The operative term in the memorandum is an administrative one: the agencies will investigate common targets in parallel and then settle between themselves which jurisdiction prosecutes which suspect [3], and several overlapping cases have already been picked out [4]. That solves an allocation problem inside the prosecution business, which is worth having but is different from reaching the money.
Reaching the money has a track record. In May the Scam Center Strike Force ran an information-sharing operation with the NCA and agencies from Australia, Canada, New Zealand and Thailand [12], and the output was more than 1.4 million social media and email accounts disrupted, seven arrests in Thailand, and more than $3.8 million in crypto frozen by private companies [13]. Set the frozen figure against the $8.65 billion the FBI's complaint centre logged for 2025 [6] and you get 0.044% [1], four and a half basis points of a single year of reported losses. Coinbase supplied over $3 million of that $3.8 million, about 79% [14][2]. Meta, Microsoft and Starlink took action against accounts and infrastructure in the same effort [14], and 1.4 million accounts against seven arrests is 200,000 accounts per arrest [7], a ratio that reflects infrastructure work far more than it reflects prosecution.
The loss numbers deserve the same treatment. Reported cyber-enabled investment fraud went from $4.57 billion in 2023 to $8.65 billion in 2025 [6], which compounds at about 38% a year [4], and the DOJ separately estimates the schemes cost Americans roughly $10 billion annually while cautioning that victim-report data could substantially understate the real total [7][9]. Those two statements sit awkwardly together: $10 billion is only about 16% above $8.65 billion [3], and since investment fraud is a subset of cyber-enabled fraud, which was almost 85% of everything reported to the centre last year [8], all reported losses must have been at least $10.2 billion [5]. The government's estimate of what these schemes cost is therefore at or below the floor of what victims reported across every fraud category, so either the $10 billion is drawn narrowly, or it has not absorbed its own caveat.
The read shifts depending on three factors, each weightier than the last. The compounds are in Southeast Asia and the named targets are Chinese organised crime groups [11], so arrest volume so far has tracked host-state participation rather than paperwork between the countries where the victims live: the Dubai police-led operation with the FBI and China's Ministry of Public Security produced 276 arrests and closed at least nine centres [15]. Next, the binding constraint may be domestic law where the compounds sit, and Myanmar's parliament approved a bill on 28 July carrying 10 years to life for digital currency fraud, with presidential assent unconfirmed [16]. Last, October could turn the platform coalition from occasional into standing, which is the version where a memorandum compounds.
My read, or rather the narrower version of it: this is a cheap allocation of prosecutorial venue against a loss series growing at 38% a year, described by the DOJ as the first agreement of its kind built to disable scam centres [2], and nothing the sources disclose commits money, headcount or seizure authority to it. If the London session [5] reports a frozen total in the tens of millions, coordination is reaching cash. If it reports another $3.8 million, it is intelligence sharing with a signing ceremony.
