Teardown

Retail / Ecommerce · Deep dive

AiFi

Santa Clara autonomous-retail platform from Duke computer-vision PhDs Steve Gu and Ying Zheng, powering 300+ camera-only cashierless stores globally by September 2025 across Zabka Nano, Aldi pilots, Verizon's Destination Store, Compass cafeterias and 74+ NFL/NBA stadium concessions — the last well-capitalised challenger left standing after Grabango's October 2024 collapse and Amazon's retreat from running its own Just Walk Out stores.

emerging

The question that decides it: AiFi has out-survived the first cashierless generation — Grabango shut in October 2024, Amazon retreated Just Walk Out from Amazon Fresh the same year, and Standard AI abandoned autonomous checkout for vision analytics — but the category is small, computer vision remains fragile at supermarket-basket size, and the buyer story has narrowed from 'reinvent grocery' to 'concession stands and c-stores under 3,000 sq ft.' **For AiFi to be a durable business rather than a graceful exit into Amazon or Microsoft, three things must be true by year-end 2027: (1) at least one anchor format — stadium concessions or Zabka-style c-store nano-formats — has to prove positive store-level unit economics (transaction-time and revenue-per-cap lifts that outrun licensed CV-per-store fees plus shrink), (2) AWS Just Walk Out's third-party licensing push, above 240 stores globally by end-2024 per Amazon, must not become the default procurement bundle, and (3) AiFi's 105-employee headcount (mid-2025, down from ~145 in 2023) must be a sign of gross-margin discipline rather than the same slow-motion cash burn that killed Grabango.**

My take

HQ
Santa Clara, CA
Founded
2016
Ownership
VC-backed private — Qualcomm Ventures, Cervin Ventures, Translink Capital, ALDI South, Zabka Group, REWE, Verizon Ventures, HP Tech Ventures, Plum Alley, Plug and Play
Funding
~$90.7M total through August 2024 (Tracxn/Crunchbase): ~$4M seed (2017-2018) + $11M Series A led by Cervin (2019) + $14.5M Series A+ led by Qualcomm Ventures (Oct 20, 2020) + $65M Series B (Mar 11, 2022) + $6.7M Series B extension (Aug 7, 2024) at a reported $100M pre-money vs. $250M in the Series B
Valuation
Reported ~$100M pre-money at the August 2024 extension; ~$250M pre-money at the March 2022 Series B (per investor deck circulated to LPs, mid-2024)
Revenue
~$24.2M reported (Kona Equity, associated with 2024 filings); 1,208% three-year revenue growth 2021-2024 per Inc. 5000 (August 2025). No audited disclosure. August 2024 investor materials targeted $30M TTM revenue and $12M ARR run-rate; Series C targeted early 2025, not publicly announced through September 2026.
Headcount
~105 (2025, per Revelio Labs, down 15.2% from 2024 and 34%+ from a ~145 peak in 2023); ~136 on LinkedIn as of June 2025; ~95 by early 2026 per Revelio
Screen
Scaled private — near $100M aggregate raised across primary and follow-on strategic rounds; recognized on 2025 Inc. 5000 at #326 with 1,208% three-year revenue growth
Published
2026-09-10
Web
aifi.com
Elsewhere
LinkedIn · Crunchbase

Founders and leadership

  • Steve Gu Co-founder & former CEO (stepped down September 2025)

    PhD in computer science from Duke; earlier a senior engineer at Apple working on 3D Touch and Force Touch, then at Google X on Google Glass computer-vision applications (Microsoft Startups profile, 2019; TechCrunch, February 2018). Founded AiFi with his wife Ying Zheng in January 2016.

  • Ying Zheng Co-founder & President

    PhD in computer vision from Duke; EMBA from Stanford GSB; earlier a research scientist at Google Research and later senior research scientist at Apple, per Nasdaq Entrepreneurial Center profile (December 2022). Runs commercial, partnerships and international expansion.

  • João Diogo (JD) Falcão CEO from September 2025; previously CTO and technical founder

    PhD in electrical and computer engineering; joined AiFi as technical founder and led engineering for nearly a decade. Named inventor on multiple AiFi patents. Promoted to CEO on September 11, 2025 succeeding interim CEO Steve Carlin (Translink Capital venture partner), per Retail Technology Innovation Hub.

