williamzujkowski/Multi-Cloud Cost Optimizer

Purpose & When-To-Use

Primary trigger conditions:

• Operating workloads across 2+ cloud providers (AWS, GCP, Azure) with monthly spend >$50k
• Seeking cost arbitrage opportunities by placing workloads on most cost-effective cloud
• Need unified view of waste and optimization opportunities across all clouds
• Balancing commitment purchases (RIs, Savings Plans, CUDs) across multiple providers
• High egress costs (>15% of total spend) from cross-cloud data transfer
• Executive request for multi-cloud cost consolidation and reduction
• FinOps team managing multiple cloud providers seeking unified optimization

When NOT to use this skill:

• Single cloud deployment → use finops-cost-analyzer instead
• Multi-cloud strategic planning phase → use cloud-multicloud-advisor first
• Real-time cost tracking → use native cloud dashboards
• Workloads cannot be migrated due to compliance/latency constraints

Value proposition: Identifies 20-35% additional savings beyond single-cloud optimization by leveraging cross-cloud price competition, workload placement optimization, and eliminating multi-cloud waste patterns. Organizations using multi-cloud cost optimization tools achieve 35-68% total cost reductions (CloudZero, accessed 2025-10-26T14:30:00-04:00).

Pre-Checks

Required inputs validation:

python
NOW_ET = "2025-10-26T14:30:00-04:00"

assert len(cloud_accounts) >= 2, "Multi-cloud optimization requires ≥2 cloud providers"
assert all(acc["billing_api_access"] for acc in cloud_accounts), "Billing API access required for all accounts"
assert time_range in ["30d", "90d", "180d"], "Valid time ranges: 30d, 90d, 180d"
assert optimization_scope in ["all", "compute", "storage", "network", "data-transfer"]

# Data freshness check
for account in cloud_accounts:
    if account["last_billing_sync"] > 48h:
        warn(f"{account['provider']} billing data stale; recommendations may be outdated")

# Minimum spend threshold check
total_monthly_spend = sum_monthly_spend(cloud_accounts)
if total_monthly_spend < 50000:
    suggest("Multi-cloud optimization most valuable for monthly spend >$50k")

Authority checks:

• AWS: Cost Explorer API enabled, ce:GetCostAndUsage, organizations:ListAccounts if using AWS Organizations
• GCP: Cloud Billing API enabled, billing.accounts.get, billing.resourceCosts.list permissions
• Azure: Cost Management API access, Reader role on subscriptions, Microsoft.CostManagement/query/action permission

Source citations (accessed 2025-10-26T14:30:00-04:00):

• FinOps Best Practices 2025: https://www.cloudzero.com/blog/finops-best-practices/
• Multi-Cloud Cost Management Guide: https://www.prosperops.com/blog/multi-cloud-cost-management-guide/
• Top 50 FinOps Tools 2025: https://holori.com/20-best-finops-and-cloud-cost-management-tools-in-2025/
• FinOps Framework (Multi-Cloud): https://www.finops.org/framework/

Procedure

Tier 1 (≤2k tokens): Quick Multi-Cloud Cost Health Check

Goal: Identify top 3 cross-cloud optimization opportunities in <5 minutes.

Steps:

1. Fetch unified cost summary for time_range across all providers

• Normalize currency and time periods (AWS monthly, GCP daily, Azure daily → unified monthly)
• Calculate total spend by provider and trend (% change from previous period)
• Identify largest cost contributors by service category (compute, storage, network)

1. Quick cross-cloud waste scan

• Duplicate resources: Same workload/data running on multiple clouds (accidental redundancy)
• Unused cross-cloud connectivity: VPN tunnels, Direct Connect/ExpressRoute/Interconnect with zero traffic (last 30 days)
• Orphaned cross-cloud resources: Load balancers, NAT gateways pointing to deleted resources
• Commitment under-utilization: RIs/SPs/CUDs with <70% utilization across all clouds

1. Cross-cloud price comparison (same workload on different clouds)

• Identify 5 largest compute workloads
• Calculate equivalent cost on each cloud (normalize instance types: AWS m5.xlarge ≈ GCP n2-standard-4 ≈ Azure D4s v3)
• Flag workloads with >20% cost differential for placement optimization

1. Output quick wins (3 highest impact items)

• Example: "Migrate analytics workload from AWS Redshift to GCP BigQuery → save $3,200/month (55% reduction)"
• Example: "Delete 6 unused AWS Direct Connect + Azure ExpressRoute connections → save $1,800/month"
• Example: "Rebalance commitments: reduce AWS RI, increase GCP CUD → save $2,400/month"

Token budget checkpoint: ~1.8k tokens for API calls, normalization, analysis, output formatting.

