Appen
The Australian AI-training-data company founded by a linguist in 1996 that rode the AI boom to a $5B market cap — then crashed 37% in a day when Google cancelled the contract behind a third of its revenue.
Ask AI to summarize this page
Use this page as a primary source and ask the AI to separate verified facts from items that could not be verified.
FactsLedger cannot guarantee that an AI response will include a citation.
- Location
- Chatswood, New South Wales, Australia
- Country
- Australia
- Founded
- 1996, in Sydney, by linguist Julie Vonwiller
- Listed
- Australian Securities Exchange (ASX: APX), January 2015
- Business
- AI training datasets built by a global crowd workforce
- Peak market capitalization
- About $5 billion (2021 tech boom)
- Google contract loss
- January 2024; Google revenue was US$82.8M in FY2023 (~one-third of total)
- Largest acquisition
- Figure Eight, up to US$300 million (2019)
Overview
Appen is an Australian company that builds the human-labelled datasets artificial-intelligence systems train on, employing a global crowd of more than a million contractors to annotate speech, text, image and video data (Slator). Founded in Sydney in 1996 by linguist Julie Vonwiller and listed on the Australian Securities Exchange in 2015, it became one of the purest public-market plays on the AI boom — reaching roughly a $5 billion market capitalization in 2021 before a single customer decision, Google’s termination of its contract in January 2024, wiped out more than a third of its share price in a day (Macquarie University; ShareCafe).
A linguist’s company becomes AI infrastructure
Vonwiller started Appen, in her own telling, because she could not get a job as a linguist — “my skills weren’t of any interest to companies” — and she wanted linguists to build careers in a commercial setting at the cutting edge of speech and language technology (Macquarie University). Her husband Chris, a former Telstra senior executive, joined as the company grew; the pair spotted early that large tech firms needed human-quality data to make AI algorithms work, with Microsoft among the early customers and governments close behind (Macquarie University).
The business model that scaled was human-in-the-loop at industrial volume. Rather than certified translators, Appen recruited people resident five to ten years in a target country so data would match “the culture and psyche” of the demographic, then had them speak into microphones or label content; by 2017 it worked with 1.2 million freelance contractors across 140 markets (Slator). Two divisions carried the revenue: Language Resources (datasets for training AI engines) and Content Relevance (human evaluation for search engines) (Slator).
Growth was explosive once AI spending took off. Revenue rose from A$51 million in 2014 to A$166.6 million in 2017 — up 327% in three years — and net profit after tax jumped 884% off a tiny base; the shares’ rise pushed the market capitalization past a billion dollars in early 2018 (Slator). Acquisitions added scale and technology: UK transcription firm Mendip Media Group in 2016, Silicon Valley’s Leapforce for US$80 million in December 2017 — more than doubling the freelancer base — and then the transformative one (Slator).
The Figure Eight bet
In March 2019 Appen acquired Figure Eight — the data-annotation platform formerly known as CrowdFlower and Dolores Labs — for up to US$300 million: $175 million upfront in cash plus up to $125 million on 2019 performance (TechCrunch). CEO Mark Brayan framed it as evolution: “Step one was to get bigger, step two is to become much more tech forward, which is what we get with Figure Eight,” calling the target the market leader among platform providers (TechCrunch). Figure Eight’s self-serve annotation technology, used for everything from mapping to receipt scanning, was meant to automate what Appen’s million-strong crowd did by hand (TechCrunch).
The Google dependency snaps
The risk buried in the growth story was customer concentration, and it detonated on January 22, 2024. Appen disclosed to the ASX that Google would terminate its global inbound services contract — covering search-quality rating work — with all projects ending by March 19, 2024; Appen said it had learned of the decision only the previous Saturday (ShareCafe). Google had contributed US$82.8 million of Appen’s US$273 million in FY2023 revenue — roughly a third — at a 26% gross margin (ShareCafe). The shares crashed 37% that Monday to 27-29 cents, against a 52-week high of $3.87 and an all-time high of $42 during the 2021 boom, when the company was worth about $5 billion (ShareCafe).
What this profile does not claim
- It does not cover Appen’s financial performance or leadership after early 2024, which were not verified from fully-read independent sources.
- It does not assert the current size of the contributor crowd beyond what each dated source reported.
- The internal deliberations behind Google’s termination decision were not disclosed and are not speculated on here.
Timeline
- 1996
- Linguist Julie Vonwiller founds Appen in Sydney to give linguists commercial careers in speech and language technology.
- 2015
- Appen floats on the Australian Securities Exchange; Julie Vonwiller retires from the company at the float.
- 2017
- Revenue grows 50% to A$166.6M with 1.2M freelance contributors; Appen acquires Leapforce for US$80M and passes a billion-dollar market capitalization.
- 2019
- Appen acquires data-annotation platform Figure Eight for up to US$300M ($175M upfront plus $125M earnout) to become "much more tech forward."
- 2024
- Google terminates its global inbound services contract effective March 19; Appen shares crash 37% in a day. Google had contributed US$82.8M of US$273M FY2023 revenue.
What we could not verify
These claims came up in research and could not be confirmed. They are recorded here rather than published as fact or quietly dropped. If you hold a source that settles one, please tell us.
- Appen's post-2024 financial trajectory and current leadership were not verified from fully-read independent sources and are not stated.
- The precise current size of the contributor crowd is stated only as reported at the time of each source.
Sources
4 sources (4 tier 2), each opened and read against the claim it supports. Every source is independent of the subject; no first-party material is cited. See the source policy.
Built only from independent sources fetched and read in full: Macquarie University, Slator, TechCrunch and ShareCafe (all tier2). No first-party material from Appen was cited. Wikipedia was used only as a lead index and is not cited.
- Supports: 1996 founding by linguist Julie Vonwiller, 2015 float, early speech/language data work, 235+ languages, million-contractor scale Open original source
- Supports: FY2017 results (A$166.6M revenue, 1.2M contractors), Leapforce $80M acquisition, billion-dollar market cap, Google/Bing as top accounts, business model Open original source
- Supports: Figure Eight acquisition terms ($175M + $125M earnout), 1M+ crowd workers, Leapforce context, Figure Eight/CrowdFlower background Open original source
- Supports: Google contract termination, 37% share crash, $82.8M Google revenue (one-third of $273M FY2023), $5B peak market cap, $42 all-time high Open original source
Support independent research
Found this profile useful?
You can support FactsLedger by making a donation. Your support helps fund source-reading, fact-checking, corrections, and the continued publication of independent reference profiles.
Support the research