NVMe and CXL storage hardware for AI and data center scale
ScaleFlux designs computational storage hardware—NVMe drives and CXL memory subsystems—using a deep embedded-systems stack (Synopsys, Cadence, ARM, SystemVerilog, Verilog). The project list (CXL firmware, NAND controller design, chip micro-architecture) and hiring shape (12 engineers, 1 sales role, mostly senior/staff IC) reveal a hardware-first, engineering-driven company focused on silicon and protocol-level optimization rather than software abstraction. Current pain points center on power efficiency and data-pipeline complexity—core constraints in data center hardware.
ScaleFlux builds specialized storage and memory hardware for data centers running AI, HPC, and data-driven workloads. The company designs NVMe SSDs and CXL memory solutions that embed computational capabilities directly into the storage layer, reducing data movement and latency. Founded in 2014 and based in Milpitas, California, the company operates as a privately held hardware innovator with engineering offices in the United States and India. The product line spans enterprise flash storage, computational storage drives, and CXL protocol implementations designed for modern data infrastructure.
ScaleFlux uses industry-standard EDA tools (Synopsys, Cadence, ARM), hardware description languages (SystemVerilog, Verilog), and protocols (NVMe, PCIe, CXL, I2C, UART). Core firmware and tools run on Linux with Python, C/C++, and Perl for automation and validation.
Current projects include CXL firmware development, CXL protocol optimization, NAND memory controller design, chip micro-architecture work, and embedding computational storage into flash drives. Work also focuses on modernizing data pipelines for data center infrastructure.
Other companies in the same industry, closest in size