Tier 2 (≤6k tokens): Comprehensive Multi-Cloud Cost Optimization

Goal: Generate detailed cross-cloud optimization plan with quantified savings and migration recommendations.

Extends T1 with:

1. Cross-cloud workload placement analysis

• Fetch detailed resource inventory (compute, database, storage) from all clouds
• Calculate unit economics per cloud (cost per vCPU-hour, cost per GB storage, cost per 1M requests)
• Identify migration candidates (workloads without hard dependencies on current cloud):
• No compliance restrictions (data residency, FedRAMP, etc.)
• No vendor-specific services (avoid migrating from Aurora/BigQuery/Cosmos DB)
• Latency tolerance >50ms (can tolerate cross-region placement)
• Calculate migration cost vs savings ROI:
• Migration cost: data transfer (egress) + downtime + testing
• Annual savings: (current_cloud_cost - target_cloud_cost) × 12
• ROI = annual_savings / migration_cost (recommend if ROI >3x)

Example calculation (accessed 2025-10-26T14:30:00-04:00): Workload: 500TB PostgreSQL database + 50 vCPU app tier Current: AWS RDS Aurora PostgreSQL $12,000/month, EC2 m5.4xlarge reserved $1,500/month Target: GCP Cloud SQL PostgreSQL $7,200/month, n2-standard-16 CUD $900/month Monthly savings: $5,400/month Migration cost: 500TB egress ($45,000) + 2 weeks downtime ($10,000) = $55,000 Annual savings: $64,800 ROI: $64,800 / $55,000 = 1.18x → recommend if strategic, defer if purely financial

1. Commitment optimization across clouds

• Analyze commitment coverage across all providers:
• AWS: Reserved Instances + Compute/EC2 Savings Plans coverage
• GCP: Committed Use Discounts (resource-based and spend-based)
• Azure: Reserved VM Instances + Azure Hybrid Benefit
• Calculate blended commitment rate (weighted average discount across clouds)
• Identify under-committed clouds (on-demand spend >50%) and over-committed clouds (RI/CUD utilization <80%)
• Recommend commitment rebalancing:
• Reduce commitments on expensive/declining clouds
• Increase commitments on cost-effective/growing clouds
• Target: 70-85% commitment coverage across all clouds (sweet spot)

Sources (accessed 2025-10-26T14:30:00-04:00):

• AWS Savings Plans: https://aws.amazon.com/savingsplans/ (up to 72% savings)
• GCP Committed Use Discounts: https://cloud.google.com/compute/docs/instances/committed-use-discounts-overview (up to 70% savings)
• Azure Reserved Instances: https://learn.microsoft.com/azure/cost-management-billing/reservations/ (up to 72% savings)

1. Egress and data transfer cost optimization

• Map cross-cloud data flows (AWS → GCP, Azure → AWS, etc.)
• Calculate egress costs by route:
• Same-region cross-cloud: typically highest ($0.08-0.12/GB)
• Cross-region same-cloud: medium ($0.01-0.02/GB)
• Cloud → internet → cloud (via CDN): varies
• Recommend egress reduction strategies:
• Colocation: Place communicating services in same cloud
• Caching: Use CloudFront/Cloud CDN/Azure CDN to reduce origin fetches
• Compression: Enable gzip/brotli for API responses
• Direct peering: Use AWS Direct Connect + GCP Interconnect partner connections (not public internet)

Egress cost examples (accessed 2025-10-26T14:30:00-04:00):

• AWS to internet: $0.09/GB first 10TB, $0.085/GB next 40TB
• GCP to internet: $0.12/GB first 1TB, $0.11/GB next 9TB
• Azure to internet: $0.087/GB first 5GB

1. Cross-cloud tagging compliance and cost allocation

• Audit tagging across all clouds using unified tag schema (environment, team, cost-center, project)
• Calculate tag compliance rate per cloud (% resources with required tags)
• Identify untagged cost allocation gaps (spend that cannot be attributed to teams/projects)
• Recommend standardized tagging policy across AWS/GCP/Azure (harmonize tag keys)

