SpaceX to Lease Compute to Reflection for $150 Million Per Month
Summary
Open source AI startup Reflection has finalized an agreement to pay SpaceX \$150 million per month for access to compute resources housed within SpaceX's Colossus 2 data center. This substantial deal, confirmed by an individual with direct knowledge, is scheduled to begin on July 1 and extend through 2029. A key term of the agreement allows either company to terminate the contract with 90 days' notice, providing a degree of operational flexibility despite the considerable monthly financial commitment. This arrangement underscores a significant development in the AI industry, where major infrastructure providers are increasingly positioning themselves to meet the intense demand for specialized compute capabilities from AI development firms such as Reflection. It also signals SpaceX's expansion into offering high-performance computing services.
Key takeaway
For AI startup founders or CTOs seeking substantial compute resources, this deal indicates a viable, albeit expensive, option from non-traditional providers like SpaceX. You should evaluate the total cost of ownership against your project's compute demands and consider the flexibility offered by cancellation clauses. This trend suggests expanding your search beyond conventional cloud providers to secure necessary infrastructure for large-scale AI model training and inference.
Key insights
Reflection will pay SpaceX $150M monthly for Colossus 2 compute, signaling a new market for infrastructure providers.
Principles
- High-demand AI compute drives large-scale infrastructure deals.
- Flexible cancellation terms are crucial for long-term contracts.
- Non-traditional tech companies enter AI compute market.
In practice
- Consider SpaceX for large-scale AI compute needs.
- Negotiate flexible terms for long-duration contracts.
- Evaluate non-traditional compute providers.
Topics
- AI Compute Leasing
- SpaceX Colossus 2
- Reflection AI
- Data Center Services
- AI Infrastructure
- Compute Market Trends
Best for: CTO, VP of Engineering/Data, Investor, Entrepreneur, Director of AI/ML
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by The Information.