Annual Report2026 EditionPublic Data

State of Restaurant Tech 2026.

An aggregated, fully-cited snapshot of where the US and LATAM restaurant industry actually is in 2026 — marketplace commissions, POS market share, AI ordering adoption, payment processing, closure dynamics, and the Brazil / WhatsApp story. Every chart names the source behind it — and where the public record stops, this report says so instead of estimating.

FoodyOS Research

Independent aggregation, no vendor sponsorship

May 4, 2026Vendor figures re-checked August 8, 2026~3,500 words · 12 minute read

1. Executive summary

Five headline findings define restaurant technology in 2026:

The three US marketplaces publish three different ladders, and the floor is much lower than the shorthand suggests. DoorDash publishes 15% (Basic), 25% (Plus), 30% (Premier) on delivery and 6% on pickup; Uber Eats publishes 20% (Lite), 25% (Plus), 30% (Premium); Grubhub publishes a marketing commission of 5% (Basic), 15% (Plus), 20% (All-access), with its own delivery fleet an add-on on top. The widely repeated “15–30%” is wrong at the bottom — Grubhub Basic is 5% — and none of the three publishes a payment-processing rate you can add to it, so this report quotes each ladder as published rather than blending them into one effective number.^[1]^[2]^[3]
Toast extended its US POS lead. Toast reported roughly 140,000 locations live by year-end 2025 and continued double-digit location growth through 2026, consolidating its position above Square for Restaurants, Clover, Lightspeed, and TouchBistro in the US independent segment.^[4]^[5]
Direct online ordering finally crossed the 60% adoption line.The National Restaurant Association’s 2025–2026 technology surveys put US operators accepting direct online orders (own site or branded app) above 60% — a step-change driven by post-2020 marketplace fatigue and lower-cost SaaS direct-ordering platforms.^[6]^[7]
AI ordering left pilot phase. Voice-AI drive-thru, chat-based ordering (WhatsApp in LATAM, SMS and web chat in the US), and predictive prep are running in live service on US and LATAM menus, anchored by OpenAI, Anthropic, and Google model APIs. No vendor publishes a location count for these deployments, so §5 names the rollouts operators have documented themselves rather than estimating a total.^[8]^[9]^[10]
The Brazil iFood economics gap is the loudest in the world. iFood publishes 12% (Plano Básico), 23% (Plano Entrega) plus a 3.2% platform payment fee — 15.2% to 26.2% all-in before any monthly fee — which has triggered Abrasel-led organizing and a measurable shift to WhatsApp + own-site ordering across mid-market Brazilian restaurants.^[11]^[12]^[13]

2. Marketplace economics in 2026

The three major US delivery marketplaces — DoorDash^[1], Grubhub^[2], and Uber Eats^[3] — each publish a three-plan merchant structure, but the three ladders are notthe same ladder, and the widely repeated “15–30%” shorthand describes none of them exactly. DoorDash runs 15% (Basic), 25% (Plus), 30% (Premier) on delivery; Uber Eats runs 20% (Lite), 25% (Plus), 30% (Premium)on its marketplace; Grubhub’s published plans are a marketing commission of 5% (Basic), 15% (Plus), 20% (All-access), a different fee with a different base. Quoting the span from its real floor — 5% on Grubhub Basic — to its real ceiling matters, because a restaurant comparing quotes against a made-up 15% floor will read a Grubhub Basic quote as a discount it is not.

What you cannot add up.A single all-in “effective take rate” cannot be built from these pages, and this report no longer prints one. DoorDash states card processing is included inside the marketplace commission without publishing the rate; Uber Eats says the same for its marketplace plans; Grubhub excludes processing from every quoted rate and does not publish what it charges. Sponsored-listing spend is merchant-elected and unpublished. Any number layering those on top is an estimate, not a sourced figure — Restaurant Business Online^[14] and Nation’s Restaurant News^[15] have tracked operator-reported all-in figures well above the listed plan rate since 2022, and that trade-press reporting, not arithmetic of ours, is the honest citation for it.

