Direct answer: at official list prices on 8 November 2026 (USD, US East/US West, excluding tax and discounts), "general-purpose SSD block storage" has converged: AWS EBS gp3 0.08/GB-month, Azure Premium SSD v2 about 0.0803/GiB-month, GCP Hyperdisk Balanced 0.08/GiB-month, and all include roughly 3,000 IOPS + 125-140 MBps of baseline performance; above that, AWS/Azure/GCP charge about 0.005/IOPS-month for IOPS and 0.04/MBps-month for throughput. For 1 TiB at 3,000 IOPS, the storage-only monthly cost is about 81.92 / 82.23 / 81.92; for 1 TiB at 10,000 IOPS and 500 MBps it is about 131.92 / 132.97 / 131.32. The real gap is in the high-performance tier: 1 TiB, 10,000 IOPS and 500 MBps costs about 778 on io2, 793 on Azure Ultra and 448 on GCP Hyperdisk Extreme. The conclusion: in the general tier compare included IOPS/throughput plus snapshot/replication; in the high-performance tier compare price per IOPS plus performance ceilings.
This guide is for cloud platform, DBA, SRE, Kubernetes storage and procurement teams. It compares AWS EBS, Azure managed disks and Google Cloud Persistent Disk/Hyperdisk storage prices, IOPS/throughput billing, 1 TiB/10 TiB TCO, the high-performance tier and a 90-day optimization checklist. Prices come from the AWS EBS pricing page, the Azure Managed Disks pricing page and Azure Retail Prices API, and the Google Compute Engine disk/image pricing page (all accessed 8 November 2026). Official price facts, editorial calculations and exclusions are kept separate. This is not procurement advice.
(1) General-purpose SSD is commoditized: gp3 / Premium SSD v2 / Hyperdisk Balanced all sit around 0.08, with about 3,000 IOPS + 125-140 MBps baseline; differences are overage rates, minimum sizes and snapshot/replication. (2) The high-performance tier is where prices diverge: io2/Ultra charge 0.05-0.065 per IOPS, Hyperdisk Extreme about 0.032; at 1 TiB/10k IOPS the gap is nearly 2x. (3) Block storage bills provisioned capacity, not actual use; measure IOPS/throughput utilization and right-size instead of assuming "bigger disk = more performance".
Billing basis: capacity + IOPS + throughput
Block storage is not "GB only": general-purpose tiers include baseline IOPS/throughput and charge separately above it; high-performance tiers split capacity, IOPS and throughput; HDD/cold tiers are cheap per GiB but slower. At minimum, align capacity (1 TiB = 1,024 GiB; AWS EBS states GB = 1024^3 bytes), baseline IOPS/throughput, overage rates, minimum capacity/performance, snapshot/replication/cross-region and tax/discounts. All three bill provisioned capacity whether used or not, unlike usage-based object/file storage.
| Item | AWS EBS | Azure managed disks | GCP PD/Hyperdisk |
|---|---|---|---|
| Capacity unit | GB-month (page: GB = 1024^3 bytes) | GiB-month | GiB-month |
| General SSD | gp3: 0.08/GB-month | Premium SSD v2: about 0.0803/GiB-month | Hyperdisk Balanced: 0.08/GiB-month |
| Baseline | 3,000 IOPS + 125 MBps | 3,000 IOPS + 125 MBps | 3,000 IOPS + 140 MBps |
| IOPS overage | gp3 0.005/IOPS-month | Premium SSD v2 about 0.0051/IOPS-month | 0.005/IOPS-month |
| Throughput overage | gp3 0.04/MBps-month | Premium SSD v2 about 0.0401/MBps-month | 0.04/MBps-month |
| High-performance | io2: 0.125/GB-month + IOPS from 0.065 | Ultra: about 0.1197/GiB-month + IOPS 0.0496 + throughput 0.3497 | Hyperdisk Extreme: 0.125/GiB-month + IOPS 0.032 |
Sources: AWS EBS pricing page (gp3/io2, baseline, IOPS/throughput); Azure Managed Disks pricing page and Azure Retail Prices API (eastus, 8 November 2026); Google Compute Engine disk/image pricing page (us-central1, 8 November 2026). Editorial summary; Azure monthly prices converted at 730 hours; the bill is authoritative.