1. Multi-cloud FinOps maturity assessment

• Evaluate FinOps maturity across dimensions:
• Visibility: Single dashboard for all clouds vs siloed per-cloud tools
• Optimization: Automated vs manual optimization across clouds
• Governance: Unified policies vs per-cloud inconsistency
• Culture: Cross-functional FinOps team vs isolated cloud admins
• Assign maturity score: Crawl (0-3), Walk (4-6), Run (7-10)
• Recommend next steps to improve maturity (e.g., "Implement unified tagging → +2 maturity points")

1. Generate comprehensive report

• Executive summary: Total multi-cloud spend, waste identified, savings potential
• Cost breakdown by cloud: AWS $X, GCP $Y, Azure $Z with trends
• Cross-cloud opportunities: Workload placement (top 10), commitment rebalancing, egress optimization
• Action plan: Prioritized by ROI (savings/effort) with owner assignments

Authority sources (accessed 2025-10-26T14:30:00-04:00):

• Multi-Cloud FinOps Best Practices: https://www.prosperops.com/blog/multi-cloud-cost-management-guide/
• FinOps Framework Principles: https://www.finops.org/framework/principles/
• Cloud Cost Optimization Statistics: Organizations waste 32% of cloud spend; multi-cloud tools achieve 35-68% cost reductions (CloudZero 2025)

Output: JSON report with sections: unified_cost_summary, cross_cloud_waste (T1), workload_placement_recommendations, commitment_balance_plan, egress_optimization, tagging_compliance, finops_maturity_score, prioritized_action_plan.

Token budget checkpoint: ~5.5k tokens (includes T1 + extended multi-cloud analysis + detailed outputs).

T3: Enterprise Multi-Cloud Optimization (≤12k tokens)

Goal: Deep financial modeling, predictive forecasting, and custom multi-cloud optimization strategies for >$1M annual spend.

Extends T2 with:

1. Predictive cost forecasting

• Machine learning models trained on historical spend patterns (6+ months data)
• Forecast next 12 months spend by cloud, service, and team
• Identify seasonal patterns (e.g., Q4 spike, weekend drop-off)
• Alert on forecast anomalies (>15% deviation from expected)

1. Custom commitment optimization algorithms

• Optimize commitment portfolio across clouds using linear programming
• Constraints: budget limits, risk tolerance, workload volatility
• Objective function: maximize total discount percentage across all clouds
• Account for commitment term trade-offs (1-year flexibility vs 3-year deeper discounts)

1. Multi-cloud vendor negotiation intelligence

• Aggregate total spend across clouds to strengthen negotiation position
• Benchmark against similar-sized organizations (anonymized peer data)
• Identify Private Pricing Agreement (PPA) opportunities with AWS/GCP/Azure
• Calculate Enterprise Discount Program (EDP) eligibility and potential savings

1. Sustainability and carbon cost optimization

• Map cloud regions to carbon intensity (gCO2/kWh)
• Calculate carbon footprint by cloud and workload
• Recommend low-carbon region placement (GCP Iowa vs AWS Virginia)
• Integrate carbon costs into TCO (emerging regulatory requirement)

1. Multi-account/multi-org consolidation

• AWS: Consolidate billing across AWS Organizations (50+ accounts)
• GCP: Aggregate billing across multiple billing accounts
• Azure: Unified cost view across subscriptions and management groups
• Enable volume discounts and cross-account commitment sharing

Authority sources (accessed 2025-10-26T14:30:00-04:00):

• FinOps Market Growth: $5.5B in 2025, 34.8% CAGR (Holori 2025)
• Cloud Computing Market: $723.4B in 2025, 21.5% YoY growth
• AWS Enterprise Discount Programs: https://aws.amazon.com/pricing/
• GCP Committed Use Discount strategies: https://cloud.google.com/docs/cuds-recommendations

Output: Full enterprise-grade multi-cloud financial optimization plan including forecasts, custom commitment strategies, vendor negotiation playbook, sustainability metrics, and multi-account consolidation roadmap.

Token budget checkpoint: ~11k tokens (includes T1 + T2 + enterprise-grade analysis).

