The Sunrise 2027 deadline is likely to be met on the leg it actually measures, retail scanner capability, and missed on the leg shoppers would notice, 2D codes on the packs they buy. Signals gathered in the past three weeks point to a transition whose supply side is productising fast while its demand side is still being argued about. The working call here: through 30 June 2027 the dated announcement flow in this migration is likely to stay dominated by coding, marking and QR-lifecycle vendors rather than by named, chain-wide retailer go-lives, and by 31 December 2027 the share of US grocery units carrying a GS1 2D code on pack is likely to remain a clear minority. The linear UPC is expected to still be on almost everything.
In short
- The prediction: Sunrise 2027 likely closes 2027 as a scanner-readiness success and an on-pack deployment shortfall, with under half of US grocery units carrying a GS1 2D code by 31 December 2027.
- The near checkpoint: fewer than three of the ten largest US grocery retailers are likely to publish a dated, chain-wide commitment to accept GS1 Digital Link QR codes at all lanes before 30 June 2027.
- Signal 1: GS1 refreshed its 2D readiness framework on 22 September 2026, and that framework defines the 2027 goal as POS scanners being capable, self-attested by vendors, not as codes reaching shelves.
- Signal 2: Markem-Imaje, part of Dover, launched GS1 Digital Link QR management software plus a one-day on-site readiness audit on 23 September 2026, which is what a vendor sells to buyers still at the diagnosis stage.
- Signal 3: a W3C and GS1 workshop in Zurich on 8–9 September 2026, co-hosted with Google, Shopify, OpenAI and Mars, relocated the payoff of the 2D code from checkout speed to machine-readable product identity, and closed without formal outcomes.
Why this matters now
Sunrise 2027 is the industry programme to move retail from the linear UPC and EAN barcode to a defined set of 2D symbols: a QR Code carrying a GS1 Digital Link URI, a GS1 DataMatrix, or a Data Matrix carrying a Digital Link URI. The stated goal, in GS1’s own wording, is that by the end of 2027 retail point-of-sale systems globally should be capable of reading and processing the product identifier from both the old and the new symbols. It is worth reading that sentence twice. The commitment is about what tills can do, not about what is printed on packs.
That distinction has been blurred in most coverage, and the blurring matters commercially. A retailer reading the deadline as a hardware and firmware project scopes a scanner refresh, a middleware change and a POS software certification. A brand reading it as a packaging deadline scopes artwork revision, print plate changes, coding line investment, resolver hosting and a data governance function. Those are different budgets on different approval cycles, and only one of them is being forced by a date.
The baseline is unflattering. Research GS1 US commissioned from VDC Research, covering close to 600 retail IT decision makers, found roughly 68.5% of retailers still running laser scanners that physically cannot decode a 2D symbol, with about 84% either evaluating or planning a migration to optical scanning. GS1 US has also framed 90% POS readiness as the practical scale-up threshold and has been explicit that UPCs continue to be scanned and accepted after 2027. A deadline that grandfathers the incumbent format is a capability deadline, not a switchover.
The hardware story is genuinely moving, because scanners sit inside a normal replacement cycle and most new bioptic and presentation units are imaging devices already. Retailers buying scanners, scales, kiosks and screens in 2026 are mostly buying 2D-capable optics whether or not they have a Sunrise programme. The packaging story has no equivalent tailwind, because packs change when a brand chooses to change them.
So the interesting question for the next four quarters is not whether the deadline is hit. It is which side of the ecosystem produces dated, verifiable commitments, and which side keeps producing readiness content. Three signals from September point the same way.
Signal 1: GS1 restated the 2027 goal as self-attested scanner capability
GS1’s 2D barcodes at point-of-sale solution pages carry an update stamp of 22 September 2026, and the underlying criteria document on the GS1 reference site was republished on 24 September 2026. This is the load-bearing signal, because it is the standards body defining the test it will be graded against, 15 months out.
The criteria document states the ambition directly: the transition from legacy linear 1D barcodes to more capable 2D barcodes on pack, with an initial goal of retail POS scanners globally capable of reading and processing both old and new barcodes by the end of 2027. The phrase “initial goal” is doing quiet work. The on-pack transition is the programme; scanner capability is the first milestone.
The mechanism is self-assessment. Solution providers self-attest against published criteria, and GS1 maintains continually updated tables of declared readiness by model. Those tables are organised by product class, and the class list is itself informative: scanners, verifiers, printers, barcode creation software, and a separate group of connected data services criteria covering brand domain names, redirection from a Digital Link URI to item information, support for identifiers beyond the product number, and optional resolver conformance.