Snapshot

AiFi is a decade-old Santa Clara CV company — the largest independent operator of camera-only cashierless stores worldwide — 300+ deployments by September 2025 across Zabka Nano (Poland), Aldi and Carrefour pilots, Verizon’s Destination Store, Compass and Sodexo cafeterias, and 74+ NFL/NBA stadium concessions. Co-founded January 2016 by Duke PhDs Steve Gu and Ying Zheng (ex-Apple/Google); led since September 2025 by CEO João Diogo Falcão. Total capital raised ~$90.7M through an August 2024 Series B extension — a down-round at ~$100M pre-money vs. $250M in 2022. AiFi is alone in a category with a fresh graveyard (Grabango shut October 2024, Standard AI abandoned autonomous checkout, Amazon pulled Just Walk Out from Amazon Fresh) — the challenger left standing; the question is whether the market it stayed in is a market at all.

Founding story

Gu and Zheng met at Duke, married, and by the mid-2010s were senior researchers inside Apple and Google. Gu worked on Apple’s 3D Touch/Force Touch, then Google X on Google Glass; Zheng was at Google Research, then senior research scientist at Apple. Their thesis (Microsoft Startups 2019; Nasdaq Center 2022): the future of AI is in the physical world, and retail is the most data-rich physical space available. AiFi launched publicly February 2018 (TechCrunch), the same month Amazon opened the first Amazon Go — positioned as the independent, sensor-agnostic alternative to Just Walk Out (no retailer would want to hand Amazon their store-level data). João Diogo Falcão joined as technical founder early. Gu stepped down September 2025; interim CEO Steve Carlin (Translink partner) handed off to Falcão on September 11, 2025 (RTIH). Zheng is now President.

How it works

An AiFi store is a ceiling-mounted camera array wired to on-site or edge-compute GPUs. Camera-first and sensor-agnostic: reuses existing CCTV where possible, adds off-the-shelf ceiling cameras where not. Demo stores use ~109 cameras (Retail Brew, September 2022). No shelf-weight sensors, no shelf cameras, no gate arm: the shopper taps a payment method at entry, keypoint-tracking creates an anonymous skeletal identity, camera streams track what leaves the shelf, and on exit the payment is charged. Verizon 5G MEC lets newer deployments skip on-prem servers, cutting hardware/maintenance ~60%. The engine is eddie — Event Driven Data Interpretation Engine — on Azure, integrated with Microsoft Copilot (Microsoft customer story, 2025). AiFi is deliberately lighter than Just Walk Out (which adds shelf-weight sensors and gate arms atop ceiling cameras) — fewer touchpoints, faster deployment, lower capex, and fewer redundant signals when the CV model is uncertain which of two similar SKUs was picked.

Product and business overview

Four named formats. NanoStore — small-footprint standalone (Zabka Nano, 50+ stores by mid-2022 and larger since). Grab & Go — store-within-a-store inside Verizon’s Destination Store and stadium concourses. To-Go — mobile/pop-up on Verizon 5G MEC, dropped into outdoor venues without wired servers. Spatial Intelligence — the analytics layer the company is now pushing hardest (shopper flow, dwell, planogram compliance, sold-out detection), sold whether the store is cashierless or not. The 2025 CEO change and Microsoft partnership point the same way: AiFi is repositioning from a checkout company to a spatial-data company.

Business model and pricing

AiFi does not publish pricing. Investor decks describe a hybrid: per-store setup, monthly per-store SaaS for CV runtime and analytics, and in some deployments a transaction-linked fee. Unit economics turn on cameras per store (100+ for a Nano), whether existing CCTV is reused, whether 5G MEC replaces on-prem GPUs, and whether transaction volume beats a part-time cashier. August 2024 investor materials targeted $30M TTM revenue and $12M ARR; Kona Equity lists ~$24.2M against 2024 filings; Inc. 5000 assigned 1,208% three-year growth (August 2025). Revenue exists and grows fast; scale is still modest at year nine.