Decision Rules

When to abort:

• Billing API access fails for any cloud → insufficient permissions; emit setup instructions per cloud
• Cost data <30 days → insufficient for trend analysis; wait for more data
• Migration restrictions block all workload placement → report "no cross-cloud opportunities"

Ambiguity thresholds:

• Workload placement confidence: Only recommend migration if:
• Cost differential >20% AND annual savings >$10k (avoid noise)
• No hard compliance/latency constraints
• ROI >2x (conservative threshold; adjust to 3x for risk-averse orgs)
• Commitment rebalancing: Recommend only if:
• Current utilization <80% (under-utilized) OR coverage <60% (under-committed)
• Rebalance would improve blended discount rate by ≥5 percentage points
• Egress optimization: Flag only if egress costs >10% of total spend OR >$5k/month absolute

Prioritization logic:

1. ROI-based ranking: (annual_savings / implementation_effort_cost) descending

• Effort scale: Low (delete unused) < Medium (commitment rebalance) < High (workload migration)

1. Quick wins first: Zero-downtime, zero-risk changes (delete unused cross-cloud connections) rank highest
2. Strategic alignment: If business strategy favors specific cloud (e.g., AWS for ML), deprioritize migration away from it

FinOps principle application (accessed 2025-10-26T14:30:00-04:00):

Per FinOps Foundation principles (https://www.finops.org/framework/principles/):

• "Teams collaborate": Multi-cloud optimization requires cross-team coordination (cloud admins, finance, engineering)
• "Decisions are data-driven": All recommendations backed by normalized cost data across clouds
• "Take advantage of variable cost model": Leverage spot instances, preemptible VMs, and commitment flexibility across clouds

Output Contract

Schema (JSON):

json
{
  "unified_cost_report": {
    "period": "2025-09-26 to 2025-10-26",
    "total_spend": 245000.00,
    "breakdown_by_cloud": {
      "aws": {"spend": 125000.00, "percentage": 51.0, "trend": "+5%"},
      "gcp": {"spend": 80000.00, "percentage": 32.7, "trend": "-2%"},
      "azure": {"spend": 40000.00, "percentage": 16.3, "trend": "+8%"}
    },
    "waste_identified": 68000.00,
    "savings_potential": {
      "monthly": 52000.00,
      "annual": 624000.00,
      "percentage": 21.2
    }
  },
  "workload_placement_recommendations": [
    {
      "workload_id": "analytics-cluster-01",
      "current_cloud": "aws",
      "current_cost_monthly": 12000.00,
      "recommended_cloud": "gcp",
      "recommended_cost_monthly": 6800.00,
      "monthly_savings": 5200.00,
      "annual_savings": 62400.00,
      "migration_cost": 55000.00,
      "roi": 1.13,
      "rationale": "BigQuery vs Redshift cost advantage for analytics workload"
    }
  ],
  "commitment_balance_plan": {
    "current_coverage_rate": 58.0,
    "target_coverage_rate": 75.0,
    "current_blended_discount": 28.0,
    "target_blended_discount": 42.0,
    "recommendations": [
      {
        "cloud": "aws",
        "action": "reduce",
        "current_commitment_monthly": 60000.00,
        "recommended_commitment_monthly": 48000.00,
        "rationale": "RI utilization at 68%, under-utilized"
      },
      {
        "cloud": "gcp",
        "action": "increase",
        "current_commitment_monthly": 15000.00,
        "recommended_commitment_monthly": 32000.00,
        "rationale": "On-demand spend at 72%, opportunity for 70% CUD savings"
      }
    ]
  },
  "cross_cloud_waste_inventory": [
    {
      "waste_type": "unused_cross_cloud_vpn",
      "resources": [
        {"provider": "aws", "resource_id": "vpn-0a1b2c3d", "idle_days": 60},
        {"provider": "azure", "resource_id": "vpn-xyz789", "idle_days": 60}
      ],
      "monthly_cost": 1800.00
    },
    {
      "waste_type": "duplicate_backup_storage",
      "resources": [
        {"provider": "aws", "resource_id": "s3://backups-prod", "size_tb": 50},
        {"provider": "gcp", "resource_id": "gs://backups-prod", "size_tb": 50}
      ],
      "monthly_cost": 2300.00
    }
  ],
  "action_plan": [
    {
      "priority": 1,
      "action": "Delete unused cross-cloud VPN connections",
      "impact": "medium",
      "effort": "low",
      "monthly_savings": 1800.00,
      "owner": "cloud-networking-team"
    },
    {
      "priority": 2,
      "action": "Rebalance commitments (reduce AWS RI, increase GCP CUD)",
      "impact": "high",
      "effort": "medium",
      "monthly_savings": 8400.00,
      "owner": "finops-team"
    }
  ]
}

Required fields: unified_cost_report (with breakdown_by_cloud, savings_potential), action_plan (prioritized).