General-tier TCO: 1 TiB / 10 TiB at 3,000 IOPS
Within the baseline (3,000 IOPS, 125-140 MBps), 1 TiB costs almost the same: gp3 81.92, Premium SSD v2 82.23, Hyperdisk Balanced 81.92; 10 TiB costs 819.20 / 822.27 / 819.20. The difference is cents to a few dollars per month, so the real decision factors become overage, snapshot/replication, cross-AZ/multi-region and bundling with compute/network/commitments. Do not migrate platforms to save 0.3%.
| Option | 1 TiB/month | 10 TiB/month | Note |
|---|---|---|---|
| AWS EBS gp3 | 81.92 | 819.20 | Includes 3,000 IOPS/125 MBps |
| Azure Premium SSD v2 | 82.23 | 822.27 | Includes 3,000 IOPS/125 MBps |
| GCP Hyperdisk Balanced | 81.92 | 819.20 | Includes 3,000 IOPS/140 MBps |
Editorial calculation: 1 TiB = 1,024 GiB; capacity x unit price; excludes snapshots, replication, cross-region, IOPS/throughput overage, tax and discounts. Azure unit price converted from 0.00011/GiB-hour x 730 = about 0.0803; GCP 0.000109589 x 730 = 0.08.
Above baseline: 10,000 IOPS / 500 MBps TCO
Once above baseline, the three still track each other: 1 TiB, 10,000 IOPS and 500 MBps costs about gp3 131.92, Premium SSD v2 132.97 and Hyperdisk Balanced 131.32. The math is capacity + excess IOPS x rate + excess throughput x rate: gp3 7,000 x 0.005 + 375 x 0.04; Azure 7,000 x 0.0051 + 375 x 0.0401; GCP 7,000 x 0.005 + 360 x 0.04 (GCP baseline is 140 MBps). The general-tier competition has shifted to baseline allowances, linear overage and performance ceilings/burst.
| Option (1 TiB) | Capacity | IOPS overage | Throughput overage | Total/month |
|---|---|---|---|---|
| AWS gp3 (10k IOPS/500 MBps) | 81.92 | 35.00 | 15.00 | 131.92 |
| Azure Premium SSD v2 | 82.23 | 35.70 | 15.04 | 132.97 |
| GCP Hyperdisk Balanced | 81.92 | 35.00 | 14.40 | 131.32 |
Editorial calculation: AWS gp3 above 3,000 IOPS at 0.005, above 125 MBps at 0.04; Azure above 3,000 IOPS at 0.0051, above 125 MBps at 0.0401; GCP above 3,000 IOPS at 0.005, above 140 MBps at 0.04. Excludes snapshots, replication, tax and discounts.
High-performance tier: 1 TiB / 10,000 IOPS / 500 MBps
The high-performance tier diverges: io2 about 778 (128 capacity + 650 IOPS), Azure Ultra about 793 (122.57 + 496 + 174.85 throughput), GCP Hyperdisk Extreme about 448 (128 + 320 IOPS). GCP's price per IOPS is about 0.032, below io2's 0.065 and Ultra's 0.0496; but also check maximum IOPS/throughput, per-volume limits, AZ/multi-attach, snapshot/replication and instance constraints. Do not choose on this table alone.
| Option (1 TiB/10k IOPS/500 MBps) | Capacity | IOPS | Throughput | Total/month |
|---|---|---|---|---|
| AWS io2 | 128.00 | 650.00 | Included | 778.00 |
| Azure Ultra | 122.57 | 496.00 | 174.85 | 793.42 |
| GCP Hyperdisk Extreme | 128.00 | 320.00 | Included | 448.00 |
Editorial calculation: io2 0.125/GB-month + 0.065/IOPS-month (first 32,000 IOPS); Ultra about 0.1197/GiB-month + 0.0496/IOPS-month + 0.3497/MBps-month; Hyperdisk Extreme 0.125/GiB-month + 0.032/IOPS-month. Excludes Ultra vCPU reservation in some configurations, snapshots, replication, tax and discounts; performance ceilings/instance compatibility need separate checks.
Hidden costs and selection: do not stop at per-GiB
The hidden costs of block storage are provisioned-but-unused capacity/IOPS/throughput, snapshots/backups (covered separately), cross-region/asynchronous replication, orphan/unattached volumes, overuse of the high-performance tier and the "bigger disk = more performance" default. Start from the workload: general web/databases use gp3/Premium SSD v2/Hyperdisk Balanced; high IOPS/low latency use io2/Ultra/Hyperdisk Extreme; throughput/cold use st1/sc1, Standard HDD, Hyperdisk Throughput/Standard PD; Kubernetes also depends on CSI/StorageClass defaults.
| Scenario | Prefer | Check first |
|---|---|---|
| General web/app/small DB | gp3 / Premium SSD v2 / Hyperdisk Balanced | Whether baseline IOPS/throughput is enough |
| High-IOPS/low-latency DB | io2 / Ultra / Hyperdisk Extreme | Price per IOPS, per-volume ceiling |
| Throughput/sequential | st1 / Standard HDD / Hyperdisk Throughput | Throughput ceiling, burst |
| Cold/archive disk | sc1 / Standard HDD / Standard PD | Latency, snapshot/restore |
| Kubernetes PVC | CSI default StorageClass | Volume type, expansion, snapshots |
| Multi-AZ/DR | Regional replication/multi-zone disks | Replication traffic, RPO/RTO |
| Cost-sensitive/steady | Right-size capacity+IOPS; commitments | Discount coverage |
Editorial selection framework, not a vendor recommendation; volume types, performance ceilings, replication and discounts follow official pages/instance specs.
90-day optimization checklist
Use the first 90 days to establish which volumes exist, how large they are, actual IOPS/throughput, who uses them and their snapshot/replication. The order is inventory, utilization, tiering/right-sizing, orphan cleanup, snapshot/replication, procurement. The most common waste: provisioning for peak all year, cold data on the high-performance tier, resizing capacity without adjusting IOPS, forgotten test volumes and snapshots/replication without retention.
| Phase | Action | Output |
|---|---|---|
| Weeks 1-2 | Inventory volume type/capacity/IOPS/throughput/attachment/owner | Block-storage list, cost baseline |
| Weeks 2-3 | Review CloudWatch/Azure Monitor/Cloud Monitoring IOPS/throughput/latency | Utilization, peak/average |
| Weeks 3-5 | Tier by workload; right-size capacity/IOPS/throughput | Optimization list, savings estimate |
| Weeks 5-7 | Delete orphan/unattached/expired test volumes | Deletion list |
| Weeks 6-8 | Check snapshot/replication/backup retention and restore | Backup cost, RTO |
| Weeks 8-10 | Evaluate commitments/reservations/storage pools | Procurement recommendation |
| Weeks 10-12 | Set budget alerts, monthly review, automation | Alerts, review process |
Editorial optimization advice, not an official template; capacity/performance/backup/compliance follow enterprise standards and vendor documentation.
FAQ
How much does 1 TiB of block storage cost per month?
At official prices on 8 November 2026, 1 TiB = 1,024 GiB and 3,000 IOPS baseline: gp3 about 81.92, Azure Premium SSD v2 about 82.23, GCP Hyperdisk Balanced about 81.92. Excludes snapshots, replication, overage, tax and discounts.
Why are the three almost identical?
General-purpose SSD is highly commoditized: capacity around 0.08, baseline 3,000 IOPS + 125-140 MBps, IOPS overage 0.005 and throughput 0.04. Differences are more about minimum sizes, performance ceilings, snapshot/replication, discounts and ecosystem than per-GiB price.
Which is cheaper for high IOPS?
For 1 TiB/10k IOPS/500 MBps, the editorial estimate is GCP Hyperdisk Extreme about 448, AWS io2 about 778 and Azure Ultra about 793. Also check per-volume ceilings, instance compatibility and whether throughput is charged separately.