The technical bar for a scanner is specific enough to be checkable. To be called fully 2D capable a scanner must handle reverse reflectance (light symbol on dark background) as well as normal reflectance for all three retail 2D symbols, return a result in under 300 milliseconds on high-speed bioptic units and under 500 milliseconds on presentation and activated handheld devices, and be configurable by the retailer to transmit either the product identifier alone or the identifier plus additional data. Print and symbology conformance leans on ISO/IEC 15415, 15416, 16022 and 18004.
Two implications follow. First, the latency thresholds confirm that checkout throughput is treated as a constraint to be protected rather than a benefit to be gained. Nobody is upgrading a front end because 2D scans faster. Retailers running lanes where seconds compound, particularly those weighing where self-checkout pays and where it does not, are being told the new symbol must not make throughput worse.
Second, a self-attestation register measures vendor product capability, not installed base, and not shelf penetration. It is the right instrument for the milestone GS1 set and the wrong instrument for the outcome everyone describes. Nothing in the framework produces a number for the share of units at shelf carrying a 2D code, which is precisely the number that would falsify or confirm a claim of transition.
There is also a quiet complexity tell. Three permitted symbols and two syntaxes mean a retailer must decide what its front end transmits, and a brand must decide what it encodes. The connected data services criteria, which cover brand-owned domains and redirection behaviour, describe an ongoing web operations commitment rather than a print change. That is a governance obligation with no deadline attached to it.
Signal 2: the coding incumbent started selling readiness audits, not volume
On 23 September 2026 Markem-Imaje, the industrial marking and coding business inside Dover, launched two products: CoLOS 2D Activate and CoLOS 2D Engage. Per the company’s release, they are enterprise software-as-a-service tools that let consumer goods manufacturers create, organise, activate and manage both static and dynamic GS1 Digital Link QR codes, with the stated aims of Sunrise 2027 preparation, meeting evolving retailer and regulatory requirements, and unlocking connected packaging value.
The product is not the signal. The services attached to it are. The same release describes implementation experts providing a structured assessment of a manufacturer’s production environment, coding processes and operational readiness, delivered as a full-day on-site review, followed by recommendations to identify gaps, prioritise investments and build a roadmap for implementing GS1 Digital Link QR codes.
Read that as a market-stage indicator. A full-day gap assessment with a roadmap deliverable is what an incumbent sells when buyers have not yet decided what to do. It is the offer that precedes capital approval, not the offer that follows it. Fifteen months from a deadline, a vendor selling roadmaps to its own installed base is telling you that the installed base does not have roadmaps.
The launch venues reinforce the read. The solutions were shown at the PPMA Show in Birmingham on September 22–24 and are scheduled for PACK EXPO International in Chicago on October 18–21. Both are production and packaging equipment shows. The audience being recruited is the plant and the packaging engineer, not the retail merchant or the category buyer.
Note also what a dynamic Digital Link QR implies operationally. Static codes are a print change. Dynamic codes require resolver infrastructure, redirection rules that can change after the pack has shipped, versioning, and an owner accountable when a scan resolves to the wrong page. Selling that as managed SaaS is rational precisely because most manufacturers do not want to run it, which is another way of saying the capability is not resident in the brands yet.
On the brand side the picture is consistent with pilots rather than conversion. Procter & Gamble, Coca-Cola, PepsiCo, Nestlé, L’Oréal, Unilever and Mondelez have all been reported as running GS1 Digital Link pilots or selected ranges, and US beverage firms have been expanding ingredient-transparency QR programmes across portfolios during 2026. Pilots across selected stock-keeping units are real progress and are also, by construction, not portfolio conversion.
Signal 3: Zurich moved the payoff from the till to machine-readable identity
On 8–9 September 2026 a workshop titled “E-Commerce for Humans and AI Agents” ran in Zurich, co-hosted by W3C and GS1 with Google, Shopify, OpenAI and Mars among the hosts. The framing problem discussed there was described as the last meter: an agent can complete a transaction competently and still buy the wrong physical item, because it cannot reliably tie a listing to a specific product.
The proposed anchor was the product identifier itself, with the global trade item number positioned as the primary identifier for physical goods, and GS1 Digital Link as the mechanism that turns a pack into a brand-controlled web address resolving to brand-authoritative data. A Mars pilot with K10X was presented as a demonstration, using Digital Link together with the GS1 web vocabulary to resolve packaging variations of the same product.
