Crusoe
A U.S.-based AI data center developer that entered the market with an energy-first philosophy. The company initially used associated natural gas from oil fields to power cryptocurrency mining operations and subsequently transitioned into an infrastructure provider serving AI training and inference workloads.
Timeline (last 90 days)
For openai/gpt-oss-120b, across 13 providers the input price ranges from $0.03/M (wandb) to $0.80/M (Crusoe), a spread of about 27x (+2566.7%).
GPU-hour hosts (such as Crusoe and Hyperbolic) bill by raw compute hour rather than by token, and their per-token price depends on how fast the model runs on the hardware, causing the per-token cost of the same model to fluctuate sharply.
Qwen/Qwen3-235B-A22B-Instruct-2507 ranges from a low of $0.09/M (DeepInfra) to a high of $3/M (Crusoe) across 3 providers, a spread of +3233.3%, the largest price spread case in the article.
Dedicated GPU cloud providers such as Lambda, Crusoe, Nebius, and Fluidstack offer lower unit prices than hyperscale providers, but typically require a certain level of resource commitment.
Boom Supersonic's newly launched stationary power generation units are no longer part of Crusoe's near-term plans.
The original source material does not disclose the specific reason why Crusoe abandoned the partnership.
Crusoe abandoned a partnership plan valued at $1.25 billion that had originally intended to use Boom Supersonic's turbines to power its AI data centers.
Typical players in the offtake network space include GPU cloud platforms such as CoreWeave and Crusoe, as well as P2P compute markets such as Vast.ai and RunPod.
Crusoe originated from a business model of converting associated gas flared at oil and gas fields into electricity to power GPU computing (flare gas power generation + mobile data centers).
Crusoe completed a financing round totaling $3.9 billion.
8 more timeline events
All Facts (12)
Qwen/Qwen3-235B-A22B-Instruct-2507 ranges from a low of $0.09/M (DeepInfra) to a high of $3/M (Crusoe) across 3 providers, a spread of +3233.3%, the largest price spread case in the article.
50%UnverifiedFor openai/gpt-oss-120b, across 13 providers the input price ranges from $0.03/M (wandb) to $0.80/M (Crusoe), a spread of about 27x (+2566.7%).
50%UnverifiedGPU-hour hosts (such as Crusoe and Hyperbolic) bill by raw compute hour rather than by token, and their per-token price depends on how fast the model runs on the hardware, causing the per-token cost of the same model to fluctuate sharply.
50%UnverifiedDedicated GPU cloud providers such as Lambda, Crusoe, Nebius, and Fluidstack offer lower unit prices than hyperscale providers, but typically require a certain level of resource commitment.
50%UnverifiedTypical players in the offtake network space include GPU cloud platforms such as CoreWeave and Crusoe, as well as P2P compute markets such as Vast.ai and RunPod.
50%UnverifiedCrusoe originated from a business model of converting associated gas flared at oil and gas fields into electricity to power GPU computing (flare gas power generation + mobile data centers).
50%UnverifiedCrusoe's funding will be used to build large-scale data centers as well as small modular AI factories.
50%UnverifiedCrusoe has reached a contract worth $13 billion with quantitative trading firm Jane Street
50%UnverifiedThe $30 billion valuation is built on expectations of sustained high-speed growth in AI demand, and if computing supply becomes oversaturated, the asset-heavy model will face depreciation pressure.
50%UnverifiedJane Street's long-term contract provides Crusoe with stable revenue expectations and a credit endorsement, boosting investor confidence.
50%UnverifiedCrusoe has redirected its experience in energy integration and data center construction toward the AI infrastructure sector
50%UnverifiedCrusoe initially entered the market with an "energy-first" philosophy, providing computing power for cryptocurrency mining by utilizing flare gas from oil fields to generate electricity
50%