Voltron Data

CEO Josh Patterson was previously the engineering lead for RAPIDS, Nvidia's open-source suite of GPU-accelerated data-science libraries. CTO Rodrigo Aramburu also came from Nvidia, where he worked on GPU database kernel development. The board includes Todd Mostak, who founded GPU database company MapD (later OmniSci), giving the firm deep bench strength in the hardware-software interface for accelerated analytics.

Theseus is a distributed query engine that compiles SQL and dataframe operations directly to GPU kernels rather than executing them on CPUs. It uses the Apache Arrow columnar format as its in-memory representation, which means it can interoperate with the existing PyData ecosystem — Pandas, RAPIDS, DuckDB — without requiring users to extract, transform, or serialize data into a proprietary format. The engine is designed to replace legacy CPU-based query layers in machine-learning feature pipelines.

No. Voltron Data is a venture-backed operating company, not an investment vehicle. Its $110 million in disclosed funding across seed and Series A rounds — from Walden Catalyst, BlackRock, and others — finances product development, not portfolio deployment. The firm generates revenue through enterprise subscriptions and support for its open-source engine.

Voltron Data was co-founded by Josh Patterson and key contributors to the Apache Arrow ecosystem. Arrow's creator, Wes McKinney, is not an executive at the firm but the company is built entirely on the Arrow standard: Theseus uses Arrow as its wire format and in-memory representation, making it natively compatible with any tool in the Arrow ecosystem without serialization overhead.

The firm has publicly cited Walmart and HPE as early adopters. The HPE relationship is a formal hardware partnership, with Theseus integrated into HPE's Apollo and Cray high-performance computing systems to provide GPU-accelerated SQL analytics for enterprise customers running compute-intensive data engineering workloads.