In the same window AWS made S3 Vectors generally available (store and query embeddings for RAG directly in S3, no separate vector database) and Google shipped Rapid Storage (a zonal bucket with sub-millisecond reads/writes and 6 TB/s throughput). Neither is a price cut; both attack the cost of idle GPUs and standalone vector databases. The premium-storage fight has moved decisively from $/GB to latency and AI-readiness — the axis where the hyperscalers, not the budget providers, hold the advantage.
With new legacy-account creation already blocked and October 13 roughly three weeks away, any enterprise still on legacy or GPv1 Blob Storage is down to its final weeks to migrate proactively rather than let Microsoft apply default GPv2 settings — and a default billing baseline — automatically. Azure Storage Discovery (still public preview) is the fastest way to inventory an estate; teams that haven’t started are cutting it close on the clearest forced-action date on any provider’s 2026 calendar.
Standard internet egress runs $0.087–$0.12/GB on AWS, Azure, and Google versus $0 on Cloudflare R2, Backblaze B2, and Wasabi — at 10 TB/month, roughly $891 on S3 against $0 on R2. With real-world spend routinely two-to-five times the headline storage rate once egress and retrieval are counted, buyers increasingly model exit cost before they pick a hot tier, and zero-egress providers have moved from niche to default consideration.
With Google’s GA Model Context Protocol server for Cloud Storage already in production and Azure Blob still without a native equivalent, re:Invent is the concrete, dated checkpoint for whether AWS closes the S3 MCP gap — and whether Azure finally joins the AWS–Google Cross-Cloud Interconnect effort it said would happen “later this year.” Teams building AI-agent or multicloud storage topologies around either question now have roughly ten weeks before both likely resolve at one event.
The emerging 2026 blueprint assigns storage by role rather than standardizing on one cloud: Cloudflare R2 or Wasabi for zero-egress internet serving, Backblaze B2 (~$6.95/TB) for cheapest cold archive, and S3 with Object Lock for regulated copies. The hyperscalers keep the compliance and integration anchor while quietly ceding the high-egress serving tier — a structural concession that pricing pages don’t show but architecture diagrams increasingly do.
Amazon’s ~$200B, Google’s $175–185B, and Microsoft’s $110–120B guidance figures are unchanged again this week — a quiet hold after months of upward revisions. The spending keeps arriving into a market where AWS, Azure, and Google still hold roughly 31%, 25%, and 11% infrastructure share, reinforcing why pricing and feature experimentation, not share competition, is where the real 2026 storage action has been.
AWS Storage
AWSS3 Vectors is now GA with higher scale and performance, letting teams store and query vector embeddings directly in S3 for retrieval-augmented generation at subsecond latency with no standalone vector database to run. It reframes S3 from passive object store toward an AI-native data layer and undercuts the cost of purpose-built vector DBs for embedding-heavy workloads — the clearest sign AWS is answering the AI-storage competition on features, not price.
AWS S3 Vectors →S3 now supports sort and z-order compaction for Apache Iceberg tables, cutting scan times and downstream engine costs for lakehouse workloads. With re:Invent roughly ten weeks away, it also sets up this digest’s standing watch item — whether AWS ships a native S3 MCP server to match Google Cloud Storage’s GA offering. No public AWS signal on the MCP question has surfaced yet.
AWS Blog →Azure Storage
AZURENew legacy-account creation is already blocked, and on October 13 Azure auto-migrates remaining legacy Blob accounts to GPv2. With roughly three weeks left, any enterprise that hasn’t inventoried its Azure Blob estate (Storage Discovery, still in public preview, is the fastest path) is working against a hard calendar constraint — inaction itself becomes a migration Microsoft executes on the customer’s behalf.
Microsoft Learn →Still in preview, Massive v2 targets the same GPU-utilization problem Google is chasing: keep high-throughput accelerators fed without forcing customers to shard hot data across accounts. It positions Azure to compete on AI-pipeline performance rather than $/GB — but as a preview with no GA date, it remains a roadmap signal, not a purchasable capability, heading into Ignite season.
Microsoft Learn →GCP Storage
GCPRapid Storage, a new zonal Cloud Storage bucket, delivers under-1ms random read/write latency and up to 6 TB/s of throughput — roughly 20x faster data access aimed squarely at GPU-bound AI training and inference. Like GCSFS prefetching before it, it targets the cost of idle accelerators rather than headline storage price, extending Google’s strategy of competing on AI-pipeline performance over $/GB.
