3-Year Cost Breakdown
| Platform | Compute | Storage | ML / AI | Pipelines | Year 1 | Year 2 | Year 3 | 3-Year Total |
|---|
3-Year Total Comparison
Key Assumptions & Unit Prices
⚠ Illustrative ROM — based on public H1 2026 list pricing. Actual costs vary by region, negotiated discounts, and committed use. Data egress costs excluded.
Base unit prices (before profile multipliers)
| Platform | Storage/TB | Compute/user | Pipeline/day | ML model | FTE/mo |
|---|---|---|---|---|---|
| Snowflake | $6.50¹ | $120 | $18 | $380 | $0 |
| Databricks | $20 | $130 | $22 | $95 | $4,167 |
| MS Fabric | $21 | $110² | $20 | $150 | $4,167 |
| AWS Native | $18 | $95 | $65³ | $220 | $8,333 |
| OSS on K8s | $15 | $60 | $28 | $75 | $25,000 |
Platform licenses (flat/mo): Snowflake $2k · Databricks $1.5k · Fabric $1.8k · AWS $0 · OSS $0
¹ Snowflake compresses ~3.5× ($23 list ÷ 3.5)
² Fabric adds capacity floor: +$800/mo at ≥50 users, +$2,400/mo at ≥150 users
³ AWS Glue DPU-hour model — 0.25 hr/run × $0.44/DPU
Snowflake concurrency multipliers (active users)
| Users | Multiplier | Reason |
|---|---|---|
| < 50 | 1.0× | Single warehouse |
| 50–99 | 1.3× | Concurrency scaling begins |
| 100–199 | 1.7× | Multi-cluster active |
| 200+ | 2.2× | Full multi-cluster |
Workload profile ML multipliers
| Profile | Snowflake | Databricks | Fabric | AWS | OSS |
|---|---|---|---|---|---|
| Mixed | 1.0× | 1.0× | 1.0× | 1.0× | 1.0× |
| SQL Analytics | 2.0× | 0.8× | 1.0× | 1.5× | 0.5× |
| ML-First | 3.5× | 0.6× | 1.2× | 2.5× | 0.4× |
| Real-Time | 1.0× | 1.0× | 1.0× | 1.0× | 1.0× |
FTE assumption: $100k/year loaded cost ($8,333/mo). OSS requires 3 platform engineers. Annual growth compounds variable costs only — FTE and license are fixed.
Estimates are illustrative ROM only — based on public H1 2026 list pricing. Actual costs vary by region, committed use discounts, and negotiated agreements.