Skip to main content

DataPelago

Category: AI Infrastructure

DataPelago develops a Universal Data Processing Engine (Nucleus) that accelerates data processing for GenAI and analytics workloads using heterogeneous accelerated computing (CPU, GPU, FPGA), enabling 10x faster processing and up to 80% cost reduction. DataPelago was founded in 2021. The company is led by Rajan Goyal. Based in Mountain View, California. Team size: 101-500. Total funding raised: $75M. Latest round: Series A. Key investors include Eclipse Ventures, Taiwania Capital Management, Qualcomm Ventures, Alter Venture Partners, Nautilus Venture Partners, Silicon Valley Bank (First Citizens Bank).

Founded
2021
Headquarters
Mountain View, California
Team size
101-500
Total funding
$75M

Value proposition

10x faster data processing at 80% lower cost for GenAI and analytics workloads, with zero code changes, zero data migration, and zero vendor lock-in — by processing data where it resides using heterogeneous accelerated computing

Products and solutions

DataPelago Nucleus (Universal Data Processing Engine) — comprising DataVM (domain-specific ISA virtual machine), DataOS (heterogeneous compute orchestration layer), and DataApp (pluggable integration layer for Spark, Trino, etc.), DataPelago Accelerator for Spark

Unique value

World's first Universal Data Processing Engine that accelerates any query engine (Spark, Trino) on any hardware (CPU, GPU, FPGA) for any data type (structured, semi-structured, unstructured) without requiring application changes or data migration

Target customer

Enterprise data and AI teams running large-scale analytics and GenAI workloads on Apache Spark, Trino, and lakehouse architectures; organizations with significant GPU infrastructure investments seeking better utilization

Industries served

Financial services/insurance (Akad Seguros), Cybersecurity (McAfee, HiddenLayer), Social media (ShareChat), Marketing analytics (RevSure.ai), Technology/Samsung SDS, General enterprise AI/analytics

Technology advantage

Heterogeneous accelerated computing engine that dynamically maps data operations to CPUs, GPUs, FPGAs, and custom silicon at runtime; zero-copy data processing eliminating data movement; open-source compatible (Apache Gluten, Substrait, Velox); no vendor lock-in; works with existing Spark/Trino/Presto deployments without code changes; claims up to 10x faster processing and 80% cost reduction vs traditional architectures; outperforms NVIDIA cuDF in benchmarks (up to 10.5x for project, 10.1x for filter, 4.3x for aggregate operations)

How they differentiate

Hardware-agnostic (CPU/GPU/FPGA) vs NVIDIA cuDF's GPU-only lock-in; open-source compatible (Gluten/Substrait) vs Databricks Photon's proprietary approach; zero-code deployment vs requiring application rewrites; dynamic runtime hardware mapping vs static configurations; claims superior benchmarks vs both NVIDIA cuDF and Databricks Photon

Main competitors

NVIDIA cuDF (GPU-accelerated data processing), Databricks Photon (proprietary Spark engine), Starburst, Dremio, Teradata

Key partnerships

Design partner: McAfee, Cloud partnerships: AWS, Azure, Google Cloud, Open-source ecosystem: Apache Gluten, Substrait, Velox, Apache Spark, Trino, Presto, Reseller: Twingo (Israel)

Notable customers

Akad Seguros, McAfee, ShareChat, RevSure.ai, Samsung SDS America, HiddenLayer

Major milestones

Founded 2021, Came out of stealth October 2024 with $47M funding and Universal Data Processing Engine launch, Named to Fast Company's Most Innovative Companies of 2026 (#4 in Data Science), Acquired by NetApp in July 2026

Market positioning

Positioned as a disruptive data acceleration layer between storage and query engines, competing against NVIDIA's GPU-native libraries and proprietary lakehouse engines by offering hardware-agnostic, open-source-compatible acceleration with superior price/performance

Geographic focus

Global (US-headquartered with India office; customers in US, Brazil, India, Israel)

Patents and IP

Founder Rajan Goyal holds 200+ patents; Satya Lakshmipathi Billa holds 100+ patents; company has filed patents related to universal data processing engine technology

About Rajan Goyal

Ex-CTO at Fungible (DPU startup acquired by Microsoft); Distinguished Engineer at Cavium (multi-core processors, acquired by Marvell); roles at Cisco and Oracle. Holds 200+ patents. MS in Computer Science from Stanford University; Bachelor's in Mechanical Engineering from Thapar Institute.

Latest news about DataPelago

More AI Infrastructure companies

Official website: