TikTok Hunts for 4GW of AI Computing Power for the 2030s

Why TikTok needs 4 gigawatts of capacity

The social network is planning for a decade where AI model training and automated media generation demand city-scale electrical grids.

Four gigawatts is a staggering target. That volume of electricity could power millions of homes, yet it has become the standard scale sought by major tech firms racing to train and run massive machine learning clusters. According to reporting from CNBC, TikTok has begun sounding out infrastructure suppliers about securing long-term capacity for the early 2030s.

The appetite is directly tied to what comes after the short-form video feed. Recent hiring across TikTok's US division reveals an expanding portfolio of AI projects. Beyond basic recommendation ranking, open roles describe work on commercial AI systems for business clients, model inference, and automated creative tools capable of generating media from scratch.

Oracle stays put as rumors swirl

Despite talks with alternative capacity suppliers, TikTok insists its core relationship with Oracle is not changing.

Rumors that TikTok might walk away from Oracle picked up as news of the exploratory talks broke. TikTok moved quickly to shut down talk of an imminent migration. "Oracle will continue to be our primary cloud infrastructure provider," a TikTok spokesperson said.

The reassurance carries weight because the partnership is not purely technical; it is political. Following the 2026 restructuring ordered by the White House, the platform operates domestically as the TikTok USDS Joint Venture. ByteDance retained a 19.9% stake, while Oracle, Silver Lake, and MGX each took 15%, alongside backing from Michael Dell's family investment office. Under that arrangement, TikTok's recommendation algorithm is explicitly slated for domestic training and evaluation on Oracle's American servers.

Yet retaining Oracle as primary host does not preclude TikTok from signing secondary contracts. With Oracle heavily reliant on leased data center space from third-party developers, TikTok may simply be hedging its bets to ensure power bottlenecks do not choke off future AI development—a systemic concern across hyperscalers, as seen when a Google Cloud outage took down European banking and web apps after infrastructure dependencies strained.

What it means for users

For everyday users across North America and Europe, the infrastructure search will not alter how the app functions today. However, securing this volume of computing power signals that future updates will rely far more heavily on native generative tools and algorithmic tailoring processed locally within regional data centers.