Nagare Labs Unveils Z2RAM Memory Technology
The Georgia Tech spinout's oxide-semiconductor memory design targets AI chip bottlenecks, backed by a $5 million seed round.
Updated on Nov. 6, 2026 in Semiconductors
Nagare Labs, a Georgia Tech spinout, unveiled Z2RAM technology on October 26, 2026, alongside a $5 million seed round led by Playground Global. The company is developing a memory cell designed to stack directly above AI accelerator logic.
Why it matters
As processor compute speeds outpace memory data delivery, a widening gap known as the memory wall limits AI system performance. Stacking memory directly over logic aims to shorten data fetch paths and address that constraint.
Demonstrations confirmed data retention exceeding 300 seconds at temperatures up to 85 degrees Celsius, while projections call for more than 5x the bit density of 2 nm high-density SRAM and 14x faster access than current DRAM. These figures remain company projections, and the design has not yet been produced at scale.
The players
Nagare Labs
A Georgia Tech spinout developing oxide-semiconductor memory designed for stacking above AI accelerator logic.
Playground Global
A venture capital firm that led the $5 million seed round for Nagare Labs.
Pat Gelsinger
Former chief executive of Intel who serves as a board director of Nagare Labs.
The details
Z2RAM uses a two-transistor memory cell built on oxide semiconductors and has no capacitor, which allows fabrication with standard back-end processes. The design is intended to be stacked directly above an AI accelerator's logic die.
Timeline
2026: Nagare Labs was founded by Georgia Tech professors Suman Datta and Shimeng Yu.
October 26, 2026: Nagare Labs unveiled Z2RAM and announced its $5 million seed round.
Around 2030: Nagare Labs targets working memory-tile samples.
Within a decade: The company targets mass production.
The Tech Race
Z2RAM is measured against 2 nm high-density SRAM and current DRAM, the memory technologies that define the density and speed benchmarks AI chipmakers rely on today. The projected gains would position Nagare Labs against established memory suppliers if the design reaches production.
If Z2RAM reaches production, AI systems could access memory faster and with less energy, which could affect the performance of AI tools people use. Nagare Labs has not announced commercial availability, and working samples are not expected until around 2030.
The takeaway
Stacking memory directly above processing logic is one approach to closing the memory wall that limits AI hardware. Whether Z2RAM's projected density and speed gains hold up in production will determine its relevance to the industry.
Further reading
For more on chip design work emerging from the region, see Semiconductors coverage.
Source note: This article includes information reported by Crypto Briefing.
