CyberLink subscribers had higher paid-AI overlap than Adobe subscribers in 2025
Share of established subscribers who remained active in the panel and had a standalone paid generative-AI receipt during each full calendar year
Share with paid-AI receipt
Source: YipitData U.S. e-receipt panel. CyberLink 2025 n=1,070; Adobe 2025 n=85,083. CyberLink cohorts ranged from about 875–1,100 subscribers annually.
Chart data
| Series | Share with paid-AI receipt |
|---|---|
| CyberLink — 2023 | 16.7% |
| Adobe — 2023 | 15% |
| CyberLink — 2024 | 15.7% |
| Adobe — 2024 | 10.7% |
| CyberLink — 2025 | 30.1% |
| Adobe — 2025 | 18.3% |
Which paid-AI services produced the 2025 difference?
ChatGPT appeared for 19.4% of CyberLink subscribers and 11.4% of Adobe subscribers. Gemini or Google AI appeared for 10.5% and 4.8%, Perplexity for 4.8% and 2.6%, and Claude for 4.4% and 3.3%, respectively. Users can appear in more than one service rate.
Did earlier paid-AI holders show lower CyberLink renewal?
Among CyberLink subscribers active in 2024, 30.6% of 173 paid-AI holders had another CyberLink receipt in 2025, compared with 32.4% of 926 subscribers without paid AI. With only 173 paid-AI subscribers, the 1.8-percentage-point gap was too small to support a meaningful difference.
Does the Adobe comparison explain why the cohorts differed?
No. Adobe subscribers may differ from CyberLink subscribers in profession, income, product mix, and other unobserved traits. The comparison documents a difference in payments to outside AI providers; it does not show that either company’s features caused AI adoption or that paid AI replaced creative-software use.
Assumptions & Methodologies
We compared established U.S. e-receipt panelists with CyberLink or Adobe creative-product receipts in full calendar years 2023–2025. CyberLink included PowerDirector, PhotoDirector, Director Suite, ColorDirector, AudioDirector, and PowerDVD; Adobe included six creative products. External paid AI covered ChatGPT, Gemini, Claude, Perplexity, Copilot Pro, and Midjourney. The sequential check followed 2024 CyberLink subscribers into 2025. App-store renewals can be missed, cohorts were not demographically matched, the AI-holder retention cell was modest, and receipts measure payment rather than use. Results do not establish causality or substitution.