The workshop also surfaced the gap that matters. Vocabulary fragmentation between schema.org, GoodRelations and the GS1 web vocabulary was identified as unresolved, with participants pointing to the need for a vocabulary-neutral interoperability layer before this scales. The session closed without formal outcomes, which is normal for a workshop and still relevant to timing.
The strategic content of this signal is a change of sponsor. If the value of a 2D code is faster recall handling and richer nutrition data at the shelf, the sponsor is a retail operations budget. If the value is that an AI shopping agent can resolve a pack to authoritative brand data, the sponsor is the brand’s digital and data organisation, and the competitive pressure comes from the agent platforms rather than from the grocer.
That shift has a precedent in how merchant product data is already being restratified by platform decisions, a dynamic visible in the way the Content API sunset splits merchants into two data tiers. Brands with governed, resolvable, machine-readable product data get represented accurately by agents. Brands without it get approximated.
The timing implication cuts both ways, and this is the honest tension in the thesis. Agent-driven demand is a stronger long-run motivator than checkout convenience, because it touches revenue rather than store labour. It is also, in September 2026, pre-standard. A pull that is real but not yet specified does not move a 2027 packaging budget. It moves a 2028 one.
What the pattern suggests
Put the three signals next to each other and the asymmetry is the finding. Every dated artifact in the past month comes from the supply side of the transition: a standards body refining self-attestation criteria, an equipment vendor launching software and audits, a standards workshop scoping a future use case. None of them is a retailer or a brand committing to a date.
| Signal | Date | Source type | What it establishes | What it does not establish |
|---|---|---|---|---|
| GS1 2D readiness framework refresh | 22 September 2026 (criteria doc 24 September) | Standards body, primary | The 2027 test is scanner capability, self-attested, across four product classes | Any measure of installed base or on-pack penetration |
| Markem-Imaje CoLOS 2D Activate and Engage | 23 September 2026 | Vendor release, primary | Brand-side buyers are still buying gap assessments and roadmaps | Committed line conversions or volume of coded packs |
| W3C and GS1 Zurich workshop | 8–9 September 2026 | Standards workshop, reported | Agent product identity is becoming the strategic rationale for 2D | A settled vocabulary, a ratified spec, or a deployment timeline |
The pattern suggests three testable consequences. First, scanner readiness is likely to be reported as on track through 2027, because the metric is self-declared vendor capability against a published criteria list and the replacement cycle does most of the work. Expect readiness framed as a percentage of models or of installed lanes, not as a percentage of items scanned as 2D.
Second, on-pack penetration is likely to stay low and, importantly, unmeasured. There is no authoritative public figure for the share of US grocery units carrying a 2D code, and no mechanism in the framework that would produce one. A prediction that cannot be scored is a weak prediction, so the practical test has to be built from observable proxies rather than from a headline statistic.
Third, the categories that convert first are likely to be the ones where variable data is already printed at line speed: fresh and prepared food with lot and expiry coding, private label where the retailer controls artwork, pharmacy and regulated goods, and anything already under a traceability obligation. Categories printing long runs on gravure or flexo with 12 to 24 month artwork cycles are likely to be last, and they are also where the unit volume sits.
The scoring rules I would hold this piece to, in order of usefulness: whether any top-ten US grocer publishes a dated all-lane acceptance commitment before 30 June 2027; whether GS1 or GS1 US publishes an on-pack penetration figure above 50% during 2027; whether a top-five US retailer issues a supplier requirement with a compliance date and a consequence; and whether the announcement mix shifts from enablement vendors to retailers and brands.
Wider context: deadlines move packaging only when they carry a penalty
The useful comparison set is not other barcode projects. It is other industry deadlines, sorted by whether non-compliance cost anyone money on a specific date. That variable, rather than technical merit or industry enthusiasm, has historically decided whether a retail deadline compressed behaviour or drifted.
| Programme | Deadline mechanism | Cost of missing it | Observed pace |
|---|---|---|---|
| EMV chip migration, US | Card network liability shift, October 2015 | Fraud liability transferred to the least compliant party | Terminal deployment fast, software certification slow, but the date held |
| Walmart RFID apparel mandate | Supplier requirement with chargeback exposure | Commercial penalty to the supplier | Years of slippage from the early-2000s pallet mandate before a narrower relaunch in the 2020s |
| FDA food traceability rule (FSMA 204) | Federal rule with a compliance date | Regulatory enforcement | Compliance date pushed out once industry showed it was not ready |
| EU digital product passport under ESPR | Delegated acts by product group | Market access in the EU | Staged and category-specific, credible but slow |
| Sunrise 2027 | Voluntary industry ambition, self-attested capability | None specified for any party | Supply-side productisation ahead of buyer commitment |
Sunrise 2027 is the only row with no penalty. That is not a criticism of the programme, which was designed as an enabling milestone rather than a mandate, but it is decisive for forecasting. A voluntary date coordinates suppliers efficiently and coordinates buyers weakly.
