This page compares the OpenMetal Bare Metal Medium v5 against the AWS i4i.8xlarge, the closest EC2 instance by RAM and NVMe storage profile. The comparison is structural: tenancy model, billing model, storage persistence, egress pricing, and compliance posture – not just a spec table. Both platforms can run the same workloads; the difference is in what you own, what you pay per month, and what happens at scale.
See the OpenMetal Medium v5 spec page for full hardware specifications and the egress pricing calculator for cost modeling.
Key Takeaways
- Dedicated cores vs shared vCPUs – OpenMetal provides 24 physical cores reserved entirely for your workload; i4i.8xlarge allocates 32 vCPUs from a shared multi-tenant host with no CPU isolation guarantees.
- Persistent NVMe vs ephemeral instance storage – Micron 7500 MAX data on the Medium v5 persists across reboots, power cycles, and OS reinstalls; i4i instance storage is lost on stop or termination, requiring EBS for durability at added cost and latency.
- Fixed monthly vs per-hour billing – OpenMetal pricing is a flat monthly rate lockable up to 5 years; AWS i4i charges per hour with no long-term price guarantee beyond Reserved Instance commitments.
- 95th-percentile egress vs per-GB transfer – OpenMetal bills public egress on the 95th percentile of sustained rate; AWS charges per GB regardless of traffic pattern, with costs that compound at scale.
- Full IPMI vs SSH/SSM access – OpenMetal provides out-of-band IPMI for power control, BIOS changes, and OS reinstall; AWS EC2 provides SSH or Systems Manager access only, with no equivalent hardware-layer management.
- Hardware-level TDX isolation available – the Medium v5 supports Intel TDX confidential computing via a RAM upgrade; AWS i4i runs on shared Nitro infrastructure with no equivalent per-VM hardware memory encryption.
Spec Comparison
| Component | OpenMetal Medium v5 | AWS i4i.8xlarge | AWS i4i.metal |
|---|---|---|---|
| CPU | 2x Xeon 6505P (dedicated) | 32 vCPU shared | 128 vCPU shared |
| Architecture | Granite Rapids, Intel 3 | Ice Lake | Ice Lake |
| RAM | 256 GB DDR5-6400 | 256 GB | 1,024 GB |
| Local Storage | 6.4 TB NVMe (persistent) | 7.5 TB NVMe (ephemeral) | 30 TB NVMe (ephemeral) |
| Storage Persistence | Persists across all reboots | Lost on stop/terminate | Lost on stop/terminate |
| Private Network | 20 Gbps LACP (included) | Up to 25 Gbps | Up to 75 Gbps |
| Tenancy | Dedicated physical server | Shared multi-tenant host | Dedicated (metal only) |
| Remote Management | IPMI (power, BIOS, KVM) | SSH / AWS SSM only | SSH / AWS SSM only |
| Boot/Data Isolation | Yes – separate RAID 1 boot | No equivalent | No equivalent |
| Confidential Computing | TDX (via RAM upgrade), SGX | Nitro Enclaves | Nitro Enclaves |
| Pricing Model | Fixed monthly | Per-hour (On-Demand or RI) | Per-hour (On-Demand or RI) |
| Egress Model | 95th percentile | Per GB | Per GB |
AWS i4i.metal included for reference – the Medium v5 maps more directly to i4i.8xlarge by RAM.
Ready to Compare Costs for Your Workload?
Tell us about your workload and we’ll help you configure the right deployment — bare metal or Hosted Private Cloud, in any of our four data center regions.
Processor: Dedicated Cores vs Shared vCPUs
The Medium v5 provides 24 physical cores and 48 threads across two Xeon 6505P processors, reserved entirely for your workload. There is no hypervisor tax, no CPU steal, and no interference from neighboring tenant VMs. BIOS-level control allows setting CPU governor, power states, and boot order without a support ticket.
