Crusoe
Crusoe builds vertically integrated AI infrastructure—energy systems, AI data centers, and Crusoe Cloud—to help AI builders train and serve models faster. Its energy-first approach converts stranded, wasted, renewable, and other abundant energy into reliable compute while reducing environmental impact.
At a Glance
- AI model labs and generative-AI startups
- Enterprise AI developers
- Hyperscalers and large technology companies
- HPC and research organizations
- +3 more
AI Tools by Crusoe
(1)
Crusoe
GPU Cloud for AI Model Training
Discussions
No discussions yet
Be the first to start a discussion about Crusoe
Latest News
Crusoe pledged $50 million in compute credits to the White House AI Science Initiative.
Crusoe and Thinking Machines Lab announced a $65 million annual agreement for open-model inference on Crusoe Cloud.
Crusoe announced the initial close of a $3.9 billion Series F at a $30.9 billion post-money valuation.
Crusoe and Perplexity announced a multi-year partnership spanning frontier-model training and managed inference.
Products & Services
AI-focused infrastructure cloud offering GPU and CPU compute, storage, networking, orchestration, Kubernetes/Slurm clusters, and APIs for large-scale training, inference, and data processing.
Managed AI platform for discovering and deploying models, bringing proprietary models, serverless inference, self-serve and tailored deployments, and fine-tuning.
Managed, autoscaling inference for open and proprietary models with API access and production deployment without customer-managed infrastructure.
LoRA-based supervised fine-tuning workflow in Intelligence Foundry; users upload data and fine-tune without provisioning a cluster.
Market Position
Crusoe positions itself as a vertically integrated, energy-first AI infrastructure provider spanning power, data-center design/construction, hardware deployment, and cloud/managed AI services. It competes with neocloud and GPU-cloud providers such as CoreWeave, Lambda, FluidStack, and Together AI, as well as hyperscaler infrastructure from AWS, Microsoft Azure, Google Cloud, and Oracle; its differentiation is control of energy and physical infrastructure plus an AI-optimized cloud stack.
Leadership
Founders
Chase Lochmiller
Previously a quantitative researcher and trader at GETCO and Jump Trading, and a general partner at Polychain Capital. He studied mathematics and physics at MIT and computer science/AI at Stanford.
Cully Cavness
Energy entrepreneur who previously worked as an energy investment banker at Petrie Partners and as Business Development Manager at Global Geothermal, a renewable power-plant developer. He studied geology at Middlebury and earned an MBA from Oxford.
Executive Team
Chase Lochmiller
Co-founder, CEO and Chairman of the Board
Former quantitative researcher/trader at GETCO and Jump Trading; former Polychain Capital general partner.
Cully Cavness
Co-founder, President and Chief Strategy Officer
Energy entrepreneur; previously at Petrie Partners and Global Geothermal.
Board of Directors
Founding Story
Lochmiller and Cavness founded Crusoe in 2018 to turn otherwise-wasted energy—initially flare gas from oil and gas operations—into useful electricity for modular computing. The original digital flare-mitigation model powered Bitcoin mining at well sites; the company subsequently expanded that energy-first infrastructure into AI data centers and cloud computing.
Business Model
Revenue Model
Usage-based cloud revenue from GPU/CPU compute, storage, managed Kubernetes, inference, and fine-tuning, supplemented by enterprise/reserved capacity agreements and infrastructure/data-center development and operating contracts.
Pricing Tiers
Public on-demand examples include L40S $1.50/GPU-hour, A100 PCIe $2.00, A100 SXM $2.30, H100 $3.90, and H200 $4.29; GB200/B200 and some others require sales contact.
Additional CPU/RAM configurations are listed on the pricing page.
Spot and reserved pricing, growth-aligned commitments, and tailored enterprise agreements are available.
Serverless Inference, Self-Serve Deployments, Tailored Deployments, Provisioned Throughput, and Serverless Fine-Tuning.
Target Markets
- AI model labs and generative-AI startups
- Enterprise AI developers
- Hyperscalers and large technology companies
- HPC and research organizations
- Cloud-native software companies
- Oil and gas, renewable-energy, battery-storage, and nuclear-energy partners
- Large-scale AI model training
- Fine-tuning and experimentation
- Low-latency production inference and AI agents
- HPC and distributed GPU workloads
- Data processing and model checkpointing
- Managed model serving for frontier and open-model labs
- OpenAI/Stargate
- Oracle Cloud Infrastructure
- Perplexity
- Thinking Machines Lab