French startup Antaios raises $11M for developing SOT-MRAM that can potentially replace embedded non-volatile memory and SRAM cache in microprocessors and SoCs
Antaios SA (Grenoble, France), a 2017 startup working on spin-orbit-torque (SOT) MRAM has secured US$11 million in funding.
Context & Ripple Effects
Antaios is the latest entrant in a decade-long effort to displace the workhorse memories inside chips: back in 2016, Nantero raised $21M for its carbon-nanotube NRAM on essentially the same premise — a new physics layer that could serve as embedded non-volatile memory. The thesis has since sharpened around power, with Israel's RAAAM raising a $17.5M Series A led by NXP for GCRAM, pitched at up to 10x power savings versus high-density SRAM.
What makes Antaios notable is geography and scope: a Grenoble startup claiming SOT-MRAM can address both embedded NVM and SRAM cache — a bigger slice of the memory hierarchy than most rivals target — in a French silicon ecosystem where SiPearl's €130M-backed Rhea1 ARM AI chip shows there is domestic SoC demand for advanced on-chip memory options.
First-order effects
- Antaios gains an $11M runway to move SOT-MRAM from lab toward foundry-grade integration, aimed directly at microprocessor and SoC designers now locked into SRAM caches and existing embedded NVM.
Second-order effects
- Antaios now competes for the same design-in slots as adjacent-technology rivals like RAAAM — whose round NXP led, signaling that IDMs would rather fund multiple hedges than bet on one replacement memory.
Third-order effects
- If any of these challengers clears foundry qualification, the boundary between non-volatile storage and cache collapses into one technology choice per SoC — making qualification itself, not raw cell performance, the deciding moat.
The trend: Emerging-memory startups keep raising against SRAM and embedded NVM replacement, and the funding cadence — Nantero in 2016, RAAAM and Antaios since — suggests buyers are finally willing to bank multiple alternatives at once.