AWS i4i instances allocate vCPUs from a shared host. While AWS Nitro uses a lightweight hypervisor that minimizes overhead, vCPUs are still logical allocations on physical hardware shared with other customers. Workloads sensitive to CPU scheduling variance – latency-critical databases, real-time analytics, blockchain validators – benefit from the determinism of dedicated physical cores.
The Xeon 6505P includes Intel AMX and AVX-512 instruction extensions. On dedicated bare metal, these execute at full width without hypervisor mediation. AWS i4i instances run on Ice Lake processors; AMX support depends on the specific instance generation.
Memory: 256 GB Persistent vs Cloud Instance RAM
Both the Medium v5 and i4i.8xlarge provide 256 GB of RAM at their base configurations. The structural difference is upgradeability and what the RAM enables.
On the Medium v5, RAM is upgradeable on the deployed server: 8 DIMM slots remain open at base configuration, supporting upgrades to 512 GB, 1 TB, or 2 TB without migrating workloads. The 1 TB upgrade activates Intel TDX confidential computing – a security posture change that requires a RAM upgrade on AWS would instead require migrating to a different, larger instance type, incurring downtime and reprovisioning.
AWS i4i instances cannot be upgraded in place; scaling RAM requires stopping the instance, changing instance type, and restarting – with associated downtime and potential storage migration if instance storage data cannot be lost.
Storage: Persistent NVMe vs Ephemeral Instance Storage
Boot and data isolation
The Medium v5’s 6.4 TB Micron 7500 MAX NVMe data drive persists across all reboots, power cycles, and OS reinstalls. Boot/data isolation separates the OS volume (2x 960 GB RAID 1) from data storage, so OS maintenance never risks data volumes.
Micron 7500 MAX data drives
AWS i4i instance storage (NVMe-based) is ephemeral: data does not survive instance stop or termination. Workloads requiring durable local storage on AWS i4i must use Amazon EBS, which adds per-GB-month charges, per-IOPS charges for provisioned IOPS volumes, and network-attached latency overhead compared to local NVMe.
| Metric | OpenMetal (Micron 7500 MAX) | AWS i4i NVMe |
|---|---|---|
| Sequential Read | 7,000 MB/s | ~7,000 MB/s |
| Random Read (4 KB) | 1,100,000 IOPS | ~1,000,000 IOPS |
| Read Latency (typical) | 70 µs | Not published |
| QoS | Sub-1ms at 6-nines | Not published |
| Persistence | Survives all reboots | Lost on stop/terminate |
| Drive model | Micron 7500 MAX (confirmed) | AWS-managed, model varies |
Source for OpenMetal: Micron 7500 NVMe SSD Tech Product Spec Rev. A 10/2023.
Networking
The Medium v5 provides 20 Gbps of LACP-bonded private bandwidth included at no additional cost. All east-west traffic between OpenMetal servers on the same customer VLAN is included – no cross-zone transfer charges, no per-GB metering for internal traffic. DDoS protection is included for up to 10 Gbps per IP.
AWS i4i.8xlarge provides up to 25 Gbps network bandwidth. However, data transfer between AWS Availability Zones is charged per GB, and cross-region transfer is charged at higher rates. Workloads with heavy inter-node communication – database replication, Kubernetes pod traffic, Ceph storage replication – incur measurable AWS cross-AZ costs at scale.
Egress pricing: 95th-percentile billing, not per-GB transfer.
Egress cost comparison: AWS egress from US-East is approximately $0.09/GB. At 50 TB/month, that is approximately $4,600/month in egress charges alone. On OpenMetal, 50 TB/month of bursty egress traffic billed at 95th percentile against a committed baseline would typically cost a fraction of per-GB rates – workloads that burst occasionally pay for their sustained rate, not every byte. At 5 TB/month, the AWS egress cost is approximately $460/month. The gap widens significantly as egress volume grows.
Security and Confidential Computing
The Medium v5 supports Intel TDX (via 1 TB RAM upgrade) and Intel SGX (active by default), providing hardware-level memory isolation for individual VMs and application-layer enclaves respectively. TDX Trust Domains are isolated from the hypervisor; SGX enclaves are isolated from the OS. Both technologies provide remote attestation – cryptographically signed evidence of execution environment for external verifiers.