Regulatory backdrop. New York City’s Local Law 88^[16] caps third-party delivery commissions at 15% on delivery and 5% on marketing. California’s AB 286^[17] requires marketplaces to disclose all fees to both restaurants and customers. We make no claim about how the litigation around these caps has resolved — but the commercial consequence is visible in the marketplaces’ own published pricing: Grubhub runs a separate New York City fee schedule^[18] alongside its national one, and notes on that page that the City Council changed the third-party delivery fee cap in 2025 (Grubhub NYC pricing page, as of August 2026).

In uncapped markets the published plans are explicit. DoorDash publishes^[1] 15% (Basic), 25% (Plus), 30% (Premier) on delivery, with pickup at 6% on every plan for merchants whose pickup menu prices match in-store, and orders placed through the restaurant’s own Online Ordering channel commission-free apart from payment processing (DoorDash merchant pricing page, August 2026). Grubhub’s national schedule^[2] is a marketing commission of 5% (Basic), 15% (Plus), 20% (All-access), with its delivery fleet a further 10% on top of whichever plan you are on and payment processing excluded from every quoted rate (Grubhub pricing page, August 2026). Uber Eats publishes^[3] 20% (Lite), 25% (Plus), 30% (Premium) on its marketplace, 15% when the restaurant delivers with its own drivers, and pickup at 7% where in-store pricing is validated (10% where it is not) — Uber Eats merchant pricing page, August 2026.

Figure 1 — Commission each US marketplace publishes on its own merchant pricing page, by plan, August 2026. Every bar is a rate the platform states; nothing is modelled, and no processing, ad spend or order-protection charge is added, because none of the three publishes those as a rate. Grubhub’s bars are a marketingcommission and are not directly comparable to the other two — using Grubhub’s own fleet adds a further 10% on top. Source: DoorDash merchant pricing, Grubhub pricing and fees, Uber Eats merchant pricing.

The conclusion most operators reach by year three on a marketplace is the same one independent operators in dense US cities have been writing about for half a decade: the marginal order from the marketplace is profitable only if the kitchen is already running, and the averageorder pulls the blended location margin toward zero or below. Toast’s 2026 industry report^[19] and the National Restaurant Association’s technology survey^[6] both surface a sustained shift in how operators blend channels — not abandoning marketplaces, but treating them as a customer-acquisition surface that should be converted to direct repeat orders within 30–60 days.

For an in-depth breakdown of the operator playbook, see our RMS buying guide for 2026.

3. POS market share — Toast, Square, Clover, Lightspeed, TouchBistro

Restaurant POS is the most concentrated tier of the restaurant tech stack. Five vendors dominate the US independent segment in 2026 — Toast, Square for Restaurants, Clover, Lightspeed Restaurant, and TouchBistro — together accounting for the overwhelming majority of new-store installations.

Toast. Toast (NYSE: TOST) reported 140,000 locations live as of Q4 2025^[4], up from roughly 99,000 at year-end 2023^[5]. The company’s 2024 and 2025 shareholder letters^[20] describe sustained net-location additions in the mid-five-figure range per year. Toast is vertically integrated — proprietary hardware, Toast Payments processing, and a deep app marketplace — and is the category leader by both location count and revenue.

Square for Restaurants. Block (NYSE: SQ / XYZ) reports the Square ecosystem at 4M+ active sellers across all verticals^[21]; Square for Restaurants is a subset of that base and is the dominant choice for single-location operators trading off feature depth for tablet-native simplicity. Block’s shareholder letters track the Restaurants vertical separately within the quarterly Investor Day deck.^[22]

Clover. Clover, owned by Fiserv, is bundled through the Fiserv banking and ISO channel and is widely deployed in casual / quick-service. Fiserv reports Clover annualized GPV at $300B+^[23]— that figure includes retail, but the restaurant subset is meaningful and is the channel’s largest vertical.^[24]

Lightspeed Restaurant. Lightspeed (NYSE: LSPD) reports ~165,000 customer locations^[25] across retail and hospitality globally. Lightspeed Restaurant is the hospitality-vertical product, and is particularly strong in full-service multi-location and in Europe / APAC.

TouchBistro.TouchBistro is private and does not publish quarterly counts; the company’s public website cites 29,000 restaurants^[26] served. TouchBistro is concentrated in independent full-service in North America.

