Microsoft’s 3nm AI inference accelerator with FP4/FP8 tensor cores, 216GB HBM3e at 7 TB/s. The XPU market reached $31B in 2025 according to Futurum’s research.

Microsoft’s 3nm AI inference accelerator with FP4/FP8 tensor cores, 216GB HBM3e at 7 TB/s. The XPU market reached $31B in 2025 according to Futurum’s research.

About the Author
Research Director, Semiconductors
Leads Futurum's Semiconductors Practice covering edge AI, generative AI, and hyperscale silicon architectures.