Tech Disruptors Solidigm on Satiating AI’s Data-Storage Needs
6 snips
Nov 19, 2025 Greg Matson, Senior Vice President at Solidigm, brings 25 years of expertise in memory and storage to the conversation. He discusses how Solidigm is addressing the skyrocketing data demands of AI, with insights on the pivotal shift towards high-capacity SSDs. Greg explains the significant difference between SSD needs for inference and training, while predicting a future with one-petabyte drives. He emphasizes the importance of liquid-cooled SSDs and customer collaboration in staying competitive as demand for flash storage surges.
AI Snips
Chapters
Transcript
Episode notes
From Intel Roots To Liquid-Cooled SSDs
- Solidigm adapted SK Hynix charge-trap tech and added firmware and thermal innovations.
- The company launched the industry's first liquid-cooled SSD as a customer-driven system solution.
SSDs Form Two Critical AI Storage Tiers
- SSDs sit both inside GPU servers and as larger network-attached tiers to feed GPUs.
- As inference context sizes and checkpointing grow, on-box SSD capacity and performance must scale rapidly.
Multimodal Output Explodes Storage Needs
- Multimodal and video outputs are orders of magnitude larger than text and will drive massive fast-storage demand.
- Greg Matson expects video and multimodal to be a strong long-term storage catalyst.