Boston Dynamics can now assemble a single Atlas humanoid robot in approximately one day on the Waltham, Massachusetts prototype operations floor — the company published this manufacturing video today, disclosing an ambition to compress that benchmark to one hour by 2027 and eventually to minutes.

The disclosure arrived via a roughly six-minute YouTube film titled Machine, Assemble, Repair: How We Build Atlas, the company’s clearest statement yet that manufacturing velocity — not just robot capability — has become a primary engineering objective as Atlas transitions from research platform to production unit. Humanoids Daily confirmed the details in its own coverage published the same day.

How Atlas Gets Built: In-House Gear Cutting, Robotic Arms, Same-Week Iteration

Sr. Technical Program Manager Katie Conner narrates the film’s account of Atlas’s assembly progression: from builds that took multiple days roughly a year ago, to the current roughly one-day benchmark, to an ambition of one hour by 2027 — and eventually minutes. The one-day assembly progression described in the video is the first time Boston Dynamics has publicly disclosed a specific per-unit build time for Atlas.

The prototype operations floor at Waltham functions as something between a manufacturing line and a design studio. Machinists describe precision work — including in-house gear cutting and tight-tolerance components — as the reason fabrication stays under one roof with the engineering team. An engineer can walk to the shop floor, discuss a design change directly with a machinist, and have revised parts in hand the same day or the next — then watch those parts move in an assembled humanoid one floor above.

The iteration loop is fast enough to take an idea from a Monday conversation to a physically implemented change by the end of the same week. Robotic arms assist in sub-assembly, integrating the robot-being-built with the tools that build it.

This emphasis on rapid hardware cycling has been building throughout 2026. In early October, Boston Dynamics revealed a new four-finger hand design for Atlas — explicitly chosen to balance dexterity with cost, durability, and repairability. Faster iteration on the prototype floor is the mechanism that lets engineers work through those hardware trade-offs before committing them to production tooling.

How Does Boston Dynamics’ Build Time Compare to Figure AI and Tesla?

The one-day figure needs a precise definition before any competitive comparison can be made: it describes how long it takes to assemble one Atlas robot from components, not how fast a production line produces them. That distinction matters, because the metrics circulating across the humanoid sector measure different things.

Figure AI reported its BotQ facility in San Jose achieved one robot per hour in April — a line-throughput figure describing how many robots exit the production line each hour, not how long each takes to build. BotQ’s designed capacity is 12,000 humanoids per year, which corresponds to roughly 1.4 units per hour if the line runs continuously — but line throughput is not per-unit assembly time. A line producing one unit per hour can still require 12 or more hours to complete each robot if 12 robots are progressing through different assembly stages simultaneously.

Tesla was reportedly producing several hundred Optimus robots per week as of August 2026, according to The Information, though most units are committed to internal training, data collection, and testing rather than external customer shipment. In China, AGIBOT announced its 20,000th cumulative robot produced in September — a figure spanning its A, X, and G model lines, not a single humanoid platform.

What the convergence of disclosures confirms is that manufacturing velocity has become a primary competitive axis in humanoid robotics, alongside capability and cost. Atlas is currently estimated at approximately $140,000–$150,000 per unit. Tesla has publicly targeted a sub-$30,000 price for Optimus, leaning on its automotive manufacturing scale, though external sales remain on the horizon. Boston Dynamics has argued Atlas will deliver return on investment within two years of deployment — a payback claim first articulated in September 2026.

What One Hour Would Actually Require

Moving from a day to an hour is a 24-fold compression in build time. In manufacturing terms, that kind of step-change typically signals a transition from semi-manual assembly to highly automated production: fewer manual torque sequences, more robotic sub-assembly, and purpose-built tooling replacing adjustable-fixture work.

Boston Dynamics has not published a specific roadmap for achieving the one-hour target, and Conner frames it in the video’s build-time roadmap as an ambition rather than a committed timeline. The current baseline reflects the prototype operations floor, which is designed for fast design iteration on a relatively small number of units — not for volume output.

The comparison with Figure AI’s manufacturing progression is instructive. Figure 02 relied heavily on slow CNC machining — a prototype approach where complex components could each take over a week to produce. For the Figure 03, the company redesigned for tooled manufacturing: injection molding, die casting, metal injection molding, and stamping replace CNC-heavy processes, with sub-assembly cycle times under 20 seconds per part with steel molds. BotQ also plans to deploy its Helix AI to have robots assemble components on the line, eliminating conveyor systems entirely and turning the factory into a training environment for the same AI the robots run.

That progression illustrates the structural choice Boston Dynamics faces: to achieve hour-scale builds, the prototype-floor model would need to give way to purpose-built production tooling, at the cost of some design flexibility. The all-electric Atlas architecture — chosen in 2024, replacing the hydraulic systems of its predecessor — was already a step in that direction. Electric actuators eliminate hydraulic circuits and fluid-management steps, reducing assembly complexity and component count relative to hydraulic platforms. In-house gear cutting maintains quality control on precision drivetrain components without waiting on external vendors.

