Market Analysis: Meta's Cloud Computing Strategy Focuses on Cash Flow Recovery Rather Than Halting High-End Computing Pursuits

On July 2, market analysis indicated that Meta is planning to launch a cloud computing business that rents out computing power: one type involves opening up model capabilities deployed on its own AI infrastructure to external clients, while the other involves renting out more basic 'bare metal' computing power. However, the reality seems to lean more towards Meta recovering cash flow from its existing outdated computing resources rather than stopping its pursuit of high-end computing power. In mid to late June, it was reported that Meta had signed a contract with Crusoe to obtain a total of approximately 1.6GW of AI computing capacity from two data centers in Texas and Missouri. At the same time, Meta raised its full-year capital expenditure guidance for Q1 2026 to $125 billion to $145 billion. When viewed together, these two developments appear to represent a reallocation of resources across different generations and uses: continuing to purchase new cards to train cutting-edge models, while accepting the rental of some older cards (such as the H series) for applications like inference of high-traffic products and hosting external models, without indicating a slowdown in the purchase of the most sought-after high-end cards.
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned