Tech news in 3 minutes
TSMC: Fiscal 2Q26 Financial Results
Komprise CEO Kumar Goswami outlines how the unstructured data management company is capitalizing on the memory and flash pricing crisis and the stalled enterprise AI pilot problem, positioning its platform to reduce storage costs and prepare dark data for production AI workloads. Founded in 2014 and based in Campbell, CA, Komprise has grown to nearly 200 employees with hundreds of large enterprise customers, some storing over 50PB of data. The company is growing at more than 2x the typical B2B tech rate and expects to break even by early 2027. Goswami highlights that the global memory shortage and flash pricing volatility are driving strong tailwinds, as Komprise can free up 60-80% of primary flash capacity by intelligently tiering cold data to object storage, delaying expensive storage refreshes. On the AI front, Komprise has released three core capabilities: a built-in scanner for PII/PHI/IP tagging; Komprise AI Preparation and Process Automation (KAPPA) data services using Python scriplets (Kaplets) to extract relevant data from billions of files; and Transparent File Tables (TFT), a living tabular catalog that sends unstructured data metadata to lakehouses like Databricks and Snowflake via Apache Iceberg. One healthcare customer used Komprise to beat clinical AI timelines by 300% with 97% lower cost. Differentiation comes from global visibility across all storage without vendor lock-in, and a focus on reducing data fed to AI by orders of magnitude rather than vectorizing everything. Go-to-market includes strengthened partnerships with Microsoft, Everpure, IBM, NetApp, and AWS, plus a new Flash Stretch assessment for channel partners dealing with storage refresh pain amid the DRAM and flash crisis.