How are IOPS/throughput billed?
General tiers include a baseline (AWS/Azure 3,000 IOPS + 125 MBps, GCP 3,000 + 140), then charge per IOPS/per MBps-month above it; high-performance tiers split capacity, IOPS and throughput. Measure utilization before provisioning.
Are snapshots/replication included?
No. Snapshots, backups and cross-region/asynchronous replication are billed separately; see the cloud snapshot cost guide for incremental/archive/retrieval logic. Include them in the block-storage TCO.
Block vs file vs object storage?
Block storage attaches to a single VM/instance with low latency, suited to databases/system disks; file storage is shared NFS/SMB; object storage is for massive unstructured/archive/HTTP API data. Do not use block storage for shared files or massive archives.
How can block-storage costs be optimized?
Right-size capacity/IOPS/throughput by utilization, tier cold data down, delete orphan volumes, review snapshot/replication retention, use commitments/reservations and appropriate Kubernetes StorageClasses, and set budget alerts.
Will prices change?
Yes. Region, currency, discounts, commitments and product updates all matter; recheck the official pages and bill before publishing or purchasing. This article uses US East/US West list prices only.
Research method and limitations
This article separates "official price facts" from "editorial calculations." AWS gp3 0.08/GB-month, 3,000 IOPS/125 MBps baseline, 0.005/IOPS and 0.04/MBps overage, and io2 0.125/GB-month + 0.065/IOPS come from the AWS EBS pricing page. Azure Premium SSD v2 about 0.0803/GiB-month, 3,000 IOPS/125 MBps baseline, about 0.0051/IOPS and 0.0401/MBps overage, and Ultra about 0.1197/0.0496/0.3497 come from the Azure Managed Disks pricing page and Azure Retail Prices API (eastus, 8 November 2026). GCP Hyperdisk Balanced 0.08/GiB-month, 3,000 IOPS/140 MBps baseline, 0.005/IOPS and 0.04/MBps overage, and Hyperdisk Extreme 0.125/GiB-month + 0.032/IOPS come from the Google Compute Engine disk/image pricing page (us-central1, 8 November 2026). TCO, selection framework and the 90-day checklist are editorial.
Eight limitations: (1) only public list prices, excluding tax, enterprise discounts, commitments, support and migration; (2) only US East/US West, other regions differ; (3) capacity normalized to 1 TiB = 1,024 GiB; AWS states GB = 1024^3 bytes; (4) snapshots/backups/replication/cross-region/egress are excluded; (5) no real IOPS/throughput/latency benchmark; (6) not all volume types (st1/sc1, Standard HDD, Hyperdisk Throughput, etc.) are covered; (7) Kubernetes CSI/instance specs/multi-attach limits are not assessed; (8) prices/products change, so recheck official pages and the bill before publishing or purchasing.
Sources and basis
- AWS, Amazon EBS Pricing: aws.amazon.com/ebs/pricing. gp3/io2, 3,000 IOPS/125 MBps baseline, IOPS/throughput overage, GB = 1024^3. Accessed 8 November 2026.
- Microsoft Azure, Managed Disks Pricing: azure.microsoft.com/en-us/pricing/details/managed-disks. Premium SSD v2/Ultra, 3,000 IOPS/125 MBps baseline, IOPS/throughput billing. Accessed 8 November 2026.
- Azure Retail Prices API (armRegionName=eastus, productName contains Premium SSD/Ultra): prices.azure.com/api/retail/prices. 0.00011/GiB-h, 0.000007/IOPS-h, 0.000055/MBps-h, Ultra 0.000164/0.000068/0.000479. Accessed 8 November 2026.
- Google Cloud, Disk and image pricing: cloud.google.com/compute/disks-image-pricing. Hyperdisk Balanced/Extreme, 3,000 IOPS/140 MBps baseline, IOPS/throughput rates. Accessed 8 November 2026.
Data cut-off 8 November 2026; prices/specs follow the official AWS/Azure/Google pages and bills; this is not procurement, commercial or legal advice.