Paragraf Expands into Memory Market

Moving into memory matters because it shifts Paragraf from selling specialty parts into one of the biggest and most strategic chip categories in computing. Sensors are usually attached to a specific instrument or battery pack. Memory sits inside almost every server, phone, AI device, and edge system. If Paragraf can make graphene memory with much lower power draw, it is no longer limited to niche sensing workloads, it can sell into the core hardware budget of data centers and intelligent devices.

• The memory effort is not just exploratory lab work. In February 2025, Paragraf and the University of Cambridge received Innovate UK funding to build a proof of concept memory device that combines ferroelectric materials with graphene on silicon, with Paragraf saying the goal is around 10x lower power than existing memory technology.
  2 paragraf
• This is a different market structure from sensors. Memory is a huge volume semiconductor category dominated by a few large players and sold into standard compute platforms, so success would give Paragraf access to far larger unit volumes than quantum sensors or diagnostic chips, even if it begins in narrow low power niches first.
  1 sacra 3 marketsandata
• Paragraf already has the manufacturing logic for this move. Its core capability is growing graphene directly on silicon wafers with standard semiconductor processes, and that same platform has already been extended from magnetic sensors into molecular sensing through the 2023 Cardea Bio acquisition, showing a pattern of reusing one materials stack across multiple device categories.
  1 sacra 4 paragraf 5 sacra

The next step is a wedge into workloads where electricity and heat matter more than lowest chip cost. If graphene memory proves meaningfully more efficient, Paragraf can start in edge AI and power constrained systems, then move upstream into larger data center memory budgets. That would turn graphene from a specialty sensor material into a broader semiconductor platform.