The RFID precedent is the closest analogue and the most cautionary. The technology worked, the retailer pushing it was the largest in the world, the business case was legible, and the mandate still slipped for years before returning in narrower, category-specific form. What eventually moved it was a retailer attaching a date and a financial consequence to specific categories.
Retail capital allocation is the other constraint. Grocers in 2026 are choosing between store labour, price investment, refrigeration, e-commerce economics and front-end technology, and the dominant pattern has been to fund store fundamentals first, which is the logic behind the view that Kroger’s next reset likely comes from stores rather than e-commerce. A firmware and configuration project with no deadline penalty is a reasonable thing to schedule behind those.
Geography adds a nuance that complicates simple pessimism. National GS1 organisations in smaller, more concentrated retail markets can move faster than large fragmented ones, and GS1 Lanka’s 2026 activity, with more than 5,500 member companies registered on its identification platforms and over 20,000 product records uploaded, is an example of a national migration being coordinated centrally. Expect the first credible claims of high 2D coverage to come from markets like these, and expect them to be read, incorrectly, as evidence about the US.
Implications for retailers, brands and platforms
For retailers the operative question in the next two quarters is not whether to comply but what to transmit. A scanner that can decode three symbols still needs a decision about whether the front end passes the product identifier alone or the identifier plus lot, expiry and serial data, and every downstream system that receives it needs to tolerate the longer string. The cheap version of this project is a firmware and configuration exercise. The expensive version is a data contract change across POS, promotions, inventory and loss prevention.
There is real operational upside available for retailers willing to take the expensive version, and it sits in perishables. A code carrying lot and expiry turns recall scope from a category-wide sweep into an item-level query, and it makes date-based markdown automatable at the lane. Retailers already building that discipline through lot and expiry tracking for perishables and cosmetics are the ones positioned to capture it, because the constraint there is data quality rather than symbology.
For brands the decision is a sequencing problem, and the September evidence suggests most are still sequencing. The cost stack is artwork revision, plate or digital print change, coding line capability, resolver hosting, redirection governance and analytics. The revenue case is consumer engagement plus, increasingly, accurate representation to shopping agents. The risk of waiting is not a retail penalty in 2027, it is being approximated by agents in 2028.
| Scenario | What would have to happen | Observable by 30 June 2027 | Assessment |
|---|---|---|---|
| Base case: capability met, shelves lag | Nothing changes; replacement cycle carries scanners, brands stay in pilot | Enablement vendors dominate announcements; no top-ten grocer all-lane commitment | Most likely on current signals |
| Fast case: a mandate appears | A top-five US retailer sets a dated supplier requirement with a consequence | A supplier bulletin or investor-day slide with a compliance date | Possible; this is the single change that would break the thesis |
| Fast case: regulation forces coding | Traceability or product-passport rules require item-level data in scope categories | A compliance date inside 2027 or 2028 for a major category | Plausible in fresh food and EU-exposed categories, narrow in scope |
| Slow case: agent standards stall | Vocabulary fragmentation persists past 2027 | No ratified interoperability layer following Zurich | Would push brand-side conversion into 2028 and beyond |
For platforms and agent builders the asymmetry is an opportunity. Whoever makes brand-authoritative product data easy to publish and cheap to resolve gets to define the interface, and the Zurich host list suggests the contest for that position has already started. The commercial prize is not the barcode. It is being the layer agents trust when they resolve a physical item.
For investors the near-term exposure is cleanest in the enablement layer: scanning and imaging hardware, coding and marking equipment, QR lifecycle software, verification, and the consultancies doing readiness work. That is where dated revenue events are appearing now. Exposure premised on rapid consumer-facing 2D ubiquity in 2027 looks early on the evidence available.
Caveats: what could go wrong with this call
The strongest counter-signal is the one that has broken every similar forecast: a retailer mandate. If a top-five US retailer publishes a dated supplier requirement for GS1 Digital Link codes with chargeback or listing consequences, brand packaging timelines compress quickly, because packaging follows commercial risk rather than industry ambition. This is a single-event risk to the thesis, it is not currently visible in any September signal, and one announcement would be enough to invalidate the near-term checkpoint.
