Google Cloud will delete remaining RPC endpoints and shut down Blockchain Node Engine on December 15

Google's shutdown runs on two deletion rules, and they do not line up ^[4][6]. Crypto Briefing reports that every remaining node and endpoint will be deleted on December 15, 2026 ^[4]. The same account says Google will automatically delete leftover RPC endpoints at shutdown, while dedicated nodes are on the customer's list to clean up ^[6]. Google's migration checklist ends with the customer deleting its own legacy Blockchain Node Engine nodes ^[5]. On that reading, a team with dedicated nodes owns its own deletion ^[6].

The window looked long when it opened. New node creation and new RPC endpoints stopped on June 15 ^[3], so the wind-down runs 183 days end to end ^[14]. Existing nodes get limited support and critical updates in that time, and customers should not expect new features ^[3]. Google has stopped building the product ^[3]. Every customer still running production workloads on it now has a migration project on its roadmap that it did not plan for, as Crypto Briefing puts it ^[11].

The money can go three ways from here (or rather, two that Google wrote down and one it left out). A team can buy Quicknode through its Google Cloud Marketplace listing, which comes with Google Cloud incentives, or sign up with Quicknode directly ^[7]. Or it can take on the work Blockchain Node Engine existed to remove: syncing the chain, keeping up with client upgrades, monitoring uptime and paying for storage ^[8]. Crypto Briefing's report does not include customer counts, Quicknode's prices or the size of the Marketplace incentives.

I'd expect most of this spend to land on Quicknode, because Google named it the officially recommended RPC infrastructure provider and wrote the path into its own documentation ^[2][13]. Crypto Briefing calls that the sort of distribution infrastructure companies usually have to fight for ^[13]. The counter-case is that a team forced to re-platform anyway has every reason to price the move against other providers, or against running its own nodes ^[8]. The view is wrong if the Marketplace incentives turn out small and teams shop the move on cost. Those incentive terms decide how much Google's default is worth to Quicknode ^[7].

Archive nodes carry the hardest part of the move. The shared RPC service covered only Ethereum Mainnet and the Holesky testnet ^[10]. For those users I'd expect the job to be mostly the checklist step of updating RPC URLs and application configurations ^[5]. Dedicated nodes, mainly for Ethereum Mainnet and its testnets, Polygon and Solana, came in full and archive versions ^[9][10]. An archive node keeps the complete history of past states, used for analytics, indexing and some auditing ^[9]. Crypto Briefing notes that historical data queries are often the hardest to replicate cleanly ^[12]. Checking those queries on a new provider is the validation step, and it has to be finished before December 15 ^[5][4].