Is Samsung Ballie Ever Coming Out? What’s Actually Confirmed in 2026

The 2020 unveiling ran as a suitcase-sized rolling ball that followed a single presenter around a demo space, took voice commands, and projected basic information onto nearby surfaces. The version shown at CES 2024 kept the sphere shape but was reworked internally, and CES 2025 confirmed that direction was still what Samsung intended to ship rather than a one-off keynote prop. Samsung has also described the assistant function in consistent terms across that span: it holds spoken conversation, takes calls, and triggers smart-home actions rather than functioning as a display accessory tethered to a phone. What changed between the 2020 concept and the 2025 version was not the pitch but the backend behind it.

Six years is a long span for a hardware concept to stay in active development without a retail unit, longer than the typical product cycle of the robot vacuums and mowers Robovations classifies most often, where a redesign usually ships within one or two years of its reveal. In 2025 the project also gained a defined feature set beyond the original concept: a partnership with Google Cloud put Gemini behind Ballie’s conversational layer, and Samsung named a summer 2025 launch window for the US and South Korean markets. That window passed without a launch. Samsung did not go silent about it. A spokesperson statement in December 2025 confirmed the delay and attributed it to a goal of delivering what the company described as

An even more impactful customer experience.

Samsung spokesperson, reported December 2025

That is a stated reason without a mechanism. Press coverage of the delay has since named four specific areas as unresolved: navigation, battery life, AI integration and pricing. As of this writing, coverage points to a possible 2026 release, not a confirmed one.

None of the four areas cited around the 2025 delay has a published answer. Samsung has not disclosed a navigation architecture. Demo footage shows Ballie moving through home spaces with apparent obstacle avoidance, but there is no published account of whether that relies on LiDAR, visual odometry, structured mapping, or some combination. Battery runtime, charging dock design, and projector brightness in daylight conditions are also undocumented. Pricing speculation online ranges from $500 to $2,000 based on projected capabilities, but Samsung has issued no official MSRP, and reporting on the delay names pricing itself as one of the unsettled questions rather than a number the company is simply withholding. Samsung has not disclosed camera specifications either; a lens mounted low on a rolling sphere would see a home from a different vantage than a countertop assistant or a floor-level vacuum, which matters for both obstacle avoidance and for what the camera can observe about a room, but no field of view, resolution or mounting height has been published.

Robovations classifies Samsung Ballie at Level II, Assisted Autonomy, on the Autonomy Ladder. The justification is narrow and specific: demo footage shows movement in structured home spaces with apparent obstacle avoidance, which supports autonomous mobility within a constrained environment. It does not support more than that, because Samsung has published no navigation architecture to verify what the demos show, and reporting on the 2025 delay names navigation itself as one of the areas still being worked out.

The Level II ceiling here is not a teleoperation cap; nothing in the record suggests a human pilot behind the demo footage. It is a documentation cap, which is a different and in some ways stricter constraint: the behavior may well be autonomous, but until Samsung publishes the architecture behind it, the classification cannot assume more than what the footage itself demonstrates. Reaching Level III, Conditional Autonomy, on the Ladder would require documentation of the navigation and obstacle-recovery architecture behind that footage, not repeated demo footage of the same behavior.

The autonomy level is one input into a broader readiness assessment, which for a robot with no shipping units, no price and no independent reviews runs on a different set of measures than a robot already in owners’ homes: how plausible the described engineering is, how far demonstrated capability tracks claimed capability, Samsung’s own track record in consumer robotics, and how many open questions stand between the current state and a shipping product. Robovations’ internal summary for this record describes Ballie as carrying high engineering risk, not because a rolling, projecting sphere is an unusual form factor to build, but because Samsung’s track record in consumer robotics is thin and a named launch window came and went in 2025 rather than quietly slipping unannounced.

• Track record: 35 out of 100, reflecting Samsung’s limited history shipping consumer robots to market.
• Engineering plausibility: 57, the strongest of the five inputs.
• Demonstrated versus claimed: 30, the gap between what demo footage shows and what Samsung has claimed the shipping product will do.
• Path to readiness: 15, the lowest input, reflecting how much remains unresolved between the current state and a shipping unit.

The spread across those five inputs is informative on its own. Engineering plausibility sits well above the other four, meaning the individual pieces Samsung is describing, a rolling chassis, a projector, a conversational layer built on an existing large language model, are each proven technologies elsewhere. What is unresolved is not whether they can exist, but whether Samsung can integrate and ship them together at a stated price with the reliability a consumer product requires. That distinction is why the path-to-readiness input sits far lower than the plausibility input rather than tracking it.

Robovations does not score Ballie on the standard 0-100 methodology used for shipping robots, because that methodology depends on inputs, failure rate, software stability, routine maintenance, that only exist once a product is in owners’ hands. The readiness score above measures a different question: not how good the robot is, but how much confidence the current evidence supports about whether and when it becomes one a reader can buy.

Ballie’s stated design puts microphones on a robot that moves through the house, plus a projector and camera lenses for wall and floor display and presumably for the obstacle avoidance its demo footage shows. Its conversational layer runs on Google’s Gemini under the 2025 Google Cloud partnership, which means voice interaction is very unlikely to be processed entirely on the device Samsung has shown. Samsung has not published a data-handling summary describing what portion of that pipeline stays local, what is sent to Google Cloud, or how long either company retains it.

That is a different kind of open question than a battery spec. A stationary smart speaker with a known cloud dependency is a well-understood privacy trade-off; a robot that moves through every room of a house, carrying a microphone and a camera, with no published account of its data boundary, is not yet a documented one for Ballie specifically. Robovations’ own accounting of open questions for this robot names the division of processing between the device and Google Cloud as a distinct item, alongside what the microphones capture, what leaves the home, how long either company retains it, and which smart-home actions Ballie can execute without a paired Samsung account or SmartThings hub.

None of this means the data is mishandled. It means the question has not been answered in public, on a robot whose entire pitch depends on carrying a microphone and a camera through every room of a house rather than sitting in one corner of it.

No MSRP has been published for any version of Ballie. Online speculation places a plausible range between $500 and $2,000 based on the capabilities Samsung has described, but that range is not sourced to Samsung and should be read as a guess rather than a figure. Reporting on the December 2025 delay lists pricing among the unresolved questions rather than a settled figure Samsung is holding back.

No maintenance record exists either, for the same reason: there is no shipped unit to document care for. The announced hardware includes a projector window and camera lenses that would need periodic wiping, and Samsung has published no cleaning procedure, no consumables list and no battery service plan for the design it has shown. Pricing and maintenance are not being withheld so much as they have never applied to a real product yet, which is itself informative: a robot six years into public development without an owner’s manual is still, functionally, a demo.

The record changes the moment a unit ships. At that point the assessment moves from a readiness score built on public statements to a full classification built on owner reports, and every open question listed here becomes a checkable fact rather than a projection. Until then, the six-year gap between the first demo and a retail listing is itself the most concrete data point available.

Six tracked years produce a specific kind of record: continuous public development, three redesigns, one named and missed launch window, and a company that has been willing to say a date slipped without saying when the new one is. That is a different posture than either a shipped product or an abandoned one, and it is why Ballie’s classification sits where it does: a real Level II robot in demo footage, an undocumented data boundary on paper, and a readiness assessment that reflects six years of announcements against zero shipped units.