Traction over time

DateMilestone
Jan 2016Incorporated by Gu and Zheng
Feb 2018Emerges from stealth same month as first Amazon Go
Oct 2020Series A+ (Qualcomm); total ~$30M
Jun 2021First Zabka Nano opens in Poznan
Mar 2022$65M Series B; ~$250M pre
Aug 202280 stores worldwide
Feb 2023100+ stores globally
Dec 2023Eagles/Verizon at Lincoln Financial Field
Aug 2024Series B extension $6.7M at ~$100M pre — ~60% down round
Oct 202474+ stadium stores by year-end; 236% YoY growth
Aug 2025Inc. 5000 #326; 1,208% 3-yr revenue growth
Sep 2025Falcão CEO; 300+ stores globally
Feb-Apr 20267-Eleven ramps AiFi stores; BlockStreet acquisition

Market analysis

The addressable universe is smaller than pitch decks suggest. eMarketer, LossPreventionMedia and Wharton all put cashierless below 1% of global in-store retail through the end of the decade. Near-term TAM is small-footprint formats where a single cashier is a meaningful cost line and basket size is small enough for CV accuracy (c-stores, campus micro-markets, stadium concessions, airport grab-and-go) in markets where labour is expensive and turnover chronic. Zabka Nano and US stadium concessions are the two proven wedges. Grocery, once treated as the endpoint, is the graveyard — Amazon abandoned it at Fresh in 2024.

Competitive intel

Amazon Just Walk Out is the structural competitor; its 2024 pivot changed the game — Amazon is now a hyperscaler licensor with 240+ third-party stores globally by year-end 2024 (CNBC, October 5, 2024) bundling AWS, identity, billing and CV in one procurement. Trigo ($194M raised) has the same camera-plus-CCTV pitch, anchored by Tesco. Zippin ($45M) competes directly on stadiums and airports. Grabango — dead October 9, 2024 — is the cautionary comparable. Standard AI pivoted to Vision Analytics, a warning even the best-funded US pure-play walked away. Shopic and Caper Cart are the smart-cart wedge Amazon itself validated with Dash Carts. Silent competitors: RFID plus phone scanning (Uniqlo, Zara, Sam’s Club Scan & Go) ships today and never asks the customer to trust a CV algorithm.

History and evolution

Series A 2019 (Cervin) and Series A+ October 2020 (Qualcomm) build the platform through the pandemic. First Zabka Nano opens Poznan, June 2021 — vaulting AiFi to Europe’s largest cashierless chain within twelve months. $65M Series B March 2022 (ALDI South, Zabka, REWE, Verizon, HP, Qualcomm, Cervin) at ~$250M pre is the high-water mark.

Then the category compresses. Amazon closes Go and Fresh cashierless locations through 2023-2024 and by late 2024 removes Just Walk Out from Fresh in favour of Dash Carts (CNBC, October 5, 2024). Standard AI shifts to Vision Analytics with layoffs (CSP Daily News). Grabango permanently shuts October 9, 2024 (CNBC). AiFi survives by raising a $6.7M Series B extension in August 2024 at ~$100M pre-money — 60% below the 2022 mark — from Translink and Cervin, while pivoting from grocery to c-store, stadium, workplace and analytics. Gu steps down September 2025; Falcão takes over. Headcount peaks near 145 in 2023, drops to ~105 in 2025 and ~95 by early 2026 (Revelio) — controlled retrenchment, not a cliff.

What people say

The case for. Zabka credits AiFi with Europe’s largest autonomous chain (~50 Nano stores, mid-2022). Detroit Lions data at Ford Field: AiFi cut queue time 50% and lifted revenue per cap 170%. Verizon’s Destination Store (AKQA case) positions AiFi as the checkout layer differentiating the flagship. Microsoft’s 2025 customer story: eddie on Azure processes ~90 PB of spatial data annually with Copilot on top. Inc. 5000 #326 (August 2025) corroborates compounding revenue.