Why this section has no market-share chart. Each figure above is the count its own vendor chose to publish, and no two are measured on the same ruler: Toast’s is restaurant locations, Lightspeed’s is customer locations across retail and hospitality worldwide, Block’s is sellers across every vertical, and Fiserv reports Clover as payment volume rather than locations. None of the five publishes a restaurant-only US location count — so a percentage split between them cannot be built from public filings, and we won’t manufacture one. Take the five numbers as five different rulers, each linked to the filing it came from.^[27]^[28]

Independent third-party trackers including Datanyze^[27] and BuiltWith^[28] produce broader share rollups including non-restaurant POS; for restaurant-only views, G2’s Restaurant POS Grid^[29] is the most current public ranking by user reviews and is refreshed quarterly.

For a vendor-by-vendor head-to-head, see our deep-dive comparison pages: FoodyOS vs Toast, FoodyOS vs Square, and FoodyOS vs Grubhub.

4. Online ordering adoption

The single largest behavioral shift inside US restaurants between 2019 and 2026 is the operator-side adoption of direct digital ordering. The National Restaurant Association’s 2026 State of the Restaurant Industry^[6] report puts US operators accepting direct online orders (their own website or branded app, not a marketplace) at 62% in 2025–2026, up from 28% in 2019. The same survey shows marketplace-only operators dropping from roughly half the industry to a minority, with the bulk of the remainder now hybrid (marketplace plus direct).

Toast Industry Report 2026. Toast’s 2026 Restaurant Success Report^[19] aggregates anonymized transaction data from 100,000+ Toast locations and shows direct online orders growing year-over year as a share of total off-premise revenue, with QSR and fast-casual concepts furthest along.

Restaurant365. Restaurant365’s industry data hub^[30] tracks similar adoption from the back-of-house side and cross-references with menu-engineering data, showing direct online orders carry meaningfully higher contribution margins than marketplace orders even at the same listed price point — because of the missing marketplace cut.

Figure 2 — % of US restaurants accepting direct online orders, 2019–2026. Source: NRA State of the Industry 2020–2026; Toast Industry Report.

The mix-shift is not zero-sum. Marketplace orders continue to grow in absolute terms — total US online food delivery is projected at ~$140B in 2026 by Restaurant Dive^[31]— but operators have correctly read the signal that the customer relationship lives wherever the loyalty record lives, and are working harder to make that the restaurant’s own CRM rather than DoorDash’s. The practical playbook is covered in our deep-dive on what to look for in an RMS in 2026.

5. AI ordering — chat, voice, predictive

Through 2024 most AI-ordering pilots were vendor demos and press releases; through 2026 the technology is in production. Three distinct surfaces are scaling.

Voice-AI drive-thru.The largest deployment cohort is the QSR drive-thru: Wendy’s FreshAI (Google Cloud), Bojangles, Carl’s Jr., and others have rolled out voice-AI to hundreds of locations through 2024–2026. Google Cloud^[32] publishes a customer page on the Wendy’s deployment; QSR Magazine^[33]has reported continuously on the rollout cadence and on the competitive responses from McDonald’s and Yum Brands.

Chat-based ordering.Outside the drive-thru, the higher-volume surface in 2026 is chat ordering — primarily over WhatsApp in LATAM and SMS / web chat in the US. Anthropic’s Claude API^[8]and OpenAI’s GPT family^[9] are the two reference models powering this category. The operator value is non-trivial: a single AI agent handles Spanish, English, and Portuguese on the same number, reads the menu correctly, and takes orders 24/7 without hold time.

Predictive ordering / kitchen prep. A third, more recent surface is predictive prep — using historical order data to forecast which items the kitchen will be asked to fire in the next 30 minutes and pre-stage them. Modern Restaurant Management^[34] covers this category continuously; the tooling is still relatively early but is the highest-ROI use of AI in the kitchen because it directly reduces prep waste at peak.

Adoption.The NRA’s 2026 technology survey^[6] reports 34%of US operators using or actively piloting AI in some form in 2026 — up from 16% in 2024. The most common entry point is the phone / chat ordering surface, not back-of-house, because it’s the highest pain point per labor dollar. Independent coverage in Restaurant Dive^[31] and NRN^[15] corroborates the doubling pattern.

Figure 3 — % of US operators using or piloting AI in some form, 2022–2026. Source: NRA technology surveys 2022–2026.

