TikTok U.S. targets four gigawatts of compute power for AI expansion
TikTok U.S. operations are aggressively seeking multiple gigawatts of computing capacity, aiming for four gigawatts by the early 2030s. S. operations are aggressively seeking “multiple gigawatts” of computing capacity, a move underscored by the global rush to secure digital infrastructure for advancing artificial intelligence workloads.
The company, formally known as TikTok USDS Joint Venture, is reportedly in discussions to acquire an unprecedented four gigawatts of TikTok U.S. compute capacity by the early 2030s, signaling a substantial investment in its U.S.-based AI capabilities.
This massive procurement drive, first reported on 2024-06-18, highlights the intense competition for energy resources and data center space across the United States. It also reflects TikTok U.S.’s commitment to localizing its algorithm training and other AI-driven services within the country, a requirement stemming from ongoing U.S. national security concerns.
TikTok U.S. compute capacity for AI workloads
The quest for four gigawatts by the early 2030s represents a staggering demand for computing capacity from a single entity. To put this in perspective, one gigawatt of power is sufficient to supply hundreds of thousands of homes, illustrating the sheer scale of TikTok U.S.’s ambitions for its artificial intelligence infrastructure. This push is fueled by the escalating computational needs of advanced machine learning models.
Industry experts have described the demand as both “unprecedented” and “staggering.” One source familiar with the discussions noted, “It’s an unprecedented amount of power for a single company to be seeking.” Another added, “They are looking for capacity that simply doesn’t exist yet in many markets,” underscoring the challenge TikTok U.S. faces in securing such vast resources.
The evolving role of Oracle Cloud Infrastructure
TikTok U.S. has affirmed that Oracle Cloud Infrastructure (OCI) remains its primary cloud provider in the United States. This partnership is central to its operational strategy, particularly concerning data security and algorithm localization. OCI typically operates within facilities that Oracle leases, forming a critical backbone for TikTok’s U.S. data handling.
The company confirmed in January that it would “retrain, test, and update” its core recommendation algorithm, which is heavily reliant on machine learning, exclusively using U.S. user data. This vital algorithm will be “secured in Oracle’s U.S. cloud environment,” a crucial step in meeting regulatory mandates and ensuring data sovereignty.
The expanding compute needs suggest a deeper integration and reliance on OCI’s capabilities for these specialized tasks.
Intense competition for digital infrastructure
TikTok U.S.’s hunt for gigawatts of power unfolds amid a global scramble for digital infrastructure. Major tech players like Microsoft, Amazon, and Google are also significantly expanding their data center footprints to support their own burgeoning AI initiatives. This widespread demand has made electricity the primary limiting factor for data center expansion, often overshadowing concerns about capital, land, or connectivity.
The scale of power TikTok U.S. is seeking is equivalent to several large data centers, reflecting a broader industry trend where AI facilities consume substantially more power than traditional ones. A single AI server rack can demand 50-150 kilowatts, a stark contrast to the 10-15 kilowatts required by conventional computing racks. This increased consumption creates significant pressure on existing energy grids and data center markets.
TikTok USDS Joint Venture: Structure and mandate
The TikTok USDS Joint Venture LLC was officially established in January 2026, marking a pivotal moment in the platform’s U.S. operations. This entity was created to directly address U.S. national security concerns surrounding potential Chinese government access to user data.
Its formation came after an Executive Order signed by then-President Donald Trump on September 25, 2025. This order extended the deadline for enforcing the Protecting Americans from Foreign Adversary Controlled Applications Act (PAFACA) until January 22, 2026, allowing the divestiture of TikTok’s U.S. operations from its Chinese parent company, ByteDance, to be finalized.
The joint venture is majority-American owned, with Silver Lake, Oracle, and MGX Fund Management Limited each holding a 15% stake as managing investors. Other notable investors include the Dell Family Office and various strategic investment firms. ByteDance retains a 19.9% minority stake.
Adam Presser serves as CEO, with Will Farrell appointed Chief Security Officer, overseeing the venture’s strict mandate to safeguard U.S. user data and its recommendation algorithm within Oracle’s U.S. cloud environment.
Regulatory pressures shaped TikTok’s U.S. entity
The current structure of TikTok U.S. is a direct outcome of years of intense U.S. government scrutiny. Concerns escalated with the Committee on Foreign Investment in the United States (CFIUS) review of ByteDance’s 2017 acquisition of Musical.ly. Subsequent executive orders, including threats to ban the app, eventually culminated in the Protecting Americans from Foreign Adversary Controlled Applications Act (PAFACA) in April 2024, which mandated divestiture.
While previous efforts like “Project Texas” aimed to create a data firewall with Oracle’s involvement, they were ultimately deemed insufficient by the Biden administration. The establishment of the TikTok USDS Joint Venture represents a more comprehensive approach to data security and operational independence for the platform in the U.S. Such regulatory landscapes frequently influence curbs on U.S. investment in foreign technology sectors.
Data localization and algorithm training in the U.S.
A core part of the TikTok USDS Joint Venture’s mandate involves retraining, testing, and updating its prized recommendation algorithm solely with U.S. user data. This critical component, which shapes the user experience, will be fully secured within Oracle’s domestic cloud environment.
This localization effort requires substantial computational resources, directly contributing to the demand for multiple gigawatts of power. It signifies a complete overhaul of how TikTok’s core intellectual property operates within the United States.
Broader implications of AI power consumption
The enormous computing capacity sought by TikTok U.S. underscores a broader, rapidly accelerating trend: the insatiable energy demands of artificial intelligence. AI data centers consume vastly more power than their traditional counterparts, often three to five times more per square foot. Modern GPUs, essential for AI workloads, can draw 700-1,200 watts per chip, significantly more than the 150-200 watts typically used by traditional CPUs.
Globally, projections indicate that AI data centers could require 68 gigawatts of power by 2027, potentially skyrocketing to 327 gigawatts by 2030. This compares to a total global data center capacity of 88 gigawatts in 2022. The dramatic increase in power consumption poses significant challenges for energy infrastructure and raises questions about the sustainability of rapid AI expansion.
Securing this level of power involves complex negotiations with major data center providers and utilities, often requiring new infrastructure development. The competitive landscape means companies like TikTok U.S. are not just bidding for existing capacity but are also driving the creation of new energy and data center solutions.
This push for new infrastructure reflects broader economic trends impacting tech companies, as seen in recent earnings estimates and market performance across the sector.
The pursuit of such immense power also points to the advanced AI products TikTok U.S. is developing. The company has listed numerous machine learning roles and advertised for positions focused on “shipping a flagship B2B AI product at the frontier of large-model inference and creative automation.”
These ambitious product plans necessitate the foundational compute power now being sought, signaling a strategic pivot towards more intensive AI integration beyond its consumer-facing app.
Future outlook for AI infrastructure and energy demands
TikTok U.S.’s aggressive push for four gigawatts of compute power offers a stark glimpse into the future of artificial intelligence development. As AI models grow more complex and widespread, the demand for underlying infrastructure—especially electricity—will only intensify. This scenario compels data center operators, energy providers, and tech companies to innovate rapidly in power generation, distribution, and energy efficiency.
The competition for these resources is set to reshape global energy markets and infrastructure investment over the next decade. Companies that can secure reliable, large-scale power will gain a significant competitive edge in the AI race.
For TikTok U.S., this investment is not just about keeping pace but about establishing a robust, independent operational foundation that aligns with both its product ambitions and stringent regulatory requirements.