AI Spend ROI & Budget Governance

Crestline AI · Q3 2024 · Updated 2 hours ago

AI Efficiency Ratio
4.2x
-0.9 vs Target
QTD AI Spend
$1.84M
$310K over projected
Budget Burn Rate
113%
vs Q cap
Experiment Spend %
31.4%
+9.4% over limit
Idle Compute Cost
$127K
MTD wasted GPU hrs
Utilization68%
Avg ROI Payback
8.2mo
-1.3mo vs Q2

Monthly AI Spend by Category

Last 6 Months

Workload Portfolio: Spend vs ROI

Bubble size = monthly spend. Target ROI ≥4.5x (green zone).

Team Budget Burn Rate vs Quarterly Cap

Red = over cap; teams ranked by % over budget.

Spend Trend by Category (90d)

Optimization Savings Tracker

MTD savings: $58K from caching, $31K from prompt compression.

ROI Initiative Payback Timeline

Green = already profitable; Orange = in payback period; Red = not yet launched.

AI Workload Cost Attribution

Workload Team Type MTD Spend Cost/Unit ROI % Budget
Doc Intelligence APIData ScienceProd$284K$0.0426.8x15.4%
Search Embedding GenPlatformProd$241K$0.0185.2x13.1%
Customer Support BotCX OpsProd$198K$0.314.9x10.8%
Code Gen AssistantEngineeringProd$176K$0.143.8x9.6%
LLM Fine-tuning ExpData ScienceExp$162K—TBD8.8%
Sentiment AnalysisProductProd$134K$0.0225.6x7.3%
Image ClassificationData ScienceProd$119K$0.0894.1x6.5%
Multimodal Exp SuiteResearchExp$107K—TBD5.8%
Report SummarizationAnalyticsProd$94K$0.383.4x5.1%
RAG Pipeline R&DData ScienceExp$81K—TBD4.4%

Experiment vs Production Spend

Target: ≤28% experiment spend. Current: 31.4% (over limit).

Forecast vs Actual Spend

Month 5 spike: $94K unattributed LLM experiments (Data Science).