The complaints. CV accuracy at supermarket basket size — 30+ SKUs, similar variants, crowded aisles — is the recurring technical complaint. LossPreventionMedia and Wharton put self-checkout and cashierless shrink at 3.5-4% vs <1% for cashiered lanes; AiFi publishes no audited data, and Amazon’s 2024 pullback from Fresh is the most credible admission that CV alone struggles at supermarket scale. Labour scepticism: cashiers do not disappear from P&Ls, they move to restocking, security and confusion resolution. Customer confusion: shoppers who misplace items, share payment methods or wear occluding bags generate false charges. Biometric privacy — AiFi says anonymous keypoint tracking only, no facial recognition — is a live objection for EU regulators and Illinois BIPA. The 60%-below-mark extension and headcount slide from ~145 to ~95 are the market’s own vote of no confidence.

Outlook: the open question

AiFi is the last well-capitalised independent CV challenger standing — an achievement, and a smaller prize than the 2022 Series B implied.

For AiFi to be durable rather than a graceful exit to Amazon, Microsoft or Compass/Verizon, three conditions must hold by year-end 2027. First, at least one anchor format must show positive store-level unit economics — stadium concessions (the 170% revenue-per-cap lift Detroit Lions reported) or Zabka-style Nano c-stores — meaning licensed CV-per-store SaaS plus shrink beats the fully-loaded cashier line it replaces. Second, AWS Just Walk Out’s third-party licensing push (240+ stores by end-2024) must not become the default procurement bundle — the CIO reflex to standardise on AWS is enough to end this without AiFi losing on technical merits. Third, the 105-employee headcount must be gross-margin discipline, not the slow-motion cash burn that killed Grabango — meaning the Series C targeted for early 2025 (not publicly closed by September 2026) closes at a mark that clears the 2022 Series B, or the company becomes cash-flow-positive. If all three hold, AiFi is a $300M-$500M strategic acquisition candidate. If any one fails, the graceful exit becomes the not-graceful one.

How to attack it

First, concede the CV-inference layer and win on services. Amazon’s 2024 pivot makes AWS Just Walk Out a licensed product; an attacker can build cashierless-stores-as-a-managed-service on top of it — site survey, install, POS/loyalty/planogram integration, staff training, shrink audit, 24/7 monitoring — priced per store per month plus a share of documented shrink recovery. It avoids the CV R&D moat, matches procurement to where CIOs sit, and eats AiFi’s professional-services revenue directly. Weaknesses this exploits: (a) AiFi asks retailers to buy an independent CV platform when the hyperscaler offers one (CNBC, October 5, 2024); (b) the Series B extension at 60% below the 2022 mark is a public credit signal customers use to negotiate (Fundz, August 7, 2024); (c) headcount fell 34% from the 2023 peak (Revelio), thinning the field-services bench large retailers need.

Second, unbundle the hardware. AiFi and Amazon both require 100+ ceiling cameras per store (Retail Brew, September 2022). A gate-less, shelf-only sensor stack — smart shelves with weight and NFC/IoT tags plus one overhead wide-angle camera for gait tracking, payment at exit — deploys in a fraction of the time without a rewire. The wedge is small-footprint formats (airport lounges, hospital cafeterias, hotel pantries, corporate micro-markets) where AiFi’s per-store camera bill is the biggest objection. This is what Amazon argued moving to Dash Carts: intelligence sometimes belongs on the object, not the ceiling. Third, analytics-only: spatial-intelligence SaaS on existing CCTV priced below AiFi’s bundled add-on. AiFi is pivoting to a data company just as the checkout segment contracts; RetailNext, Trax and Standard AI’s Vision Analytics sit here already.

Adjacent-segment play

The clearest adjacency is not another retail format — it is industrial and logistics spatial intelligence. AiFi’s core stack is anonymous keypoint tracking and event-driven interpretation of ceiling-camera streams; retrained, it is what warehouse safety, cycle-count robotics and dock-door optimisation need. Cognex, Osaro and RightHand Robotics operate nearby with less CV depth. Repackaging eddie for warehouse floors — worker-safety, pick-path analytics, ceiling-camera cycle counting — is a natural Azure extension and a much larger spend line than retail cashierless.