AWS i4i instances run on the AWS Nitro System. Nitro provides isolation between instances and between customers using a dedicated hardware security chip and lightweight hypervisor. AWS Nitro Enclaves provide application-layer isolation similar in concept to SGX enclaves. Full VM-level memory encryption equivalent to TDX is not confirmed as generally available on i4i instances.
| Feature | OpenMetal Medium v5 | AWS i4i |
|---|---|---|
| Hardware memory encryption (VM layer) | TDX (via RAM upgrade) | Not confirmed |
| Application enclave | SGX (128 GB EPC, default) | Nitro Enclaves |
| Remote attestation | TDX + SGX (hardware-rooted) | Nitro Attestation |
| Physical isolation from other customers | Yes – dedicated server | No – shared Nitro host |
| IPMI / firmware access | Full IPMI, BIOS control | No equivalent |
HIPAA and Regulatory Compliance
OpenMetal holds HIPAA compliance at the organizational level and offers BAAs. The Ashburn facility (NTT DATA VA1) holds HIPAA as a facility-operator certification alongside SOC 1/2 Type II, ISO 27001, PCI DSS, and NIST 800-53 HIGH. Other OpenMetal locations hold PCI DSS and ISO 27001 at the facility level (see primary spec page for details).
AWS maintains a list of HIPAA-eligible services and signs BAAs. AWS operates under the shared responsibility model: customers are responsible for configuring HIPAA-eligible services correctly, with AWS providing the compliant infrastructure layer. OpenMetal provides dedicated hardware, meaning the physical isolation layer is under direct customer control rather than shared with other AWS customers.
When Each Platform Wins
The right platform depends on workload type, billing sensitivity, and compliance posture. Below is a direct summary of where each platform wins.
When OpenMetal Wins
- Sustained 24/7 workloads – fixed monthly pricing is materially cheaper than On-Demand EC2 for workloads running continuously; even Reserved Instances require 1 – 3 year upfront commitments with usage restrictions.
- Egress-heavy applications – video streaming, data distribution, large-dataset exports, and log shipping generate per-GB charges on AWS that compound rapidly above 10 TB/month. OpenMetal’s 95th-percentile billing rewards bursty-but-not-constant egress patterns.
- Compliance and isolation requirements – HIPAA, PCI DSS, and NIST 800-53 workloads that require dedicated physical hardware and auditable facility certifications. TDX and SGX provide hardware-level isolation that exceeds software-only tenant separation.
- Predictable multi-year budgeting – fixed pricing lockable up to 5 years supports OpEx planning for regulated workloads with multi-year audit cycles. AWS pricing can change with 30 days’ notice.
- Full hardware control – BIOS access, IPMI, custom OS installs, and bare-metal Kubernetes or Proxmox deployments that require direct hardware access.
When AWS Wins
- Scale-to-zero and event-driven workloads – serverless and highly variable demand patterns where paying only for active compute time is more economical than fixed hardware.
- Deep AWS ecosystem integration – workloads that rely on managed services (RDS, SQS, DynamoDB, Lambda, SageMaker) that have no direct equivalent on bare metal infrastructure.
- Global edge and multi-region – workloads requiring dozens of geographic regions or AWS CloudFront/Route 53 global edge infrastructure.
- Total spend below $10k/month – at low scale, the operational simplicity of managed cloud and the ecosystem breadth of AWS may outweigh the cost advantage of dedicated hardware.
- Managed compliance tooling – workloads that rely on AWS-native compliance services (Macie, Security Hub, GuardDuty, Config) for continuous compliance monitoring.
Ready to Compare Costs for Your Workload?
Tell us about your current cloud spend and workload profile. We’ll provide a side-by-side cost analysis for the Medium v5 vs your current AWS deployment.
