This position piece advocates for increased research attention on tabular foundation models (LTMs), which are underexplored compared to text and image models despite being the dominant modality in many fields. The authors argue that LTMs could revolutionize how tabular data is used, enabling contextualization with related datasets and offering applications such as few-shot learning, data augmentation, and automated meta-analyses. They highlight the potential impact of LTMs on scientific discovery, privacy, and reproducibility. The paper discusses the challenges and opportunities in developing LTMs, including handling mixed-type columns, cross-dataset modeling, textual context, and invariance to column order. It also explores the current state of LTMs, their real-world applications, and the need for robust evaluation methods. The authors conclude by comparing the impact of LTMs and large language models (LLMs) and emphasize the importance of shifting research priorities to focus on LTMs.This position piece advocates for increased research attention on tabular foundation models (LTMs), which are underexplored compared to text and image models despite being the dominant modality in many fields. The authors argue that LTMs could revolutionize how tabular data is used, enabling contextualization with related datasets and offering applications such as few-shot learning, data augmentation, and automated meta-analyses. They highlight the potential impact of LTMs on scientific discovery, privacy, and reproducibility. The paper discusses the challenges and opportunities in developing LTMs, including handling mixed-type columns, cross-dataset modeling, textual context, and invariance to column order. It also explores the current state of LTMs, their real-world applications, and the need for robust evaluation methods. The authors conclude by comparing the impact of LTMs and large language models (LLMs) and emphasize the importance of shifting research priorities to focus on LTMs.