A second adjacency is airports and airline lounges — American, Delta, Marriott, Hilton and IHG run F&B micro-outlets where guests are pre-authenticated and cashier labour is expensive. A third is cannabis retail — age-gating, reconciliation and shrink pain where walk-out solves a real problem, though regulatory overhang caps ambition. What does not generalise is diverse SKUs and large baskets — full-line grocery, home improvement, apparel — where CV science is unsolved and every walk-out competitor has failed publicly. If AiFi becomes a big company, it will be because retail cashierless was the origin story, not the destination.

Sources and further reading

Capital history

DateRoundAmountValuationLead(s)
2017-2018 Seed ~$4M Undisclosed Plug and Play, angels, early Cervin check
2019 Series A $11M Undisclosed Cervin Ventures (lead); TransLink Capital, Plug and Play
Oct 20, 2020 Series A+ $14.5M Undisclosed Qualcomm Ventures (lead); Cervin, TransLink, Plum Alley, SeaX — bringing total to ~$30M
Mar 11, 2022 Series B $65M ~$250M pre-money (per investor deck referenced in mid-2024 extension memo) Qualcomm Ventures and Cervin co-led; ALDI South Group, Zabka Group, REWE, Verizon Ventures, HP Tech Ventures — total to ~$80M (TechCrunch, Mar 11, 2022)
Aug 7, 2024 Series B extension $6.7M ~$100M pre-money — a ~60% down-round from the 2022 mark Translink Capital and Cervin Ventures co-lead (existing investors); Fundz filing

Investors / owners: Qualcomm Ventures, Cervin Ventures, TransLink Capital, Verizon Ventures, ALDI South Group, Zabka Group, REWE Group, HP Tech Ventures, Plum Alley, Plug and Play, SeaX Ventures, Microsoft (strategic, non-equity)

Competitive set

  • Amazon Just Walk Out (AWS) — Amazon pulled Just Walk Out from all Amazon Fresh US stores by late 2024, replaced with Dash Carts, but doubled third-party licensed deployments to 240+ globally by year-end 2024 (CNBC, Oct 5, 2024). AWS bundles CV, edge, billing and identity into one procurement — the bundle AiFi and Microsoft try to counter with Azure + Verizon 5G MEC.
  • Trigo — Israeli rival; ~$194M raised total through 2024; anchored by Tesco (UK, both customer and investor) and Shufersal. Trigo's pitch is CV overlaid on existing CCTV — the same 'sensor-agnostic' framing AiFi uses. Strongest remaining independent challenger.
  • Zippin — US rival; ~$45M raised through 2024; focused on stadiums and airports — direct overlap with AiFi's fastest-growing segment. Smaller balance sheet but the same core business.
  • Grabango (defunct) — Berkeley challenger founded by Pandora co-founder Will Glaser in 2016; raised ~$73M cumulatively including a $39M round in June 2021 (Aldi US, 7-Eleven, Circle K, Giant Eagle). Announced permanent shutdown October 9, 2024, unable to raise a bridge (CNBC). The most important recent category data point: even with tier-1 grocery and c-store logos, the CV-per-store cost structure could not clear a venture bar.
  • Standard AI (formerly Standard Cognition) — Once the best-funded US CV challenger (~$235M raised through 2022, ~$1B valuation). Pivoted away from autonomous checkout in 2024 to 'Vision Analytics' — cameras for shopper behavior insights — with layoffs and a CEO change to former COO Angie Westbrock (CSP Daily News, 2024). No longer a direct competitor.
  • Shopic / Caper Cart (smart-cart alternative) — Israeli smart-cart maker Shopic (~$46M through 2023) and Instacart's Caper Cart take the opposite bet: put the intelligence on the cart. Amazon's own retreat to Dash Carts at Amazon Fresh in 2024 validates the smart-cart wedge as a competitor for the same 'reduce cashier labor' budget.
  • In-house RFID and phone-scan alternatives — Uniqlo and Zara have deployed RFID self-checkout at scale; Sam's Club Scan & Go works on phone-based scanning. For any single-format retailer with volume, RFID plus phone scanning ships today and never asks a customer to trust a CV algorithm.