Voltage Park
Voltage Park provides owned, high-performance GPU infrastructure and software for AI and machine-learning workloads. Its mission is to accelerate access to AI transformation by making scalable, flexible and cost-effective compute available to startups, enterprises, researchers and nonprofits.
At a Glance
- AI startups and seed-stage companies
- Large AI labs and technology companies
- Enterprise AI teams
- Research universities and scientists
- +3 more
AI Tools by Voltage Park
(1)
Voltage Park
GPU Cloud for AI Workloads
Discussions
No discussions yet
Be the first to start a discussion about Voltage Park
Latest News
Lightning AI and Voltage Park completed their merger to create a unified AI cloud built for training, inference and AI applications.
JBK Group partnered with Voltage Park and Matrice.ai for vision-AI factory deployments across Qatar.
Voltage Park launched its AI Factory positioning as a faster path to enterprise AI transformation.
Voltage Park led strategic funding for Matrice.ai to industrialize real-time vision AI.
Products & Services
Self-serve bare-metal H100 GPU clusters, with hourly usage, no long-term contract, and startup times advertised at about 15 minutes; pricing starts at $1.99 per GPU-hour.
Reserved H100 capacity for longer-term workloads, with six- or 12-month-plus terms, large single-cluster configurations and 3,200-Gbps InfiniBand.
Fully managed Kubernetes control plane for AI/ML containers on bare-metal GPU clusters, including provisioning, security, patching and monitoring. Defaults include NVIDIA GPU Operator, Prometheus/Grafana and SentinelOne.
An AI-agent project for achieving state-of-the-art benchmark scores on web tasks in an operating-system setting, powered by Voltage Park compute.
Market Position
Voltage Park positioned itself as an AI-native neocloud and AI factory rather than a general-purpose hyperscaler: it owns the GPU hardware, offers transparent on-demand and reserved access, uses high-speed InfiniBand and adds managed Kubernetes, storage, security and support. Its competitors include CoreWeave, Lambda, Crusoe, Vast.ai and other GPU neoclouds; after the Lightning AI merger it differentiates by combining GPU capacity with integrated AI development, inference and deployment software.
Leadership
Founders
Eric Park
CEO at Voltage Park when it launched in 2023; previously associated with the San Francisco Compute Company, where he served as CTO, and worked on deploying large-scale machine-learning infrastructure.
Jed McCaleb
Founder of the Navigation Fund, the nonprofit/endowment that provided Voltage Park's initial capital foundation; McCaleb is known as a co-founder of Stellar and an early founder of Ripple and Mt. Gox.
Executive Team
Ozan Kaya
President, Lightning AI; former CEO of Voltage Park
Former President of CarLotz before its merger with Shift Technologies; previously CFO/COO at ShopRunner, CFO at Kapow, Treasurer at Milestone Aviation Group, and an investment banker at Deutsche Bank and Wachovia Securities.
Saurabh Giri
Chief Product & Technology Officer
Previously led teams that built and operated Amazon Bedrock and other AWS AI products; earlier co-founded a payments platform and an algorithmic trading firm and worked in Deloitte corporate restructuring.
Board of Directors
Founding Story
Voltage Park was started in response to the shortage and high cost of cutting-edge ML compute. Its initial vision was to make machine-learning infrastructure accessible to everyone—from seed-stage startups and nonprofits to research universities and large enterprises—by owning and operating a large H100 fleet and offering more flexible access than traditional hyperscalers.
Business Model
Revenue Model
Usage-based hourly GPU rentals for on-demand compute, plus longer-term reserved/dedicated-capacity contracts and enterprise managed infrastructure/services.
Pricing Tiers
Self-serve, pay only for usage, no minimum term, advertised startup in about 15 minutes.
Self-serve configurations from 8 to 1,016 HGX H100 GPUs, with no minimum terms.
Six-month-plus contracts, 32 to 8,000-plus H100 GPUs in a configuration, and 3,200-Gbps InfiniBand.
Target Markets
- AI startups and seed-stage companies
- Large AI labs and technology companies
- Enterprise AI teams
- Research universities and scientists
- Nonprofits and government-funded research
- GPU brokers and AI-as-a-service providers
- Large-scale model training
- Fine-tuning and experimentation
- Production inference and model serving
- AI-agent and web-task workloads
- High-performance computing and scientific research
- Containerized AI workloads on Kubernetes
- Imbue
- Character.ai
- Atomic AI
- Phind