The second counter-signal is regulatory forcing that bypasses retail entirely. Food traceability rules, product passport requirements in the EU, and deforestation and sustainability reporting obligations can all require item-level or lot-level data regardless of what happens at the till. Where those bite, 2D coding arrives as a compliance byproduct and the shelf penetration number rises for reasons that have nothing to do with Sunrise 2027.
The third is a measurement objection to my own prediction, and it is fair. Nobody publishes an authoritative share-of-units-with-2D figure, so the year-end claim risks being unresolvable rather than wrong. That is why the near checkpoint is framed around the announcement mix and around named retailer commitments, which are countable, and why the resolution proxies are stated explicitly above rather than left implicit.
The fourth is that the cost asymmetry I lean on is category-dependent. For lines already printing variable data, including fresh, prepared food, pharmacy and much of private label, adding a 2D symbol is a modest change rather than a capital project. If those categories convert broadly, aggregate penetration could rise faster than a plate-change model predicts, particularly in retailers with large own-brand ranges.
The fifth is timing risk on the agent side in the opposite direction. If a major platform ships an agent shopping surface that visibly privileges products with resolvable brand data, brand urgency could reprice inside two quarters rather than four. Competitive visibility loss moves marketing budgets faster than operational efficiency moves capital budgets, and the parallel with how quickly commerce teams reacted to modern POS platform shifts is instructive.
Finally, this piece treats the US grocery channel as the reference market. Concentrated markets with active national GS1 bodies may well hit high coverage earlier, and claims drawn from them are likely to be quoted as though they generalise. They do not, and readers should discount cross-market extrapolation accordingly.
FAQ
What exactly does Sunrise 2027 require?
It sets an industry goal that by the end of 2027 retail point-of-sale systems globally should be capable of reading and processing the product identifier from a defined set of 2D barcodes as well as from existing linear barcodes. It is a capability milestone for scanning infrastructure. It does not require brands to have converted their packaging by that date.
Does the UPC stop working after 2027?
No. GS1 US has been explicit that UPCs continue to be scanned and accepted after 2027 while the industry scales toward its readiness threshold. Dual printing, with both a linear barcode and a 2D symbol on pack, is the recommended transition approach, and the linear code is likely to remain on most packs well beyond 2027.
Is there a penalty for missing the deadline?
None is specified for any party. That is the analytical crux of this piece. Unlike the EMV liability shift or a retailer supplier mandate, Sunrise 2027 carries no financial or regulatory consequence, which is why supply-side productisation is running ahead of buyer commitment.
Are you predicting Sunrise 2027 fails?
No, and this is the most likely misreading. On the metric the programme set, scanner capability, it is plausibly on track and may well be declared met. The prediction is narrower: the on-pack leg is likely to lag materially, and readiness reporting is likely to describe vendor and lane capability rather than the share of items at shelf carrying a 2D code.
What would prove this prediction wrong fastest?
A dated, chain-wide acceptance commitment from a top-ten US grocer before 30 June 2027, or a dated supplier requirement with a consequence from a top-five US retailer. Either would signal that the forcing mechanism the programme lacks has been supplied privately, and brand packaging timelines would likely compress in response.
Which categories are likely to carry 2D codes first?
Categories already printing variable data at line speed: fresh and prepared food with lot and expiry coding, pharmacy and regulated goods, and private label where the retailer controls the artwork. Long-run centrally printed packaging with slow artwork cycles is likely to be last, and that is where most unit volume sits.
Why would AI shopping agents matter to a barcode deadline?
Because they change who benefits. An agent resolving a physical product to brand-authoritative data needs a stable identifier and a resolvable address, which is what a Digital Link code provides. That moves the business case from store operations to brand representation, and it moves the sponsor from a retail budget to a brand digital budget.
Is the agent case strong enough to accelerate 2027 packaging decisions?
Probably not in 2027, on current evidence. The Zurich workshop identified an unresolved vocabulary gap between competing product data standards and closed without formal outcomes. A pull that is real but not yet specified tends to move the following budget cycle rather than the current one.
What should a mid-sized retailer actually do in the next two quarters?
Confirm which installed scanner models are already 2D capable against the published readiness criteria, decide what the front end will transmit, and test that downstream systems tolerate the longer data string. That sequence is inexpensive, it is reversible, and it converts a deadline into a configuration decision rather than a capital request.
The one primary source worth reading in full is GS1’s own criteria for solution provider 2D readiness, which is where the wording of the 2027 goal and the self-attestation mechanism can be checked directly: GS1 solution provider 2D readiness criteria.