Appen

Appen is a publicly traded Australian data services company founded in 1996 by linguist Dr.

Updated 18 Jun 2026 Reviewed by 7wData

Visit Website

Profile

Appen provides human-annotated datasets and quality evaluation services to train and benchmark AI models, with particular strength in speech recognition, RLHF alignment, and reasoning-trace data across 180+ languages.

Appen is a publicly traded Australian data services company founded in 1996 by linguist Dr. Julie Vonwiller as a specialist in linguistic annotation and speech recognition. The company went public on the ASX in 2015 and has since expanded into a major provider of AI training datasets through a series of acquisitions, including Figure Eight (2019, $175M upfront) and Quadrant (2021, $25M).

Under CEO Ryan Kolln, who took the helm in February 2024 following a brief leadership transition, Appen operates a global crowd of over 1 million vetted contributors across 500+ locales and 180+ languages. The company's core offerings now span frontier model alignment (RLHF, reasoning traces, safety evaluation), agentic AI training data, speech and audio transcription, multimodal datasets for vision-language models, and robotics/physical AI datasets. Appen's financial trajectory reflects the volatility of AI-dependent customers: Google's contract termination in January 2024 (which represented roughly one-third of revenue, or $82.8M annually) caused the stock to drop 41% in a single day, but the company stabilized and repositioned.

FY 2025 revenue was $230.8M (excluding Google), up 4.5% year-over-year, with generative AI projects now driving 44.1% of Q4 revenue, up from 34.8% prior year. The company achieved $10M in annualized cost reductions through automation and technology initiatives. FY 2026 guidance targets $270–$300M revenue with 5–10% EBITDA margins.

Geographic diversification is a key post-Google priority, with China operations growing 74.8% in FY 2025 to $102.9M and now representing 45% of total revenue. The company remains profitable and competitive in the crowded AI data-labeling space despite ongoing workforce challenges on its CrowdGen platform, including worker concerns about project availability and pay compression since the platform transition in 2024.

Who buys this

Foundation model builders (LLM developers needing frontier alignment, RLHF, and reasoning data)
Large tech companies developing AI search, voice assistants, and generative applications
Enterprise and mid-market organizations building internal AI agents and agentic workflows
Companies developing multimodal and vision-language models requiring annotated image and video datasets
Robotics and autonomous systems companies collecting training data for embodied AI

Publicly disclosed clients

Google (contract ended March 2024)
Multiple undisclosed 'frontier model builders' (per company filings)

Strengths and what to watch

Strengths

Global crowd scale unmatched among competitors: 1M+ contributors across 180+ languages in 200 countries, enabling rapid scaling of specialized annotation tasks
Deep domain expertise in speech/audio transcription (30+ years), combined with modern platform capabilities for multimodal and reasoning-trace annotation
Geographic diversification and China momentum: China revenue grew 74.8% YoY to $102.9M (45% of total in FY 2025), reducing reliance on single large US/UK customers

Watch for

Customer concentration: Despite diversification, over 80% of revenue is concentrated among top 5 clients; loss of any single major customer (as demonstrated by Google contract termination) creates sharp financial volatility
CrowdGen platform worker experience: Crowd contributors report declining project availability, lower pay rates, and technical issues since 2024 platform transition; quality control and retention remain publicly scrutinized concerns
Competitive margin pressure: Rivals including Scale AI, Labelbox, and SuperAnnotate are consolidating around hybrid managed-services and platform models; Appen's primarily crowd-based model faces pricing pressure in favor of lower-cost synthetic data and in-house annotation teams

Recent moves

Key Information

Industry: Data Generation & Labelling
Founded: 1996

Frequently Asked Questions

What is Appen?

Appen is a publicly traded Australian data services company founded in 1996 by linguist Dr. Julie Vonwiller. It provides comprehensive AI training data and model evaluation services through human annotation, operating a global crowd of over one million vetted contributors across many languages and locales worldwide.

What products and services does Appen offer?

Appen delivers RLHF and frontier model alignment, agentic AI training data, speech and audio transcription, multimodal datasets, and robotics or physical AI datasets. These offerings support large language model training, multilingual speech recognition, reasoning trace annotation, and overall model quality evaluation for customers.

What is Appen's global capacity and expertise?

Appen operates a global crowd of over one million vetted contributors spanning more than 500 locales and 180 languages. This scale, combined with 30-plus years of speech and audio expertise, supports multilingual datasets, crowd-based labeling, and diverse data collection across geographic regions worldwide.

What are Appen's recent financial results and outlook?

Generative AI projects drove 44.1% of Appen's fourth-quarter revenue, up from 34.8% previously. FY 2025 revenue reached $230.8 million, growing 4.5% excluding Google. The company guides FY 2026 revenue between $270 and $300 million, reflecting momentum in generative AI demand.

How important is Appen's China business?

Appen's China operations grew 74.8% year-over-year to $102.9 million, representing 45% of total revenue. This expansion reflects strong regional momentum and geographic diversification, helping offset the loss of the Google contract that previously contributed roughly one-third of company revenue annually.

What are Appen's main business risks?

Appen faces customer concentration risk, with 80% of revenue from its top five clients. The Google contract ending March 2024 caused a 41% stock drop. Additional risks include CrowdGen platform worker issues and competitive pressure from Scale AI and synthetic data alternatives.

How Appen compares

Direct head-to-head against 3 competitors. Picked by 7wData.

Positioning: Appen provides human-annotated datasets and quality evaluation services to train and benchmark AI models, with particular strength in speech recognition, RLHF alignment, and reasoning-trace data across 180+ languages.
Customer segments: Foundation model builders (LLM developers needing frontier alignment, RLHF, and reasoning data)
Strengths: Global crowd scale unmatched among competitors: 1M+ contributors across 180+ languages in 200 countries, enabling rapid scaling of specialized annotation tasks
Watch for: Customer concentration: Despite diversification, over 80% of revenue is concentrated among top 5 clients; loss of any single major customer (as demonstrated by Google contract termination) creates sharp financial volatility
Recent moves: Appen reports FY 2025 results: $230.8M revenue (+4.5% ex-Google), FY 2026 guidance $270–$300M

Positioning: Enterprise data labeling platform combining ML automation with human review, positioned above Appen on automation depth and foundation model client reach.
Customer segments: Foundation model builders (OpenAI, Nvidia), autonomous vehicle teams, robotics firms, and US defense agencies via procurement or direct contract.
Strengths: Hybrid ML plus human-in-the-loop pipeline sustaining near-perfect label accuracy across text, image, video, and 3D sensor data at enterprise volume.
Watch for: Meta's 49% stake caused OpenAI and Google to reduce or pause engagements over data confidentiality, concentrating revenue risk in one strategic partner.
Recent moves: Meta acquired a 49% non-voting stake for $14.3 billion (June 2025), valuing Scale at $29 billion. Founder Wang stepped down as CEO.

Positioning: API-first annotation platform for in-house ML teams, billing per screen-activity time, contrasting with Appen managed labor model.
Customer segments: Fortune 500 enterprises, LLM and foundation model builders, in-house ML teams with DataOps capability requiring platform control.
Strengths: Model-assisted labeling via natively integrated foundation models cuts annotation time up to 70 percent, keeping efficiency gains with the customer.
Watch for: Usage-based LBU pricing creates unpredictable costs at scale, with customers reporting surprise bills and pressure to negotiate volume lock-in clauses.
Recent moves: Acquired Upcraft, an agentic sales automation startup, in February 2026 to scale its Alignerr expert network for AI model training.

Positioning: Legacy data labeling provider differentiating on ethical labor practices and workforce quality, serving 30% of Fortune 50 clients.
Customer segments: Fortune 50 enterprises, foundation model builders including OpenAI, and AI/ML teams in retail, financial services, and healthcare.
Strengths: Expert-led human-in-the-loop annotation with a 70,000-person global workforce delivering over 40 billion data points.
Watch for: Meta terminated its contract in 2026 after Sama workers in Kenya viewed private user content, exposing serious governance and privacy risks.
Recent moves: Launched Bulk Annotation in November 2025, using ML grouping to cut repetitive labeling work and deliver 80% throughput gains.

Sources

appen.com — Company overview, products (frontier alignment, RLHF, multimodal, physical AI, safety), 30-year heritage, global reach (1M+ contributors, 500+ locales)
en.wikipedia.org — Founding (1996 by Julie Vonwiller), ASX listing (January 2015), key acquisitions (Figure Eight 2019, Quadrant 2021), current CEO Ryan Kolln (February 2024)
www.cnbc.com — Google contract termination (January 2024, effective March 19), $82.8M annual revenue impact, 41% stock decline
www.appen.com — Financial results, ASX ticker APX, investor relations resources, annual reports
restofworld.org — Context on Google contract termination and Appen's business model pivot
www.tipranks.com — FY 2025 results, FY 2026 guidance ($270–$300M), generative AI revenue contribution (44.1% Q4), China growth (74.8% YoY)