Boardrooms now view strategically sensitive assets as untradeable across the US-China divide. Companies must choose between divesting or facing severe penalties for investing in blacklisted jurisdictions. To make matters worse, regulatory screening is broadening, and export controls are proliferating at an accelerating pace. On that reading, TikTok USDS Joint Venture LLC – established on 22 January 2026, a day before the final statutory deadline – is perhaps the exception: a transaction so freighted that it took two presidents to broker.
That reading is wrong, and it is expensive. Consider what happened on 19 July 2026, six months after closing. The Justice Department's Office of Legal Counsel advised that TikTok is no longer barred from federal government devices, on the reasoning that the joint venture operates independently of ByteDance, is majority-owned by Americans, and has revised the algorithm and cybersecurity programme that prompted the original prohibition. A government that spent four years treating the application as a security threat now permits it on its own staff's phones. That is what resolution looks like, reached by a route most corporate planners have not understood.
The clean sale was never available
The old template for settling a sovereignty dispute over a company was to sell the company: one buyer, one jurisdiction, clean title. That failed here twice – under the Oracle-Walmart framework in 2020, once Beijing added recommendation algorithms to its export-control catalogue.
What succeeded was disaggregation. The American business was unbundled into severable layers, each allocated to whichever sovereign demanded it most. User data went into domestic infrastructure under domestic custody. Governance went to a seven-member board with six American seats, ByteDance holding one and none on the security committee. Equity went to a consortium – Oracle, Silver Lake and Abu Dhabi's MGX at 15% each – with the Chinese parent capped at 19.9%, just below the statutory ceiling. Moderation and software assurance went to the venture. But the recommendation algorithm was licensed rather than transferred, and the commercial rails – e-commerce, advertising, marketing – stayed with the parent under an interoperability framework.
Read the allocation against what each government needed to say afterwards. Washington claimed control: majority ownership, domestic data, an American board. Beijing claimed nothing had been surrendered: the algorithm remained Chinese property, exported under licence and subject to state review. Both claims are true at once, because they attach to different layers of the same business. The deal closed not because the underlying dispute was settled but precisely because it was left open – each state conceding on the dimensions it could afford, neither conceding the principle.
That is the art of the deal, and it has nothing to do with charm or leverage. It is the construction of an outcome each side can describe to its own public as a win.
The price is visible, and it is the point
However, none of this comes freely. The transaction reportedly valued TikTok's US business at roughly $14 billion, which was far below analyst estimates. That gap is not a valuation error; it is perhaps the visible political risk premium, such as the cost of a forced timeline, a constrained buyer pool, and a structure that leaves the acquirer dependent on a licence it does not control. What is more is to add two and a half years of limbo, and licence terms that remain unpublished – a lasting discount on any valuation an outsider attempts.
The framing is straightforward. Geopolitical uncertainty raises transaction costs; it does not foreclose transactions. The question is thus not whether a deal could be politically insensitive but whether the discount is bearable against the alternative, - here, in particular, considering the total loss of a market of 200 million users and 7.5 million businesses.
The strongest objection, and why it fails
Some argue that what TikTok signed, is not a deal but a hostage arrangement: the venture's core asset is a licence Beijing can condition or revoke. The scrutiny has not stopped, either – the venture's security chief is due before the House Select Committee on China in September.
However, revocable permissions have become a standard feature of cross-border commerce, rather than an anomaly specific to this transaction. Export licenses, entity listings, sanctions, and outbound investment screens all grant governments significant control. A license subject to withdrawal is not fundamentally different from a supply chain vulnerable to sanctions. The critical consideration is whether such dependencies are recognised, appropriately priced, and effectively hedged.
The decisive evidence sits in the comparison. CK Hutchison attempted the clean version – a whole-asset sale of strategic ports to American capital – and Beijing stalled it. The firm that tried to transfer cleanly was blocked; the firm that unbundled and licensed closed its deal and, this month, was cleared onto federal handsets.
When considering the new M&As, four propositions follow. [1) Map sovereign demand by attribute, not by entity: states rarely want the whole company, only a guarantee about one thing. 2) Build severability into systems and contracts before pressure arrives. 3) Concede governance before equity, since board seats and audit rights are cheaper than ownership and usually satisfy the political requirement. And 4) recruit third-jurisdiction capital early: neutral ballast is now a market. Happy for suggestions]
Strategic assets no longer transfer seamlessly between geopolitical blocs, yet transactions still occur. Firms that delay action in anticipation of improved conditions risk falling behind those that proactively develop resilient arrangements.