Google Cloud Blog →Google’s Cross-Cloud (and Partner) Interconnect moves data from AWS S3 or Azure Blob into Cloud Storage over dedicated private links rather than the public internet, trimming the egress and unpredictability that dominate multicloud transfer bills. It is a quiet but strategic wedge: making it cheaper to leave a rival cloud’s storage is itself a competitive lever as egress economics move to the center of buyer decisions.
Google Cloud →Industry
CROSS-CLOUDStandard 2026 internet egress runs $0.087–$0.12/GB on AWS, Azure, and Google versus $0 on Cloudflare R2, Backblaze B2, and Wasabi — at 10 TB/month that is roughly $891 on S3 against $0 on R2. Real-world spend routinely runs two to five times the headline storage rate once egress, requests, and retrieval are counted, which is why buyers now model exit cost before they pick a hot tier.
Finout →Zero-egress providers have moved from niche to default consideration for any workload with meaningful data movement, and the emerging blueprint splits by role rather than standardizing on one cloud: R2 or Wasabi for internet-facing serving, Backblaze B2 for cheapest cold backup, and S3 with Object Lock for regulated copies. The hyperscalers keep the compliance and integration anchor while quietly ceding the high-egress serving tier.
CloudZero →| Metric | EBS gp3 | Premium SSD v2 | pd-balanced |
|---|---|---|---|
| Storage $/GB/mo | $0.08020% cheaper than gp2 | $0.100/GiBHourly billing | $0.100Default disk type |
| Baseline IOPS | 3,000Free — independent of size | 3,000Free — independent of size | 6/GBScales with disk size (500 GB = 3K) |
| Baseline Throughput | 125 MB/sFree — independent of size | 125 MB/sFree — independent of size | 0.28 MB/s/GBScales with disk size |
| IOPS Provisioning | Independent$0.005/IOPS/mo · up to 16K | IndependentSeparately provisionable · up to 80K | Tied to sizeNo independent IOPS control |
| Max IOPS (family) | 256,000io2 Block Express | 400,000Ultra Disk | 120,000pd-extreme |
| Snapshots | Incremental, S3-backedFast Snapshot Restore available | Instant Access (Pv2/Ultra)Instant creation + restore · free until Jul ’26 | IncrementalCloud Storage-backed · regional |
| Dimension | AWS EBS | Azure Managed Disks | GCP Persistent Disk |
|---|---|---|---|
| Price Leader | gp3 at $0.08/GB Cheapest | Pv2 at $0.10/GiB | pd-balanced at $0.10/GB |
| IOPS Flexibility | Independent (gp3) Winner | Independent (Pv2) Winner | Scales with disk size only |
| Latency | io2: sub-ms | Ultra: sub-500µs avg Winner | pd-ssd: single-digit ms |
| Snapshots | Incremental + Fast Restore | Instant Access (Pv2/Ultra) Winner | Incremental, Cloud Storage-backed |
| Shared/Multi-Attach | io2 multi-attach (16 instances) | Broadest: Ultra/Premium/Standard SSD Winner | Multi-writer (up to 2 VMs) |
| Cold Block $/GB | sc1: $0.015/GB Cheapest | Std HDD: ~$0.04/GB | pd-standard: $0.040/GB |
Azure Premium SSD v2 charges $0.0005/IOPS/month above its 3K baseline — that’s 130x cheaper than AWS io2 Block Express ($0.065) and 10x cheaper than AWS gp3 and GCP Hyperdisk ($0.005). At 64,000 IOPS, the gap is staggering: Azure Pv2 costs $30.50/mo for IOPS alone, while AWS io2 charges $4,160/mo and GCP Hyperdisk Extreme charges $640/mo. If your workload is IOPS-intensive and you’re not on Azure, you’re leaving significant money on the table.
| IOPS Level | AWS gp3 | AWS io2 BE | Azure Pv2 | GCP HDB | GCP HD-Ext |
|---|---|---|---|---|---|
| 3,000 | $0 (free) | $195 | $0 (free) | $0 (free) | $30 |
| 10,000 | $35 | $650 | $3.50 | $35 | $100 |
| 64,000 | $305 | $4,160 | $30.50 | $305 | $640 |
| 256,000 | N/A (80K max) | $16,640 | N/A (80K max) | N/A (160K max) | $2,560 |