NVIDIA's Gretel Acquisition Boosts $9.2B Synthetic Data Market as Speed Claims Hit 100x

Context Synthetic tabular data are AI‑generated datasets that mimic the statistical properties of real tables without containing any actual personal information. Organizations use them to train machine‑learning models while staying compliant with privacy regulations such as GDPR, CCPA, and emerging AI‑specific rules. The push for privacy‑first analytics, combined with the need for large, high‑quality training sets, has fueled rapid expansion of the synthetic‑data sector. Regulatory bodies worldwide are also issuing guidance that encourages the use of synthetic data for compliance testing.

Key Facts Market researchers project the AI‑generated synthetic tabular dataset market to be worth $9.24 billion by 2030. In March 2025, NVIDIA acquired Gretel Labs to integrate its privacy‑focused synthetic‑data technology into its developer and cloud ecosystems. The financial terms of the deal were not disclosed. Separately, MOSTLY AI released an open‑source SDK that claims training speeds up to 100 times faster than conventional approaches.

What It Means By embedding Gretel's tools, NVIDIA can let customers generate realistic synthetic data directly inside its AI Enterprise platform and cloud services. This includes support for tabular, time‑series and text datasets, broadening the range of AI workloads that can benefit. This tight integration could shorten the data‑preparation stage of model development, letting teams iterate more quickly on AI projects. The speed claim from MOSTLY AI highlights a growing emphasis on performance as a differentiator, and enterprises will likely weigh both privacy guarantees and training efficiency when choosing a provider. Watch for NVIDIA's roadmap to expose Gretel's APIs through its AI Enterprise suite and for independent benchmarks that test the 100 times speed claim in real‑world AI training workloads.