Microsoft Sustainability Profile
Environmental impact data for 4 tracked Microsoft models.
Sustainability Profile
Renewable Energy
34 GW contracted
Emissions Trend
+25% in FY2025 (total, Scopes 1-3)
Water Trend
WUE improved 39%
Transparency Rating
High
Carbon negative by 2030 goal. FY2025 rise driven by datacenter expansion and pausing unbundled renewable certificates.
How Microsoft Compares
Microsoft's average metrics compared to the full dataset of 199 models
Energy / query
CO2 / query
Water / query
Environmental Overview
Microsoft offers 4 models in our database, focusing on text / chat and multimodal. Their models are notably efficient, averaging 0.11 Wh per query — less than half the dataset average of 49.0 Wh. Their lineup spans from Phi-4 Mini (0.04 Wh) to Phi-4 Reasoning (0.27 Wh), showing meaningful variation in energy needs. All of their tracked models have open-source weights, enabling community-driven optimization and independent efficiency benchmarking. Microsoft demonstrates strong environmental transparency, publishing detailed sustainability metrics. Carbon negative by 2030 goal. FY2025 rise driven by datacenter expansion and pausing unbundled renewable certificates.
All Microsoft Models
Text / Chat3
Multimodal1
Frequently Asked Questions
How many AI models does Microsoft have?
Microsoft has 4 AI models tracked in the Know Your Compute database, spanning 2 categories: Text / Chat, Multimodal.
What is Microsoft's most energy-efficient model?
Microsoft's most energy-efficient model is Phi-4 Mini, consuming 0.04 Wh per query. Their least efficient model is Phi-4 Reasoning at 0.27 Wh per query.
Does Microsoft use renewable energy?
Microsoft's renewable energy claim: 34 GW contracted. Carbon negative by 2030 goal. FY2025 rise driven by datacenter expansion and pausing unbundled renewable certificates.
Calculate your Microsoft footprint
Use our calculator to estimate your personal environmental impact from Microsoft models.
Calculate My Compute