There is a second dimension the video does not name: a faster build cycle generates more hardware variation, more failure-mode observations, and more repair data — all of which feed back into Atlas’s AI training pipeline. More units built and iterated faster means more diverse physical experience for the learning systems that control them. Figure AI made this logic explicit in its BotQ strategy: the robots will build more robots at the facility, with the AI learning assembly as an additional task class. Boston Dynamics’ prototype floor operates the same feedback mechanism at an earlier scale.

Atlas Enters Serial Production After Three Decades as Research Platform

The production context sharpens what the one-day build time represents. Boston Dynamics unveiled the commercial Atlas at CES 2026 in January, announcing that production had begun immediately at its Waltham headquarters. The entire 2026 output was committed from day one — reserved for parent company Hyundai Motor Group and for Google DeepMind, which is using Atlas units to develop its Gemini Robotics AI foundation models.

The company has announced plans for a new 30,000-unit robotics factory targeting 30,000 Atlas units per year. In June 2026, Hyundai moved to take full ownership of Boston Dynamics, acquiring SoftBank’s remaining 9.65% stake for $325 million, completing the ownership transition that tightens the link between the robot maker and its primary customer.

Hyundai’s deployment roadmap calls for Atlas to begin work at its Ellabell, Georgia facility in 2028, starting with parts-sequencing tasks — organizing components in the correct order before they reach the assembly line. Component assembly is projected to follow in 2030, with Hyundai ultimately targeting tens of thousands of units across its global facilities.

Zachary Jackowski, VP and General Manager for humanoid robots at Boston Dynamics, described the commercial Atlas at CES 2026 as “the most production-friendly robot we have ever designed” — a claim the prototype floor’s current metrics are now beginning to quantify.

Former CEO Robert Playter, who led Boston Dynamics until February 2026, framed the commercial Atlas’s ambition in terms of functional breadth at the time of its CES launch: “If you’re going to have a robot that’s actually useful in the factory, it’s got to do a hundred different tasks, not just one or two,” he told Business Insider in January. Amanda McMaster, Boston Dynamics’ CFO, has been serving as interim head since Playter’s departure from Boston Dynamics in February. No permanent CEO appointment has been announced as of October 2026.

Goldman Sachs has revised its humanoid robotics market forecast to $138 billion by 2035, up from its earlier $38 billion projection, reflecting accelerating progress in AI software and manufacturing economics. A May 2025 Morgan Stanley report placed the long-term ceiling at over $5 trillion by 2050, contingent on widespread industrial and commercial adoption.

Boston Dynamics sits in an unusual position within that landscape: over three decades of robotics experience and a deeply technical heritage, facing well-funded competitors with different structural advantages — Tesla’s automotive manufacturing scale, Figure AI’s early BMW partnership and production-purpose factory, and a rapidly expanding cohort of Chinese manufacturers producing at cumulative volumes that already outpace Western peers in unit count, if not in industrial task sophistication.

The one-day build time and one-hour ambition are, ultimately, engineering benchmarks. But in the humanoid sector of 2026, where the race to prove robots can be built reliably, quickly, and at cost is every bit as important as the race to make them capable, they are also strategic signals.

Frequently Asked Questions

How long does it take to build a Boston Dynamics Atlas robot right now?

As of October 2026, a single Atlas can be assembled in approximately one day on Boston Dynamics’ prototype operations floor in Waltham, Massachusetts. That is the per-unit assembly time — from components to a walking robot — not the rate at which a full production line produces units. Boston Dynamics targets reducing this figure to approximately one hour by 2027, with a longer-term ambition of minutes.

How does Boston Dynamics’ one-day build time compare to Figure AI’s one-robot-per-hour claim?

The two figures measure different things and are not directly comparable. Boston Dynamics’ one-day metric describes how long it takes to assemble a single Atlas from start to finish on its prototype floor. Figure AI’s one-robot-per-hour milestone refers to line throughput at BotQ — how many robots exit the production line each hour — which reflects how many robots are in production simultaneously, not how long each takes to complete. A line producing one unit per hour can still take 12 or more hours to finish each robot if 12 units are at different stages at the same time. Both metrics describe manufacturing progress; they are measuring different things with different instruments.

Why does building robots faster matter beyond reducing costs?

Manufacturing speed in humanoid robotics is also a learning accelerant. Each robot built on the prototype floor generates hardware variation data and failure-mode observations that feed back into the AI training pipeline governing the robot’s behavior. A faster build cycle produces more units to vary, study, and iterate — generating richer training data for the reinforcement learning systems that make Atlas useful in unstructured environments.

Figure AI made this logic explicit in its BotQ strategy, where robots will eventually help assemble more robots, with the Helix AI learning factory assembly as an additional task class. Boston Dynamics’ prototype floor operates the same feedback mechanism at an earlier scale.

How much does Boston Dynamics Atlas cost, and when will it be available?

Atlas is currently estimated at approximately $140,000–$150,000 per unit. Boston Dynamics has argued the robot will deliver return on investment within two years of deployment for manufacturing customers. The entire 2026 production run is committed to Hyundai Motor Group and Google DeepMind, with broader commercial availability expected as production scales toward the company’s announced 30,000-unit-per-year factory capacity. Hyundai plans to begin deploying Atlas units at its Ellabell, Georgia facility in 2028.