Tabular data, the structured rows and columns inside spreadsheets and databases, sits at the core of nearly every enterprise, powering decisions across healthcare, finance, and business. Yet while deep learning transformed how we work with text and images, tabular data remained the domain of gradient-boosted trees for two decades. This talk introduces Tabular Foundation Models, a new class of AI built specifically for structured data. Prior Labs' TabPFN, published in Nature and validated across hundreds of independent studies, delivers instant and reliable predictions without task-specific training, outperforming previous methods by a wide margin, and has been improving at the exponential pace typical of foundation models. It is already used in thousands of real-world applications, from early disease detection to financial risk modeling, with the latest TabPFN-3 again topping the benchmarks and agentic integrations allowing anyone to use state-of-the-art tabular predictions. I will share where the technology stands today and the road ahead: in May 2026, eighteen months after Prior Labs was founded, the company entered into a definitive agreement to be acquired by SAP, which is investing more than €1 billion over four years to scale Prior Labs into an independent, globally leading frontier AI lab for structured data in Europe.