@Solidigm, @MinIO, & @Intel partnered to design and benchmark S3-compatible object storage, with a focus on performance and concurrency scaling. This white paper goes deeper into what this benchmark means for AI builders. Check out the full results! solidigm.com/products/techno…
RAG pipelines don't have to be DRAM-constrained. SSD-offloaded vector retrieval delivered 65% lower memory footprint and 14% higher query throughput vs DRAM-bound alternatives while maintaining 99% recall. Storage is earning its place as an active AI memory tier. #AIIFD5
Well, what if your system could remember instead of recompute? With KV cache offload to NVMe, @Solidigm and @MetrumAI cut context switching from 39 seconds to 1.43 seconds! #AIIFD5#AIInfrastructure#SSD
As context windows grow, GPU memory can't keep up. Long-context AI inference forces systems to recompute the same work over and over, burning cycles and slowing response times. At scale, that's not sustainable. #AIIFD5#Inference#GenerativeAI
Forget bolting KV cache onto enterprise storage. MemKV strips it down — KV-addressed, RDMA-native, NVMe-backed. Purpose-built for the way AI inference actually works. #AIIFD5#GenerativeAI#SSD
The solution? @MinIO MemKV is built like memory, not enterprise storage. GPU registered memory raw NVMe blocks RDMA. No metadata database. No overhead. Just fast, direct KV access via io_uring. #AIIFD5#ObjectStorage#MinIO
The problem? At just 8 concurrent 32K sessions, KV cache requirements blow past GPU HBM capacity. At 32 concurrent sessions? You're at 4.3x the HBM ceiling. The memory wall is real, and it's happening now. techfieldday.com/appearance/…#AIInfrastructure#Inference#AIIFD5
Modern object storage and high-density SSDs are helping organizations build scalable, efficient AI factories. Looking forward to sharing insights alongside the @MinIO team! techfieldday.com/appearance/…
From feeding GPUs during training to supporting RAG pipelines, vector databases, and long-context inference, the ability to move and manage data efficiently is increasingly the key to unlocking AI performance. #RAG#Storage#AIIFD5techfieldday.com/appearance/…
Great conversations happen when infrastructure leaders come together to solve the challenges of scaling AI. Excited to see @Solidigm and @MinIO on stage together at AI Infrastructure Field Day. #AIIFD5techfieldday.com/appearance/…
This week, we invite you to join @Solidigm & @MinIO as we present “Density Without Compromise: How Storage & Memory Became the Defining Problem in Modern #AI Inference” at #AIIFD5 with @TechFieldDay! Tune in LIVE June 11 at 10:30am PT for the full session: techfieldday.com/appearance/…
We are pleased to announce Richard Chin as Co-CEO of Solidigm, joining Xin Guo to help lead the company’s next phase of technology development for AI infrastructure leadership. Read the full announcement: news.solidigm.com/en-WW/2661…
With growing RAG pipelines and KV caches, modern storage has evolved from helpful to necessary. Join Avi Shetty, Vice President of AI Ecosystem and Market Enablement at @Solidigm as he explores this topic at Computex Forum 2026 in Taipei, Taiwan on June 4! events.computextaipei.com.tw…
In another, Submer's Gabriel Lazar discusses data center sustainability efforts, heat reuse, infrastructure efficiency, and why the industry needs to look beyond PUE and toward IT utilization and efficiency as AI infrastructure continues to scale. (3/3) techarena.ai/content/submers…
In one episode, @Scaleway's Albane Bruyas shares what it takes to build a sovereign European cloud infrastructure, the company’s collaboration with OCP & how AI infrastructure is evolving to support longer-context & agentic AI workloads. (2/3) techarena.ai/content/scalewa…
At the OCP EMEA Summit in Barcelona, @Solidigm’s Ace Stryker joined TechArena.ai's Allyson Klein for another round of Data Insights conversations spanning sovereign cloud, sustainability, liquid cooling, and the infrastructure demands of AI at scale. (1/3)