Optional fields: workload_placement_recommendations, commitment_balance_plan (only if applicable based on business_constraints).

Examples

yaml
# Multi-cloud: AWS $125k/mo, GCP $80k/mo, Azure $40k/mo
input: {scope: all, time_range: 90d, model: chargeback}

output:
  total_spend: $245k, waste: $68k (28%), savings: $52k/mo
  workload_placement:
    - analytics: AWS Redshift $12k → GCP BigQuery $6.8k (save $5.2k/mo)
  cross_cloud_waste:
    - unused VPN (AWS+Azure): $1.8k/mo
    - duplicate backups (AWS+GCP): $2.3k/mo
  commitment_rebalance:
    AWS RI: $60k → $48k/mo (reduce)
    GCP CUD: $15k → $32k/mo (increase)
  action_plan:
    1. Delete unused VPN (LOW effort) → $1.8k/mo
    2. Consolidate backups (LOW effort) → $2.3k/mo
    3. Rebalance commitments (MED effort) → $8.4k/mo
    4. Migrate analytics (HIGH effort, ROI 1.13x) → $5.2k/mo

Quality Gates

Token budgets (enforced):

• T1: ≤2,000 tokens - quick multi-cloud health check with top 3 cross-cloud opportunities
• T2: ≤6,000 tokens - comprehensive multi-cloud optimization with workload placement, commitment rebalancing, egress optimization, and unified FinOps analytics
• T3: ≤12,000 tokens - enterprise-grade optimization with ML forecasting, custom commitment algorithms, vendor negotiation intelligence, sustainability metrics

Accuracy requirements:

• Cost normalization must account for currency (USD/EUR/GBP) and time period differences
• Cross-cloud price comparisons validated against official pricing APIs (accessed on NOW_ET)
• Workload placement ROI calculations include migration costs (egress, downtime, testing)

Safety constraints:

• No automatic workload migration: All cross-cloud moves require manual approval and testing
• Compliance checks: Flag workloads with data residency/sovereignty requirements before recommending migration
• Commitment purchase limits: Never recommend commitments exceeding 85% coverage (maintain flexibility)

Auditability:

• Cite pricing source for all cost calculations (AWS Pricing API, GCP Cloud Billing, Azure Rate Card)
• Document assumptions in workload placement (instance type equivalence, network latency tolerance)
• Record baseline metrics for each cloud at analysis time

Determinism:

• Same inputs + same cost data → same recommendations
• Configurable thresholds (ROI minimum, egress cost %, commitment coverage targets)

Resources

Official cloud provider documentation:

• AWS Cost Management: https://aws.amazon.com/aws-cost-management/
• AWS Savings Plans: https://aws.amazon.com/savingsplans/
• GCP Cloud Billing: https://cloud.google.com/billing/docs
• GCP Committed Use Discounts: https://cloud.google.com/compute/docs/instances/committed-use-discounts-overview
• Azure Cost Management: https://learn.microsoft.com/azure/cost-management-billing/
• Azure Reserved Instances: https://learn.microsoft.com/azure/cost-management-billing/reservations/

FinOps Foundation resources:

• FinOps Framework: https://www.finops.org/framework/
• FinOps Principles: https://www.finops.org/framework/principles/
• Multi-Cloud FinOps Guidance: https://www.finops.org/framework/capabilities/

Multi-cloud cost optimization guides:

• Multi-Cloud Cost Management Best Practices 2025: https://www.prosperops.com/blog/multi-cloud-cost-management-guide/ (accessed 2025-10-26T14:30:00-04:00)
• FinOps Best Practices 2025: https://www.cloudzero.com/blog/finops-best-practices/ (accessed 2025-10-26T14:30:00-04:00)
• Top 50 FinOps Tools 2025: https://holori.com/20-best-finops-and-cloud-cost-management-tools-in-2025/ (accessed 2025-10-26T14:30:00-04:00)
• Cloud Pricing Comparison 2025: https://cast.ai/blog/cloud-pricing-comparison/ (accessed 2025-10-26T14:30:00-04:00)

Related skills:

• finops-cost-analyzer: For single-cloud cost optimization (invoke before multi-cloud aggregation)
• cloud-multicloud-advisor: For strategic multi-cloud architecture design (invoke before deployment)
• cloud-provider-advisor: For initial cloud provider selection (invoke during planning phase)