2026-06-16 Author : ZCS
Amazon One's retail palm-payment service ended on June 3, 2026. The Whole Foods palm scanner had allowed enrolled shoppers to pay by holding a hand over a reader rather than presenting a card or phone at checkout. The retail shutdown should not be confused with proof that every Amazon biometric service, or palm-based authentication as a category, has ended. American Express's Amazon One service notice identifies June 3 as the discontinuation date at participating businesses.
Amazon One's history raises two separate questions: why one retail service was discontinued, and what its technology can establish about palm recognition. The available evidence supports a more limited conclusion than either “palm payments failed” or “palm vein technology has replaced Amazon One.” For background on the broader category, see our guide to palm biometrics for identity verification.
Amazon introduced Amazon One in September 2020 at two Amazon Go stores in Seattle. The service subsequently expanded to Whole Foods Market and third-party locations. Amazon's original launch announcement described uses including payment, identification and entry.
Amazon announced a wider Whole Foods rollout in July 2023. At that announcement, more than 200 Whole Foods locations already offered the service, and Amazon said all 500-plus U.S. locations would offer it by the end of the year. That announcement described an expansion plan, not a completed 500-store rollout on the announcement date.
Amazon One was an optional service. Enrolled users could associate a payment method with their palm signature and use a palm scan to identify themselves at a participating checkout. The palm scan acted as an authentication mechanism within a wider payment service, not as a new card network or an independent source of funds.
Amazon One combined information from the palm's surface and underlying vein structure. The service used those features to create a numerical representation called a palm signature for identity matching. It was not simply a surface palm-print scanner. Our explanation of how palm vein scanning works provides background on the vein-recognition component, rather than a specification for Amazon One's complete implementation.
Amazon distinguished matching from image handling. Its technical explanation said matching used the palm signature rather than raw images, but palm and vein images were encrypted and sent to a restricted area of the AWS Cloud where the signature was created. Amazon also said palm images were not stored on the reader itself. “Not stored on the device” therefore did not mean “images were never sent to or stored in the cloud.” These are descriptions of Amazon's design, not independent security-test findings. Amazon explains both points in its palm-recognition and data-security overview.
Amazon One linked a successful identity match to the user's enrolled payment method at checkout. Enrollment, payment-provider support and the merchant's integration remained part of the complete service. A scanner recognizing a palm did not, on its own, constitute a payment system.
Amazon reported more than 3 million uses in its July 2023 Whole Foods announcement. The service covered identification, payment, loyalty membership and entry; that figure should not be rewritten as 3 million purchases or Whole Foods payment transactions. The Whole Foods expansion announcement provides the dated scope of that milestone.
Amazon's retail discontinuation was reported in January 2026, not first announced in February. GeekWire's January 28 report cited customer messages and an Amazon support notice identifying June 3, 2026 as the retail service end date. The report also quoted an Amazon spokesperson citing limited customer adoption and saying associated customer data would be securely deleted after the service ended. This records the announced policy; it is not an independent audit confirming that deletion was completed. See GeekWire's contemporaneous report.
The June 3 service deadline does not establish the date every physical reader was removed. Ending payment-service availability and removing hardware are separate events. The retail notice also should not be expanded into an unsupported statement about every hospital, enterprise deployment or venue.
Amazon One had attracted scrutiny over biometric data practices before the shutdown. An August 2021 announcement from Senator Amy Klobuchar's office documented questions from three U.S. senators about Amazon's collection and use of biometric information. That biometric-data inquiry establishes that privacy concerns existed; it does not establish that those concerns caused the 2026 shutdown.
Limited adoption is the reason attributed to Amazon in the cited reporting. The relative contributions of enrollment friction, privacy preferences, competing payment methods and operating costs cannot be determined from that statement alone.
Amazon One already used both palm-surface and vein information. The distinction is therefore not “Amazon One used palm prints, whereas palm vein systems use veins.” Other systems may use palm prints, palm veins or a combination, with different sensors, matching methods and deployment architectures.
