Synthetic Data Generation

Overview

Synthetic data generation -- creating artificial but statistically representative datasets to train AI models where real data is scarce, expensive or privacy-restricted -- has grown alongside broader concerns about AI training-data scarcity, with the global market estimated at $710 million in 2026 and forecast to reach $3.67 billion by 2031. NVIDIA acquired category leader Gretel for a reported $320 million in March 2025, folding it into Omniverse and rebranding it 'Synthetic Data Generation for Agentic AI,' while smaller independent player MOSTLY AI reportedly ceased operations in March 2026 after open-sourcing its synthetic data technology, illustrating the category's consolidation around larger AI infrastructure platforms even as underlying demand grows.

Dependency Map

One hop in each direction — see Strategic Connections below for the full, sourced breakdown of every relationship.

Strategic Connections

Strategic Risks

• Rapid consolidation of category leaders into large infrastructure platforms (NVIDIA/Gretel) and apparent exits of independents (MOSTLY AI) suggest a difficult standalone business model despite growing end-market demand
• Synthetic data quality and fidelity concerns remain a technical limitation for the most sensitive AI training use cases

Future Outlook

• Continued growth in end-market demand for synthetic data as real training data becomes scarcer and more expensive
• Further consolidation of independent synthetic data vendors into larger cloud and AI infrastructure platforms

Sources

• CO/AI / Stocktwits (Mar 2025) — NVIDIA acquires synthetic data startup Gretel for ~$320M
• Startups Magazine / industry reporting (Mar 2026) — MOSTLY AI open-sources synthetic data toolkit amid reported shutdown