Transposable elements (TEs) are repetitive DNA sequences typically silenced in normal tissues. Their dysregulation in cancer can significantly impact oncogene activation and tumorigenesis. However, due to their repetitive nature, TEs are often excluded from gene-centric single-cell analyses. Although recent advances have enabled single-cell TE quantification, a systematic resource for exploring single-cell TE dynamics in cancer has remained lacking. To address this, we developed TE-SCALE (https://ngdc.cncb.ac.cn/te-scale/), a single-cell database for integrative analysis and visualization of TE expression across human cancers. It incorporates over 1.3 million cells from 330 samples across 20 cancer types and 12 tissue origins. Built on our in-house pipeline scTEfinder, TE-SCALE provides a comprehensive pan-cancer TE expression atlas, enabling multi-scale exploration from tissue to cell population, supported by well-curated TE annotations. The platform offers three key analytical modules: differential TE expression, TE-gene co-expression network, and functional enrichment analysis. A user-friendly web interface supports flexible browsing, searching, analysis, and data download. Notably, TE-SCALE identifies tumor-specific TEs preferentially expressed in particular cancer types or disease states, underscoring their potential as biomarkers for diagnosis, monitoring, and immunotherapeutic targeting. Collectively, TE-SCALE provides an essential resource for decoding TE biology in cancer and expedites its translation into clinical applications.