Palm recognition should be evaluated as a complete implementation. Matching performance, presentation-attack defenses, enrollment procedures, data protection and recovery arrangements all affect the result. For an access-control context, our guide to palm vein security considerations explores implementation-level safeguards. An internal biometric feature does not, by itself, establish that one implementation is more secure than another. Amazon's own technical description also included liveness detection, so the original article's implied contrast overlooked part of Amazon One's design.
The retail shutdown is a commercial service decision, not a comparative biometric benchmark. The cited evidence does not demonstrate that Amazon One was technically inferior to vein-only systems, nor that palm-based authentication is suitable for every retail checkout. Those conclusions require evidence beyond the discontinuation notice.
Palm-based identification remains distinct from payment authorization. A patient check-in system verifies identity within a healthcare workflow; an exam-center system verifies a candidate; a palm-payment service additionally connects authentication to a supported payment arrangement. Examples from one category do not establish adoption or success in another.
Pearson's January 7, 2026 announcement illustrates why the technology labels matter. Pearson said it already used palm vein recognition and announced the addition of Redrock Biometrics' PalmID, described as palm-print-based authentication software. Rollout was scheduled to start in mid-2026 at test centers already equipped with palm vein devices, subject to local regulatory permission. The announcement is evidence of a plan, not confirmation that a global rollout has since been completed. See Pearson's PalmID announcement.
Bupa Hong Kong and Tencent announced a healthcare check-in deployment on January 29, 2026. The launch covered Express Check-in at 20 designated clinics, but palm scanning was specified at the Central flagship medical centre and the Tsim Sha Tsui and Admiralty centres; other clinics offered QR-based check-in. The announcement described Tencent PalmAI as combining palm-print and palm-vein recognition. This was a patient-identification application, not evidence that all 20 clinics accepted palm payments. The distinction is explicit in Bupa's launch announcement.
Biometric deployment should follow the actual workflow. An existing need to verify identity does not establish that biometric enrollment is mandatory, universally appropriate or accepted by users. Buyers need to assess enrollment choices, applicable requirements, integration, fallback procedures and service continuity for the specific project. For projects involving access control, our comparison of palm vein, fingerprint and face offers a separate framework for evaluating biometric options; it is not a comparison of complete payment services.
POS-integrated readers also require application and system integration. Combining a biometric sensor with checkout hardware does not automatically provide account linking, payment acceptance, consent management or a supported replacement for Amazon One. Hardware form factor and end-to-end service readiness must be assessed separately.
Amazon One brought palm-based identification and payment to retail, and Amazon later ended its retail service after citing limited adoption. That history does not establish that palm recognition failed as a technology, that privacy concerns alone caused the decision, or that another biometric modality is automatically a better replacement.
Palm-based systems should be judged by their intended task and verified deployment evidence. Payment, patient check-in and access control have different requirements. The useful lesson for buyers is to evaluate the full service, including enrollment, integration, data handling and continuity, rather than treating a palm scanner as a complete solution.
Amazon ended Amazon One's retail service on June 3, 2026. The service end date should not be described as independently verified removal of every physical scanner by that date.
An Amazon spokesperson cited limited customer adoption in reporting published in January 2026. Privacy scrutiny is documented, but the cited evidence does not establish its contribution to the shutdown decision.
The cited notice establishes the retail service end date. It is not sufficient evidence for a claim that every non-retail deployment ended at the same time. Current availability for any specific healthcare, enterprise or venue project requires direct confirmation.
Amazon One already combined palm-surface and vein information. Palm vein recognition describes the use of vein patterns for biometric matching; it is not a feature that Amazon One categorically lacked. Differences between systems require implementation-level comparison.
Palm-payment availability must be checked with the particular merchant and payment service. A healthcare or exam-center palm-identification deployment does not establish payment acceptance, and an integrated biometric POS terminal is not automatically a complete palm-payment service.