-label restrictions tighten sharply.
  2. Robotaxis are normal in dozens of large cities, and absent from most small towns. Wrong if: a serious safety event triggers a national pause.
  3. Driverless trucks run on fixed highway corridors, mostly hub-to-hub. Last-mile trucking stays human. Wrong if: the 2030 fleet targets operators talk about land early.
  4. Humanoids do paid work in warehouses and factories, in the thousands, not millions, mostly carrying totes and loading fixtures. Wrong if: hands get dramatically cheaper and more reliable.
  5. Home humanoids remain a premium, partly teleoperated product. Wrong if: a model does a week of laundry without a remote human.
  6. Drone delivery is routine in suburbs with space for it once BVLOS rules settle. Wrong if: noise complaints win.
  7. Cleaning robots are boring infrastructure, priced per month like a copier lease.
  8. Security robots stay niche; cameras, sensors and remote monitoring win most of that budget.
  9. Robot work gets priced by the hour, acre or job, and published prices become normal. (We're biased. We run a price index.)
  10. The biggest bottleneck is integration and people, not the robot. That's the theme of our hidden costs piece.

Bot or Not? The 2026 version

A robot makes sense now if…Hire a human (or wait) if…
The job is repetitive, at scale, and the robot version has published pricesThe job changes every day and needs hands and judgement
There's a clear permit path where you areThe legal path is "apply for a special exemption"
You can find three suppliers and compare quotesOnly one company offers it, and they won't share a price
A failure is annoying, not dangerousA failure could hurt someone and you can't control the area

And the Ladder Test still applies: if a competent person with a ladder could do it in an afternoon, the future of robotics can wait.

What to do with this

  • Buyers: rent or buy service in the "real" column. Pilot the "pilot" column with exit terms. Watch the "hype" column on YouTube, where it's free.
  • Suppliers: the fastest way to move your job from "pilot" to "real" in a buyer's head is to publish a price range and a reference customer. See how suppliers win more quotes.
  • Everyone: read the incident reports, not just the launch posts. We collected the best ones in how robot jobs go wrong.

Compare live prices for crop spraying, warehouse picking, robotaxi rides and humanoid rentals.

Sources

All accessed October 2026.

More Insights