Google object storage is not a block or file based volume to mount on a GCP instance. Applications are intended to read and write blobs directly. Or perhaps copy into and out of small local scratch spaces.
An advantage to doing so is access to a near-unlimited number of blobs, without having to mange large block volumes. Disadvantages include expensive per "file" access, due to network latency and billing. And not looking like a file means applications will need to treat storage differently for the best results, which is work to develop.
Technically, there exists a FUSE file system to sort of look like a file system. It will be unusably slow and very expensive at petabyte scale.
As to traditional block storage, GCP persistent disks can be up to 256 TB per host, in 64 TB LUNs. Create LVM volumes, then file systems, and mount them. Maximum capacity disks will hit per instance performance limits though, a distributed storage system over multiple hosts may become necessary. Also gets expensive, these are fast tiers of hot storage and priced accordingly.
That covers object, and block. For completeness, GCP does have managed file share option, over NFS. While this does give you mountable volumes, the Filestore capacity limit is 100 TiB.