6. Payment processing — Stripe vs vendor-bundled, Tap-to-Pay

Payment processing in restaurants in 2026 has bifurcated into two distinct philosophies: vendor-bundled (the POS owns the merchant relationship, Toast Payments / Square Payments / Clover Payments / Lightspeed Payments) and operator-owned (the restaurant brings its own Stripe / Adyen / Worldpay merchant account, the software platform integrates).

Stripe. Stripe’s published online pricing^[35] — 2.9% + $0.30 for cards-not-present and 2.7% + $0.05 for Tap-to-Pay / in-person — is the de facto reference rate that the rest of the market is benchmarked against. Stripe Connect^[36] is what platforms like FoodyOS use under the hood to onboard a restaurant’s own merchant account, so payouts go directly to the operator’s bank rather than through the SaaS vendor.

Tap-to-Pay adoption. Apple’s Tap to Pay on iPhone^[37] opened in 2022, and by 2026 Square’s operating-business publication^[38] reports that the majority of new in-person merchants on its platform never buy a dedicated card reader. The same pattern holds for Stripe Terminal’s Tap to Pay surface^[39] on iPhone and Android. For an independent restaurant rolling out a tablet-based POS in 2026, the hardware question — did you bundle a proprietary terminal — is no longer the right question; the right question is who owns the merchant of record on the resulting transactions.

The vendor-bundled trade-off.Vendor-bundled processing is operationally easier on day one (one vendor, one statement, one integration), and where a vendor publishes a rate at all it is broadly competitive with Stripe’s. The first problem is that most of them do not: of the five POS vendors in §3, only Square publishes a restaurant card rate on its pricing page. The second is portability — the restaurant’s processing history, dispute history, and customer payment tokens belong to the POS vendor. Operators who care about multi-vendor optionality (which is most of the multi-unit segment) increasingly choose Stripe Connect-based platforms; operators who want one bill choose the bundle. Modern Restaurant Management^[34] covers the trade-off continuously.

For the operator playbook, see our blog post on why owning your Stripe account matters for direct ordering.

Figure 4 — Card-present rates each processor publishes itself, as of August 2026. Sources: Stripe, Square. Square’s plotted rate is its Free-plan in-person rate (2.5% + 15¢ on Plus, 2.4% + 15¢ on Premium). Only two bars is the finding, not an omission: Toast publishes no card rate anywhere on its pricing page — an earlier edition of this report plotted a bar for Toast anyway, and that figure was not Toast’s to plot, so it has been removed. Lightspeed publishes no processing rate on its restaurant pricing page either, and Clover quotes card-present pricing only inside a financed hardware bundle. An estimate in any of their places would be a guess. Per-transaction fixed fees (5¢–30¢) not shown.

7. Restaurant closures — rate, lifetime, factors

The single most-cited statistic about restaurants — that 60% close within the first year — is not accurate and never has been. The real numbers from the Bureau of Labor Statistics Business Employment Dynamics^[40] series, which tracks every employer establishment in the US including restaurants, show roughly 17% of new restaurants close in their first year and 50% within five years — which is approximately the same survival profile as US small business overall, slightly worse than retail, and better than the urban-legend version.

2024–2026 trend. NRA Economist’s Notebook^[41] tracks net openings vs closings monthly. Through 2024 and most of 2025, the US ran net-positive — more restaurants opening than closing — but with elevated closure rates in the highest-cost-of-occupancy markets (NYC, LA, Bay Area, Seattle). Yelp’s Local Economic Impact Reports^[42] cross-reference openings / closings against consumer-search demand and identify the same pattern: concept failure correlates more with rent + labor cost exposure than with cuisine type.

Drivers in 2026. The NRA State of the Industry 2026^[6] lists the top operator-cited closure drivers, in order: (1) labor cost and availability, (2) food cost inflation, (3) occupancy / rent, (4) third-party marketplace fees, and (5) credit-card processing. Notably, marketplace fees have moved up the list each year since 2020 even as marketplace volumes have grown — operators report that the all-in commission has become a structural margin drag rather than a marginal customer-acquisition cost.

