How much does a work-capable humanoid robot cost?
Humanoid robot price, capability-adjusted · $/unit · measured 2021–2026 · fitted live from 5 observations by the Many Minded cost-curve engine.
The last measured value is $50K (2026, vendor pricing + press). Over the fitted window the series falls 27.5% a year (80% interval −39.2% to −13.6%) — a halving every 2.2 years (80%: 1.4–4.7). Carried forward on that fit, 2031: $10K (80%: $3K–$35K).
Estimate-grade series. This one is constructed, not read off an authoritative sheet — capability-adjusted price of a WORK-capable general-purpose humanoid — a judgment series (est-grade) until autonomy benchmarks exist; the raw list-price collapse is the measured sibling below, and it runs far steeper. Treat the level as an estimate and the slope as the claim.
Thin series — n=5. Under six observations the interval is carried by the Farmer–Lafond drift-volatility prior rather than by this series' own measured volatility. The band is wide because the evidence is thin, and that is the correct response.
the curve
what this series measures
capability-adjusted price of a WORK-capable general-purpose humanoid — a judgment series (est-grade) until autonomy benchmarks exist; the raw list-price collapse is the measured sibling below, and it runs far steeper.
the gate — what crossing this threshold changes
Central fit: ~2029. 80% window: 2028–2033 · est-grade · n=5. That is 1.3 halvings away at today's value.
what the fit says
| quantity | value | how it is computed |
|---|---|---|
| fitted rate | −27.5%/yr (80%: −39.2% to −13.6%) | Farmer–Lafond drift over the trailing window · n=5 points, 5 years |
| halving time | 2.2 years (80%: 1.4–4.7) | implied by the fitted drift |
Not shown for this curve, because the machinery returns nothing: Wright's law (no cumulative-deployment series for this technology); the accelerating/decelerating check (needs ≥8 observations, this has 5); a regime break (needs ≥12 observations).
the projection, with its interval
Log cost as a random walk with drift: the forecast variance grows with horizon (τ + τ²/m), which is why these bands widen instead of staying parallel. The middle column is the least useful number on this page; the interval is the claim.
| year | central fit | 80% interval | |
|---|---|---|---|
| 2027 | $36K | $24K to $56K | near horizon |
| 2028 | $26K | $14K to $51K | near horizon |
| 2029 | $19K | $8K to $45K | near horizon |
| 2030 | $14K | $5K to $40K | the band is already wide here |
| 2031 | $10K | $3K to $35K | the band is already wide here |
| 2032 | $7K | $2K to $30K | the band is already wide here |
| 2033 | $5K | $1K to $26K | far horizon — read the interval, not the middle |
Projected to 2033, the horizon the whole engine uses.
questions this page answers
How much does a work-capable humanoid robot cost?
$50K as of 2026, the latest measured value in the series (vendor pricing + press). The fitted trend has it falling 27.5% a year, with an 80% interval of −39.2% to −13.6%. This is an estimate-grade series: the level is constructed, so read the slope as the claim.
How fast is the price of a work-capable humanoid falling?
−27.5% a year over the fitted window, an 80% interval of −39.2% to −13.6% — a halving every 2.2 years (80%: 1.4 to 4.7 years). Fitted from 5 observations spanning 5 years.
What will the price of a work-capable humanoid be in 2031?
The central fit says $10K, inside an 80% interval of $3K to $35K. The interval is the forecast; the middle number is only its midpoint. Bands widen with horizon because shocks accumulate — a constant-width band would be overconfident.
When will the price of a work-capable humanoid reach ≤ $20K/unit (capable)?
The central fit says ~2029, with an 80% window of 2028–2033, on an estimate-grade series from 5 observations. Crossing it is what the engine calls "robot build labor ≤ half of human cost", firing: the town · houses grow 2×. Treat the window, not the year, as the claim.
Where does this data come from?
vendor pricing + press. 5 observations spanning 2021–2026. The series is estimate-grade — constructed from vendor sheets and practitioner reports, and labeled as such wherever it appears. Both the observation ledger and the fitting code are public, and the engine publishes its own calibration score and its misses.
the raw numbers
Every observation behind the fit, unrounded by us and unsmoothed. This table is here on purpose: graphs make people underestimate exponential change, and the raw series beside the curve is the one correction shown to work.
| year | $/unit | change |
|---|---|---|
| 2021 | $250K | — |
| 2023 | $150K | −22.5%/yr |
| 2024 | $90K | −40.0% |
| 2025 | $65K | −27.8% |
| 2026 | $50K | −23.1% |
where this comes from
- Source: vendor pricing + press.
- Basis: estimate-grade. The series is constructed from vendor sheets and practitioner reports, labeled as such everywhere it appears, and never presented as a measured benchmark.
- Observations: n=5, spanning 2021–2026. Ledger last updated 2026-11-02.
- Fit: the Farmer–Lafond stochastic model — log cost as a random walk with drift over a 10-year trailing window, Student-t tails, and (for series under 15 points) a volatility floor at their drift-volatility prior. The code is readable source; the observation ledger is public JSON.
- Reuse: this compiled series and the numbers on this page are published under CC BY 4.0 — take them, cite the page. The underlying source keeps its own terms.
- Accuracy: the engine grades its own intervals in public — every feasible hindcast re-fitted on data as it stood, scored for whether the actual landed inside the 80% band. The calibration score is on the engine page, misses included.
the rest of the graph
Other GOODS curves:
Recent briefs from GOODS, the daily record of what moved:
- Midmarket Manufacturers Deploy Robotics for Precision Production 2026-11-08Robotic integration enables midmarket manufacturers to achieve high-precision production without full overhauls, potentially lowering costs but unproven at scale.
- Fireworks AI claims businesses overpay for AI services by avoiding full model utilization 2026-11-08Fireworks AI states businesses waste money on AI services unless they use all available models for inference tasks, with specific cost-saving claims.
- Lazarus AI realigns Alibaba models for Western business use 2026-11-08Lazarus AI's ReAligned models enable U.S. companies to deploy Chinese-trained AI without state censorship barriers
- New coding benchmark finds agents fully solve barely a quarter of real change requests 2026-11-08Alibaba-led researchers built Co-Design Bench from real user sessions; the top model fully solved only 26.4 percent of tasks.
- Electrons caught obeying a space-time limit at attosecond speeds 2026-11-08Regensburg-led experiments show electron position and timing can't both be pinned down exactly, a finding tied to far faster future electronics.
- GlobalFoundries targets 2030 for cheaper semiconductors using FDX Fusion 2026-11-07GlobalFoundries VP details FDX Fusion technology that achieves 7nm-class performance without EUV lithography, with commercial production starting in 2028 and high-volume manufacturing targeted for 203
Every tracked curve: utility solar, installed · battery pack price · battery cell price · solar module price · disk storage · human genome sequencing · transistors per chip · fastest supercomputer · DRAM memory · solid-state storage · onshore wind electricity · solar electricity (LCOE) · space launch, to LEO · wheat yield · offshore wind electricity · AI task horizon (METR) · frontier AI training compute · modular housing, turnkey · frontier-grade intelligence · AI video, finished minute · people still offline · people outside a mobile-broadband signal · farm labor share · healthcare spend per person
← the whole engine: build clock, cascade, calibration · the archive · how we know