Under Google’s design, those fragments live in separate environments. Some can sit on a customer’s own on-premises hardware. Others sit in Google Cloud. Still others can be held by different cloud providers entirely.
The result is that compromising one environment is not enough. An attacker would need to breach several independent systems at once.
Google supports threshold schemes including 2-of-3 and 5-of-7. A 2-of-3 setup means any two of three key shares can authorize a transaction. A 5-of-7 arrangement requires five of seven shares to agree.
The policy layer and the trust question
Google’s service includes a policy engine that runs inside Google’s Confidential Space. That engine enforces rules before any key signing happens. Examples include spending limits and approved lists of recipients.
There is also the obvious question of whether customers should trust Google with any piece of their keys. Google’s answer is that it does not have plaintext access to the key shares it helps manage.
Who it is built for
This is not a wallet for retail users. The service is aimed at institutional clients, with banks as a particular focus.
The intended use cases are safeguarding and transferring digital and tokenized assets.
Right now, the service is in private preview. A select group of design partners and major financial institutions has access. Google expects general availability in early 2027.
Google is also eating its own cooking. Alphabet, Google’s parent company, already uses the system to protect its own assets on the Ethereum and Solana blockchains.
Background: the single-key problem
MPC is not a new concept in crypto custody. Specialist firms such as Fireblocks and BitGo have built businesses around securing digital assets for institutions. What is new is a hyperscale cloud provider entering the same arena with its own product.
What this means
For the custody market, Google Cloud’s arrival is a competitive event. Fireblocks and BitGo now face a rival with deep enterprise relationships, an existing cloud footprint inside many banks, and a recognizable brand in boardrooms.
The multi-cloud design is also a notable choice. Allowing shares to sit with other cloud providers means a customer is not forced to put every egg in Google’s basket.
There are things to watch. The first is the general availability timeline, currently expected in early 2027. The second is which financial institutions publicly sign on beyond the unnamed design partners. The third is how the incumbents respond. Fireblocks and BitGo have years of operating history in this niche. Google has scale.
Leave a Reply