8. Latin America snapshot — Brazil, iFood, WhatsApp ordering

Latin America has its own restaurant tech narrative, and Brazil is the loudest data point. iFood is the dominant marketplace by an unusually wide margin — iFood^[43] reports 100M+ monthly orders and over 350,000 restaurants on the platform, making it one of the highest-share food delivery platforms in any large national market in the world.

iFood economics. iFood publishes two restaurant plans on its partner pricing page^[13]: 12% (Plano Básico), 23% (Plano Entrega) — the cheaper plan for restaurants delivering with their own couriers, the dearer one when an iFood courier delivers — plus a 3.2% platform fee on orders paid through iFood, and a monthly subscription of R$ 110 on Básico (R$ 150 on Entrega) charged only once monthly sales pass R$ 1.800. That puts the published all-in take between 15.2% and 26.2% of the ticket before any optional promotion participation. Earlier editions of this report quoted a higher ceiling than that; it could not be traced to anything iFood publishes, so it has been removed. Two things cut the other way and belong in the same paragraph: 0% commission below 30 orders/month on both plans, published as a limited-time offer. No lock-in: 'Cancele o plano quando quiser.' The Brazilian Restaurant Association (Abrasel)^[11] has tracked what operators actually end up paying since 2020.^[12] (iFood partner pricing page, August 2026.)

WhatsApp ordering shift. DataReportal Digital 2026^[44] puts WhatsApp at 99%+ messaging penetration among Brazilian internet users, the highest of any major market. That ubiquity is what makes WhatsApp ordering work operationally for Brazilian restaurants — there is no onboarding friction. Mesheet^[45] and other regional trackers have logged a measurable shift in mid-market Brazilian restaurants moving order volume from iFood to WhatsApp + their own site, particularly for repeat customers where loyalty makes the marketplace introduction unnecessary.

Mexico, Colombia, Argentina.Rappi is the regional marketplace equivalent across Spanish-speaking LATAM. It publishes no merchant commission schedule we could retrieve, so this report makes no claim about its take rate rather than borrowing iFood’s. The DataReportal LATAM section^[44] of the 2026 Digital report captures the per-country messaging adoption — WhatsApp leads in every LATAM country tracked — which is why WhatsApp-AI ordering has the same operator pull there as in Brazil.

For a deeper take on the WhatsApp vs iFood economics, see our blog post on pedidos pelo WhatsApp vs iFood.

9. Methodology and sources

Aggregation principle.Every figure cited in this report comes from a publicly accessible primary source — vendor pricing pages, SEC filings or shareholder letters, the National Restaurant Association’s State of the Industry, the Bureau of Labor Statistics Business Employment Dynamics series, Abrasel publications, or published trade-press tracking. We do not include proprietary survey data, paywalled research, or vendor-supplied private numbers.

Marketplace rates. Every marketplace figure in §2 and §8 is a rate the platform states on its own merchant pricing page, quoted per plan and cheapest plan first, with the retrieval date printed alongside it. This edition no longer computes an effective take rate. The previous edition modelled one on a $30 ticket by adding a “published payment processing rate” to the plan commission — but DoorDash and Uber Eats state processing is already inside the commission without publishing a rate, and Grubhub excludes it without publishing one either. There was no published number to add, so the model has been withdrawn rather than re-based on a guess. Where operators report an all-in cost above the listed plan rate, this report cites the trade press that reported it and does not restate it as arithmetic of ours.

POS location counts. Every POS figure in §3 is a number the company itself publishes — Toast and Lightspeed from investor relations, Block from its quarterly results, Fiserv for Clover, TouchBistro from its corporate site. None of them is restaurant-only and US-only, so we print them side by side and label what each one actually counts rather than estimating a restaurant-vertical subset or a share-of-market split. Broader third-party rollups are linked for readers who want them — Datanyze, BuiltWith, and G2’s Restaurant POS Grid — but their bases differ again, so we do not blend them into ours.

License. This report is published under Creative Commons CC BY 4.0. You may quote, republish, or redistribute its data freely with attribution to FoodyOS and a link back to https://foodyos.com/state-of-restaurant-tech-2026. We refresh the figures annually; if you spot a number that has moved, email us and we will update with attribution.

10. Sources

Every URL referenced inline above is reproduced here for citation. Vendor and marketplace pricing pages (1, 2, 3, 13) were re-retrieved on August 8, 2026 and the figures in §1, §2 and §8 corrected against them; all other sources were accessed April–May 2026.