TikTok U.S. the social media platform's American entity, is seeking to acquire four gigawatts of computing capacity by the early 2030s, according to sources familiar with the discussions. The expansion underscores the escalating capital demands of large-language-model inference and content-recommendation systems at scale.
The company currently relies on Oracle Cloud Infrastructure for U.S. operations. A TikTok spokesperson confirmed Oracle will remain its primary cloud provider, with infrastructure leased by Oracle. The pursuit of additional gigawatts suggests TikTok U.S. is exploring either direct ownership or long-term capacity commitments outside its existing Oracle arrangements.
Job listings for the TikTok U.S. entity reveal aggressive hiring in machine learning and infrastructure roles. At least one posting details development of a B2B AI product focused on large-model inference and creative automation—a shift that would require substantially more GPU capacity than the company's current consumer recommendation systems.
The TikTok U.S. Joint Venture was formed in 2026 following President Trump's executive order requiring divestiture of TikTok's U.S. operations from Chinese parent ByteDance. Three managing investors—Silver Lake, Oracle and MGX—each hold 15 percent stakes. The Dell Family Office also invested. ByteDance retained 19.9 percent ownership.
In January, TikTok U.S. announced plans to retrain and update its recommendation algorithm using U.S. user data, with the system secured within Oracle's U.S. cloud infrastructure. That effort alone could justify significant additional compute spending.
