WillItSurvive

Human data labeling and annotation vendor for AI model training.

Little protection left
Assessed 26 July 2026 · Evidence as of 26 July 2026
Not much stands between this business and its pressures.
On the recordSince 26 July 2026 · Checkable from 28 August 2026 · Every call on the record Checkpoint: H1 FY26 results, testing whether Appen Global's decline has stopped and China growth holds
The call

Appen's Western labeling business is shrinking fast while its China unit grows, leaving the company's future riding on one segment.

Appen pays people to label images, text, and other data that trains AI systems. The problem is that this work is becoming easier to automate: new AI tools can now generate fake training data or label data themselves, and competitors like Surge AI and Mercor have stolen Appen's best customers by offering higher quality work. Appen's revenue has crashed from $447 million in 2021 to $233 million today, its biggest customer (Google) is gone, and the company is pinning all its hopes on fast growth in China, which could reverse or slow down at any time.

Why
01
AI pressure: OpenAI, Anthropic, and Google synthetic data and RLAIF self-grading pipelines. Labs increasingly generate and self-grade their own training data instead of buying human labels, shrinking Appen's core annotation market. Could AI do the core job? Yes, likely.
02
Bigger players: Google, plus rivals Surge AI and Mercor. Google dropped Appen as a customer, and frontier labs now route expert-eval work to Surge AI and Mercor instead. Could a giant copy it? Yes, easily.
03
What protects it. The China unit, up 75% YoY to $102.9M in FY25, is the only growing segment keeping total revenue from collapsing further.
What would change this call
Appen H1 FY26 results released 28 Aug 2026 show Appen Global revenue decline rate. Shows whether the Western commodity-labor business has stabilized or is still losing customers.
Appen China quarterly growth rate reported in H1/H2 FY26 filings. Shows whether the single growth segment keeping Appen viable is decelerating.
Any announced Appen contract win with a frontier AI lab at Surge AI/Mercor tier. Would show Appen regaining ground in higher-value expert-eval work it has been losing.
Key numbers
  • Revenue fell from $447M (2021) to $233M (FY25) (2025-12-31)
  • Appen Global (ex-China) revenue down 37% YoY (2026-04-30)
  • China revenue up 75% YoY to $102.9M, FY25 (2025-12-31)
Named risks
  • Appen Global segment down 37% YoY in Q1 FY26, indicating core Western business contraction. Severity: high
  • Growth strategy concentrated in single volatile China segment (+75-88% YoY). Severity: high
  • Repeated dilutive capital raises (2023, 2024, ongoing performance-rights issuance) eroding per-share value. Severity: medium
  • Governance overhang from near-strike remuneration votes (27.74% against in 2025) and 2026 board-spill resolution. Severity: medium
  • Core commodity labeling work exposed to automation by synthetic data and model-based evaluation. Severity: high
Rivals, as read in July 2026
RivalDirectionEvidence
Scale AIlosing toScale AI described as far larger and better funded than Appen as of 2026-07-02 (Sacra)
Google (customer, not rival)losing toLoss of Google contract materially depressed Appen's FY24 revenue comparatives
Surge AIlosing toAnalysis indicates Surge AI has displaced Appen from frontier-lab expert-eval work, no fixed date
Mercorlosing toAnalysis indicates Mercor has taken quality-over-quantity expert-eval work from Appen
TELUS InternationalcontestingTELUS operates comparable managed data-labeling and AI data services offering
Questions people ask
Is Appen profitable in 2026?
No. Appen reported a net loss of approximately $21.8M for FY25 (ended 31 Dec 2025), and its Appen Global segment posted a $3.1M EBITDA loss in Q1 FY26 even as the China segment turned profitable, per company filings dated 2026-04-30.
Could AI replace Appen's business?
Appen's core work, high-volume human labeling and transcription, is directly exposed to synthetic-data generation and model-based evaluation. Group revenue fell from a $447M peak in 2021 to roughly $233M in FY25, a decline analysts tie partly to automation eroding commodity labeling demand.
Who is Appen's biggest competitor?
Scale AI is described as far larger and better funded than Appen as of July 2026 (Sacra research, 2026-07-02). Appen also lost Google as a major customer, a loss that materially depressed its FY24 revenue comparatives per company transcripts.
Who owns Appen Limited?
Appen is publicly traded on the ASX under ticker APX with no controlling parent identified. Its market capitalization was roughly A$220-236M as of 2026-07-20, according to stockanalysis.com.
Is Appen growing or shrinking in 2026?
Both, split by segment. Appen China revenue reached a run-rate above $144M annualized with 88% YoY quarterly growth as of Q1 FY26 (2026-04-30), while Appen Global revenue fell 37% YoY to $19.9M in the same quarter.
Has Appen been diluting shareholders?
Yes. Shares outstanding grew from about 123M in 2021 to roughly 264-268.5M by mid-2026, following equity raises including an A$30M raise in November 2023 and an A$55M placement plus SPP in October 2024 at A$1.92 per share.
Timeline
  • 2026-07-12Appen grants 6.59M performance rights to employees under incentive plan
  • 2026-06-30Total unquoted performance rights reach ~28.6M against ~268.5M ordinary shares
  • 2026-05-22AGM board-spill resolution defeated, 91.31% voted against; Vanessa Liu confirmed as chair
  • 2026-04-30Q1 FY26 report: China revenue run-rate exceeds $144M annualized; Global revenue down 37% YoY
  • 2026-02-25FY25 Appendix 4E filed: group revenue $230.8-233.4M, net loss ~$21.8M
  • 2026-01-01Board chair transitions from Richard Freudenstein to Vanessa Liu
  • 2025-05-2327.74% of AGM votes oppose FY2024 remuneration report
  • 2024-10-01Appen raises A$55M via placement and share purchase plan at A$1.92/share
  • 2024-02-01CEO Armughan Ahmad resigns; COO Ryan Kolln becomes CEO
  • 2023-11-21Appen completes fully underwritten A$30M equity raise
Revision history
26 July 2026
First assessed. Little protection left. The text above is fixed as of this date and is not edited afterwards.
In the same band
Sources