How the Medium v5 Compares to Public Cloud
Cost Model Comparison
| Item | OpenMetal Medium v5 | AWS i4i.8xlarge |
|---|---|---|
| Pricing model | Fixed monthly | Per-hour On-Demand or RI |
| Price lock | Up to 5 years | 1 – 3 year RI only |
| Egress model | 95th percentile | Per GB (~$0.09/GB US-East) |
| Private/inter-node traffic | Included (VLAN) | Per GB cross-AZ |
| Remote management | IPMI included | SSM (additional cost at scale) |
| DDoS protection | Included (10 Gbps/IP) | AWS Shield Standard (included) |
| Storage persistence | NVMe persists by default | EBS required for persistence ($/GB/mo) |
| RAM upgrade path | In-place (no migration) | New instance type required |
| Ramp pricing | Available for migrations | No equivalent |
Illustrative 12-Month TCO Comparison
The following estimates illustrate structural cost differences. All AWS figures require verification against current aws.amazon.com/pricing before use in proposals, marketing, or customer-facing materials.
| Cost element | OpenMetal Medium v5 | AWS i4i.8xlarge |
|---|---|---|
| Compute (monthly x 12) | [See openmetal.io/bare-metal-pricing] | ~$X,XXX/mo x 12 |
| Egress – 20 TB/mo | Included in 95th-pct billing | ~$1,800/mo x 12 = ~$21,600 |
| EBS for durable storage | Not required | ~$0.10/GB/mo x 6,400 GB = ~$640/mo |
| Cross-AZ replication traffic | Included | ~$0.01 – 0.02/GB |
| DDoS protection | Included | Shield Standard included; Shield Advanced additional |
Medium v5 Deployment Options
Bare Metal Dedicated Server
Deploy a Medium v5 as a dedicated server in Ashburn (VA), Los Angeles (CA), Amsterdam (NL), or Singapore. Full root access, IPMI, fixed monthly pricing, and optional 5-year price lock. See openmetal.io/bare-metal-pricing for current rates.
→ View pricing: openmetal.io/bare-metal-pricing
Hosted Private Cloud
A three-node Medium v5 cluster running OpenStack and Ceph provides a managed private cloud environment with Day 2 operations included, deploying in under 45 seconds. No VMware licensing, no hypervisor fees.
| Location | Region | Facility Certifications | Location Page |
|---|---|---|---|
| Ashburn, VA | US-East | SOC1/2 Type II, ISO 27001, ISO 50001, PCI DSS, NIST 800-53 HIGH, HIPAA (facility) | Ashburn |
| Los Angeles, CA | US-West | SOC2, SOC3, ISO 27001, PCI DSS | Los Angeles |
| Amsterdam, NL | EU-West | SOC1/2, PCI-DSS, ISO 27001, ISO 50001, ISO 22301 | Amsterdam |
| Singapore | Asia | BCA Green Mark Platinum | Singapore |
→ View pricing and configuration: openmetal.io/cloud-deployment-calculator
Both deployment paths: available across OpenMetal’s Tier III data center locations. Fixed monthly pricing applies regardless of utilization. No per-hour, per-query, or per-GB billing.
Get a Medium v5 Quote
Tell us about your infrastructure needs and we’ll provide a custom quote for the Medium v5 — as a standalone bare metal server or as part of a Hosted Private Cloud cluster.
- Bare metal: Single-server or multi-server deployments with full root access and IPMI
- Hosted Private Cloud: Three-node OpenStack + Ceph clusters with Day 2 operations included
- Custom configurations: RAM upgrades, additional NVMe drives, TDX enablement
Ramp pricing available for migrations. All deployments include fixed monthly pricing, 99.96%+ network SLA, and DDoS protection.
Specifications, pricing, and availability are subject to change without notice. The information on this page is provided for general guidance and does not constitute a contractual commitment. Contact OpenMetal for current configuration details and pricing. AWS specifications and pricing are sourced from publicly available documentation and may not reflect current rates or configurations.