• TikTok USDS Joint Venture has approached the market for several gigawatts of computing capacity, with one source citing a target of about 4 GW by the early 2030s.
  • Oracle (ORCL) will remain the primary U.S. cloud infrastructure provider, but TikTok did not confirm the size of the target or any signed contracts.
  • The reported capacity search underscores the venture's need for domestic machine-learning infrastructure and potential new AI products, though financial details remain undisclosed.

TikTok’s U.S. joint venture is hunting for a massive slug of computing power, according to a CNBC (VSNT) report on Friday, as the entity established to oversee American data security seeks to scale up its domestic operations. The venture has approached the market about securing several gigawatts of capacity, with one source telling CNBC it wants about 4 gigawatts by the early 2030s. The important distinction is that this is a reported capacity search—not an announced construction program or a signed contract.

TikTok confirmed that Oracle will remain its primary U.S. cloud infrastructure provider, but did not publicly confirm the size of the target. The report, citing two people familiar with the discussions, does not identify selected sites, additional suppliers, contract values, or a deployment schedule beyond the early-2030s horizon. The specific 4-GW figure comes from one source. TikTok did not respond to a request for comment.

Gigawatts describe the electrical scale associated with computing infrastructure, not a direct measure of processing performance. Four gigawatts equals 4,000 megawatts; the workload supported would depend on the chips, facility design and utilization. The headline should therefore not be read as a disclosed GPU purchase or a precisely defined amount of AI performance.

Two developments help explain the potential demand. First, TikTok’s U.S. venture has committed to retraining, testing and updating its recommendation algorithm using U.S. user data within Oracle’s U.S. cloud environment. That creates an ongoing requirement for domestic machine-learning infrastructure. Second, CNBC identified machine-learning hiring and a job advertisement describing a “flagship B2B AI product” focused on large-model inference and creative automation. This suggests ambitions beyond the existing video feed, although it does not establish how much of the reported capacity would serve that product.

The corporate division matters. TikTok is a short-video, discovery and commerce platform within ByteDance’s broader technology business. This headline specifically concerns TikTok USDS Joint Venture, the U.S. security-and-operations entity established in January 2026—not ByteDance’s entire global infrastructure program. TikTok said the U.S. arrangement supports more than 200 million Americans and 7.5 million businesses; these are company-reported figures.

The joint venture oversees U.S. data security, algorithm security, software assurance and content moderation. TikTok global’s U.S. entities retain responsibility for global interoperability and certain commercial activities, including advertising, marketing and e-commerce. Consequently, the joint venture’s finances should not be equated with all U.S. TikTok sales.

Standalone financial results for the U.S. joint venture were not disclosed in the announcement or CNBC’s capacity report. Parent-company figures provide context, but do not establish the venture’s profitability or its ability to fund this particular expansion. ByteDance’s first-half 2026 revenue was approximately $120 billion, up roughly 30% year over year, according to September reporting citing The Information. Net profit was approximately $20 billion, down by a single-digit percentage, with AI infrastructure spending weighing on earnings. These are reported private-company figures, not publicly released audited results. Separately, Reuters (TRI) reported on September 4 that ByteDance secured a $29.6 billion loan from nearly 30 banks, with sources saying proceeds would mainly support AI plans outside China. That financing is separate from any funding for TikTok USDS’s reported capacity search.

The significant recent leadership change accompanied the January restructuring: Adam Presser became CEO of TikTok USDS Joint Venture; Will Farrell became chief security officer; Shou Chew remained TikTok’s global CEO and joined the venture’s seven-member, majority-American board. Oracle, Silver Lake and Abu Dhabi-based MGX each hold 15%; ByteDance retains 19.9%. Oracle’s dual position—as a managing investor and the primary cloud provider—is particularly relevant to how the expansion could be delivered. CNBC notes that Oracle Cloud Infrastructure typically operates in facilities Oracle leases, meaning expansion could occur through cloud contracts and leased facilities rather than TikTok directly owning new campuses.

The search places TikTok alongside AI companies competing for large blocks of data-center capacity. A close parallel emerged on October 20, when CNBC reported that Anthropic was discussing a second 460-MW data center with Nscale in West Virginia, potentially worth around $45 billion. Those negotiations, like TikTok’s search, illustrate competition for future infrastructure rather than proof that every proposed facility will be delivered.

The immediate political background is the U.S. campaign to separate TikTok’s American operations from Chinese control. The January 2026 venture was established under President Trump’s September 25, 2025 executive order, following the prolonged divest-or-ban dispute. Its safeguards include U.S.-based Oracle cloud storage, third-party cybersecurity assessments, source-code review and domestic control of algorithm security and moderation. That makes the computing search more than an ordinary IT expansion: U.S. infrastructure must support a governance arrangement designed to address national-security concerns. However, CNBC does not establish that the entire reported 4-GW target is required for regulatory compliance; new AI products and ordinary platform growth could also contribute.

Internationally, the structure preserves connections rather than creating complete separation: ByteDance retains a minority stake, an Emirati investor is among the three managing investors, and the platform maintains global interoperability. The key unresolved policy question is how effectively the safeguards separate sensitive U.S. operations while retaining a global service—not simply where servers are located.

For prospective host communities, the debate is more tangible: electricity demand, infrastructure costs and environmental effects. CNBC’s October 19 report cited a March Gallup survey in which 71% of Americans opposed a data center in their local area, and a summer Pew survey showing concern about environmental effects and household energy costs. Those findings concern data centers generally; they are not evidence of a backlash to today’s TikTok headline. The CNBC capacity report itself does not document a specific public reaction or identify a community facing a TikTok-linked project.

The most informative next developments would be signed capacity agreements, named locations, power-supply arrangements and clarification of whether 4 GW represents total future capacity or incremental additions. CNBC’s report does not resolve those questions, so assigning a project cost or completion probability would be premature. Oracle is positioned to remain central, based on TikTok’s explicit statement. But that does not mean Oracle has already won a new 4-GW contract, nor that all underlying facilities would be owned by Oracle.

If delivered, the expansion could support a larger domestic recommendation-system operation and new business-facing AI services. Its commercial success would depend on whether those workloads generate enough value to justify the infrastructure commitments—a relevant concern given reports that AI spending is already pressuring ByteDance’s profits. Expert analysis of comparable projects points to two competing approaches. Google (GOOGL) and Xcel (XEL) argue that large technology customers can finance cleaner, more reliable grids. Energy expert Jigar Shah argues that distributing smaller facilities where spare grid capacity already exists can reduce unnecessary construction and costs. Neither assessment is a TikTok-specific forecast, but both identify the central execution issue: obtaining computing capacity is inseparable from deciding how to supply—and pay for—the electricity behind it.

Correction: An earlier version of this article misstated the timing of CNBC's report. It was published on October 23, 2026, not October 24. The article has been updated to reflect that TikTok did not respond to a request for comment.