{"id":"a4e18a24-92a5-4ee3-93cb-b004344583f1","name":"Prime Intellect","slug":"prime-intellect","description":"Prime Intellect is a San Francisco based GPU cloud and distributed training platform that positions itself as an \"open superintelligence stack,\" combining an on-demand and reserved GPU marketplace with an in-house training and inference framework. Rather than owning a fleet of datacenters like CoreWeave or Nebius, the company aggregates capacity from more than fifty third-party datacenter operators and resells it through a single API, competing on price, geographic breadth, and open research artifacts rather than hyperscale contracts.\n\nThe company drew attention in late 2024 with INTELLECT-1, a 10 billion parameter model trained across up to 112 Nvidia H100s split between five countries and three continents using its PRIME framework, an implementation of the DiLoCo distributed training approach with a custom int8 all-reduce that cuts communication bandwidth by roughly 400x. That research pitch, that useful models can be trained across heterogeneous, non-colocated clusters, is Prime Intellect's core differentiator against the \"one giant campus\" story told by CoreWeave, Crusoe, and Nebius.\n\nIts customer roster leans toward AI-native application companies and open-model labs, including Ramp, Zapier, Character.ai, Goodfire, Browserbase, and Standard Intelligence. Backers include Founders Fund, Radical Ventures, Nvidia, and Intel, alongside angel investment from Andrej Karpathy and John Schulman. Prime Intellect competes at the long-tail and mid-market end of the neocloud pack rather than for hyperscaler training contracts.","status":"active","founded":2024,"hq":"San Francisco, United States","hqLat":null,"hqLng":null,"hqGeocodedAt":null,"hqAddress":null,"hqPrecision":null,"website":"https://www.primeintellect.ai","fundingTotal":null,"type":null,"roles":[],"reviewStatus":"reviewed","heatIndex":0,"lifecycleState":"active","supersededByCompanyId":null,"sources":[{"url":"https://www.primeintellect.ai/","title":"Prime Intellect: The Open Superintelligence Stack","publisher":"Prime Intellect"},{"url":"https://www.primeintellect.ai/blog/intellect-1-release","title":"INTELLECT-1 Release: The First Globally Trained 10B Parameter Model","publisher":"Prime Intellect"},{"url":"https://www.primeintellect.ai/blog/intellect-1-release","title":"INTELLECT-1 PRIME distributed training framework technical details","publisher":"Prime Intellect"},{"url":"https://www.primeintellect.ai/","title":"Prime Intellect pricing and GPU catalog","publisher":"Prime Intellect"}],"keyFacts":[{"label":"Positioning","value":"GPU marketplace + distributed training stack; aggregates capacity from 50+ third-party datacenters rather than owning campuses","sourceUrl":"https://www.primeintellect.ai/"},{"label":"Headquarters","value":"San Francisco, remote-friendly","sourceUrl":"https://www.primeintellect.ai/"},{"label":"GPU inventory","value":"H100, H200, B200, and B300 instances offered on-demand and via reserved clusters; total fleet size not publicly disclosed","sourceUrl":"https://www.primeintellect.ai/"},{"label":"Announced hourly pricing","value":"H200 (80 GB) spot instances from $0.47/hr; on-demand H100/H200/B200/B300 tiers","sourceUrl":"https://www.primeintellect.ai/"},{"label":"Datacenter footprint","value":"50+ partner datacenters worldwide via colocation and aggregation; no owned campuses disclosed","sourceUrl":"https://www.primeintellect.ai/"},{"label":"Silicon partnerships","value":"Nvidia (investor and primary silicon); Intel (investor)","sourceUrl":"https://www.primeintellect.ai/"},{"label":"Named customers","value":"Ramp, Zapier, Character.ai, Goodfire, Browserbase, Standard Intelligence","sourceUrl":"https://www.primeintellect.ai/"},{"label":"Investors","value":"Founders Fund, Radical Ventures, Nvidia, Intel; angels include Andrej Karpathy and John Schulman","sourceUrl":"https://www.primeintellect.ai/"},{"label":"INTELLECT-1 training run","value":"10B-parameter open model trained on 1T tokens across up to 112 H100s in five countries over 42 days using the PRIME/DiLoCo framework with int8 all-reduce (~400x bandwidth reduction)","sourceUrl":"https://www.primeintellect.ai/blog/intellect-1-release"},{"label":"Distributed training efficiency","value":"96% compute utilization US-only; 85.6% transatlantic; 83% global on the INTELLECT-1 run","sourceUrl":"https://www.primeintellect.ai/blog/intellect-1-release"},{"label":"Public/private status","value":"Private, venture-backed","sourceUrl":"https://www.primeintellect.ai/"}],"aliases":[],"collisionRisk":"low","reviewNote":null,"atsProvider":null,"atsSlug":null,"factoryLocations":null,"manufacturingCapacity":null,"orgType":null,"burnSignal":null,"litigationEvents":null,"insurancePostureDisclosed":null,"unitEconomicsDisclosed":null,"tickerSymbol":null,"stockExchange":null,"secCik":null,"revenueUsd":null,"revenueBasis":null,"revenueAsOf":null,"employeeCount":null,"employeeCountBasis":"Not disclosed publicly on the company site as of this pass.","employeeCountAsOf":null,"marketCapUsd":null,"marketCapAsOf":null,"marketCapSource":null,"createdAt":"2026-08-28T01:48:33.221Z","updatedAt":"2026-08-28T03:45:17.421Z","jsonLd":{"@context":"https://schema.org","@type":"Organization","@id":"https://registry.deploy.report/companies/prime-intellect","url":"https://registry.deploy.report/companies/prime-intellect","name":"Prime Intellect","description":"Prime Intellect is a San Francisco based GPU cloud and distributed training platform that positions itself as an \"open superintelligence stack,\" combining an on-demand and reserved GPU marketplace with an in-house training and inference framework. Rather than owning a fleet of datacenters like CoreWeave or Nebius, the company aggregates capacity from more than fifty third-party datacenter operators and resells it through a single API, competing on price, geographic breadth, and open research artifacts rather than hyperscale contracts.\n\nThe company drew attention in late 2024 with INTELLECT-1, a 10 billion parameter model trained across up to 112 Nvidia H100s split between five countries and three continents using its PRIME framework, an implementation of the DiLoCo distributed training approach with a custom int8 all-reduce that cuts communication bandwidth by roughly 400x. That research pitch, that useful models can be trained across heterogeneous, non-colocated clusters, is Prime Intellect's core differentiator against the \"one giant campus\" story told by CoreWeave, Crusoe, and Nebius.\n\nIts customer roster leans toward AI-native application companies and open-model labs, including Ramp, Zapier, Character.ai, Goodfire, Browserbase, and Standard Intelligence. Backers include Founders Fund, Radical Ventures, Nvidia, and Intel, alongside angel investment from Andrej Karpathy and John Schulman. Prime Intellect competes at the long-tail and mid-market end of the neocloud pack rather than for hyperscaler training contracts.","identifier":"a4e18a24-92a5-4ee3-93cb-b004344583f1","foundingDate":"2024","address":"San Francisco, United States","publisher":{"@id":"https://deploy.report/#organization"}},"framework_metadata":{"framework_schema_version":"0.1.0","verification_status":"verified","maturity_stage":null,"lifecycle_state":null,"architectural_position":{"cohort":null,"sub_cohorts":[]},"within_cohort_verified_vs_claimed_pair":null,"cap_flags":[],"verification_depth":{"sources_count":4,"primary_source_types":[]}}}