This report provides a comprehensive analysis of deployed and installed Meta Platforms' Meta Training and Inference Accelerator (MTIA) semiconductor ASICs for the v1 Freya, v2 Artemis, and v3 Iris as well as insights into the future v4, v5, and v6 ASICs due out in 2026.
Features
ASIC Deployments
- By geographic region
- North America
- Europe
- Asia Pacific
- By Country in each geographic region (9 in total)
- By City/State within each geographic region where applicable
- By ASIC Product Type and by geographic region
- MTIA v1, v2, v3
MTIA Server Deployments
- By Server Product Type and by geographic region
- Kings Canyon Gen2
- Grand Teton - Artemis
- Santa Barbara Iris 1U
This product will be delivered within 2 business days.
Features
ASIC Deployments
- By geographic region
- North America
- Europe
- Asia Pacific
- By Country in each geographic region (9 in total)
- By City/State within each geographic region where applicable
- By ASIC Product Type and by geographic region
- MTIA v1, v2, v3
MTIA Server Deployments
- By Server Product Type and by geographic region
- Kings Canyon Gen2
- Grand Teton - Artemis
- Santa Barbara Iris 1U
This product will be delivered within 2 business days.
Table of Contents
- RESEARCH BRIEF
- Methodology
- Background
- EXECUTIVE SUMMARY
- Conclusion
- META TRAINING INFERENCE ACCELERATOR (MTIA) PROCESSOR ANALYSIS
- V3 Iris Deployed, Waiting for V4/5/6
- META MTIA SERVER CHASSIS CONFIGURATION
- Yosemite V3 Server Chassis for MTIA v1 Freya ASICs
- Grand Teton - Artemis Server Chassis for MTIA v2/2i Artemis ASICs
- Santa Barbara AI Training Server Chassis for MTIA v3 Iris ASICs
- Meta Platforms Server Deployment Analysis
- Table 1: Meta Platforms MTIA Processor Specifications by Generation
- Table 2: Meta MTIA Processor Deployments by Country/Generation for North America
- Table 3: Meta MTIA Processor Deployments by Country/Generation for Europe
- Table 4: Meta MTIA Processor Deployments by Country/Generation for Asia Pacific
- Table 5: MTIA Processor Deployed in United States by Data Center and Generation
- Table 6: Meta Platforms MTIA Server Types (Units)
- Exhibit 1: Global Meta Platform MTIA Processors Installed by Generation
- Exhibit 2: Global Meta Platform MTIA Processors Installed by Geographic Region
- Exhibit 3: Meta Platforms MTIA v1 and v2 Package Type
- Exhibit 4: Meta Platforms MTIA v1 Floorplan (Left) and Die Image (Right)
- Exhibit 5: Meta Platforms MTIA v2 Floorplan (Left) and Die Image (Right)
- Exhibit 6: Global Meta MTIA Processors Installed by Generation
- Exhibit 7: Global Meta MTIA Processors Installed by Geographic Region
- Exhibit 8: Global Meta MTIA Processors Installed by Country
- Exhibit 9: Meta Platforms Installed MTIA Processors by U.S. State
- Exhibit 10: Meta MTIA v1 Inference 2OU M.2 Card
- Exhibit 11: Meta Glacier Point V3 2OU M.2 Inference Accelerator Module
- Exhibit 12: Meta 2x 2OU (4OU M.2) Accelerator Sled Configuration
- Exhibit 13: Meta Kings Canyon Gen2 Inference Server Sled
- Exhibit 14: Meta Yosemite V3 Chassis
- Exhibit 15: Meta MTIA v2 Artemis Inference Accelerator Module without Heatsink
- Exhibit 16: Meta MTIA v2 Artemis Inference Accelerator Module with Heatsink
- Exhibit 17: Meta Colter Bay Accelerator Carrier Tray
- Exhibit 18: Meta Grand Teton - Artemis Server Chassis Configuration
- Exhibit 19: Meta Santa Barbara Iris Server Tray
- Exhibit 20: Meta MGX-OCP AI Training Double-Wide Rack
- Exhibit 21: Meta Platforms MTIA Servers by Type (Units)
