Memory

RAAAM aims to extend Moore’s Law to memory: moving toward 2nm qualification at TSMC

Posted on by Techtime

[In the photo above: RAAAM management team (left to right): Adam Teman, Eli Lizerovitz, Robert Gitterman, Alex Fish, and Eran Rotem. Photo: Omer Cohen]

RAAAM Memory Technologies, an Israeli–Swiss semiconductor startup, is entering a critical stage in the commercialization of its on-chip memory technology, GCRAM, following the completion of a $17.5 million Series A round led by NXP Semiconductors. CEO and co-founder Dr. Robert Gitterman says the new funding will support the qualification of a 256-MB test chip in TSMC’s 2-nanometer process — the final step before mass production.

“Our technology has already been proven on silicon,” Gitterman told Techtime. “We’re now fabricating the 2 nm qualification chip, which will undergo a series of stringent tests. Once we pass qualification, any company designing chips in this process — including Apple, NVIDIA, and others — will be able to integrate our memory as a drop-in SRAM replacement.”

The memory bottleneck

According to Gitterman, roughly 50 percent of every digital chip’s area is devoted to memory, mostly SRAM. While processors continue to scale with each generation, SRAM has reached the physical limits of miniaturization in advanced CMOS nodes below 5 nm. “Moore’s Law has stopped at memory,” he says. “SRAM has become the bottleneck of the AI era. Once it runs out of space, designers have to move to external memories like HBM — which are slower and far more power-hungry.”

The surge in memory demand for AI accelerators, autonomous vehicles, and edge devices has created a pressing need for denser, more efficient on-chip memory. RAAAM’s GCRAM targets exactly that: a seamless SRAM replacement suitable for CPUs, GPUs, and low-power SoCs alike.

Three transistors and a smart refresh

RAAAM’s innovation lies not in exotic materials or transistor geometry, but in circuit-level architecture. Each GCRAM cell uses three transistors instead of six, relying on charge-retention capacitive storage with a background refresh mechanism.
“The real breakthrough is in the refresh logic we developed,” Gitterman explains. “It operates in the background without interfering with system performance. This lets us maintain high speed while solving the yield problems that plague advanced SRAM.”

The company has demonstrated GCRAM across multiple foundry nodes — from 180 nm down to 5 nm FinFET — achieving 2× density and 10× lower power than conventional SRAM, all while remaining fully compatible with standard CMOS flows.

From academic research to commercialization

Founded in 2021 by four researchers — Dr. Robert Gitterman, Prof. Andreas Burg, Prof. Alexander Fish, and Prof. Adam Teman — RAAAM grew out of nearly a decade of collaborative research between Bar-Ilan University and EPFL Switzerland.
“None of us had prior startup experience,” Gitterman recalls. “We had to learn how to build a company from scratch. But the timing was perfect — the industry was hungry for new memory solutions, and our technology was ready.”

He admits the shift from academia to semiconductors was a reality check. “In academia you have time. In this industry, you have to move at the pace of process generations — sometimes every year. If you don’t keep up, you’re out of the game.”

Strategic backing from NXP

NXP, which led the Series A, has been working with RAAAM for several years and views GCRAM as a strategic technology. “RAAAM’s solution directly addresses one of the most critical challenges in advanced chip design,” said Victor Wang, VP of Front-End Innovation at NXP. “We’ve seen its potential firsthand.”

Alongside NXP, RAAAM is also collaborating with a major networking-chip manufacturer and with GlobalFoundries on additional process integrations.

The company currently employs 22 people, operating from Petah Tikva and Lausanne. Gitterman describes the new funding as “the first major step toward full commercialization” and a sign that the semiconductor industry is again open to genuine innovation.
If RAAAM’s qualification succeeds, its memory could soon find its way into the processors powering the next generation of artificial intelligence.

Memory Market “Is currently in a free-fall”

Posted on by Techtime

IC Insights reduced its worldwide IC market growth forecast for 2022 from 11% to 7%. The downgraded expectation is almost entirely due to the collapse of the memory market in the second half of 2022.  In a recent report, the company writes: “It was as though someone flipped a switch to the off position for the memory market beginning in June.”

In early September, Kyung Kye-hyun, Samsung’s co-CEO and head of its semiconductor unit said, “The second half of this year looks bad, and as of now, next year doesn’t really seem to show a clear momentum for much improvement.”

Many memory companies have attributed the swift downturn to a massive inventory adjustment currently underway by their customers, and expect this inventory adjustment period to extend into 2023. For example, when Micron’s fiscal 3Q ended in May, the company gave an early warning of the coming developments in the memory market: It presented 4Q (ending in August) sales guidance of -17%, and later revised this figure to at least a 21% drop in sales.

Western Digital, a major NAND flash memory supplier, commented in its 2Q conference call, that the inventory adjustment currently underway is “definitely very, very sharp in the quarter we are in (3Q22).”  Western Digital’s outlook is for a company-wide sales decline of 18% this quarter.  With hard disk drives (HDDs) making up about half of the company’s sales and expected to show only a modest decline in 3Q, IC Insights believes that its NAND flash business is likely to register a drop of at least 20% this quarter.

Taiwan-based Nanya is a relatively minor player in the global DRAM market, but its monthly sales data provides some insight into how swiftly the DRAM market can shift gears from boom to bust.  The company’s August 2022 DRAM sales were 39% of what they were in August 2021 – and down 53% from March 2022 sales – just five months ago!

With the memory market currently in a free-fall, IC Insights expects foundry giant TSMC to surpass Samsung and take over the top spot in the semiconductor company sales ranking in 3Q22: Intel is expected to move to the third position (beneath Samsung) in the ranking, with 3Q22 sales that are 26% less than TSMC’s.

Weebit Nano raised $4.5 million to commercialize ReRAM technology

Posted on by Techtime

Hod Hasharon (near Tel Aviv)-based Weebit Nano, which is developing a new type of non-volatile ReRAM technology, has raised $4.5 US million through a private allotment of shares on the Australian Securities Exchange (ASX). At the moment, the company is trying to raise an additional $0.5 AUD million through a public offering. Weebit Nano’s CEO, Coby Hanoch, told Techtime that this financial round ,”Will allow us to move towards commercialization, and hopefully, within a year we’ll already be engaging in serious interactions with potential customers.”

According to the report supplied by the company to the ASX, about half of the money raised will be allocated for the development of a dedicated module for embedded systems, the company’s first target market for its ReRAM technology. “Our technology has already been proven and tested by customers. We are now developing a specific module of the memory, in order to make it suitable for the embedded systems market.”

The Best of all Possible Worlds

Approximately 20% of the amount will be allocated for the development of a component called ‘Selector’, which is designed to minimize leakage currents between the memory cells, and about 15% for transferring the technology for production at standard Fab manufacturing facilities. Weebit Nano is developing a new Resistive Random Access Memory (ReRAM), based on the use of materials that change their electrical resistance in response to electrical voltage, thus “remembering” the voltage levels after they are disconnected from the power source.

It combines the non-volatility of flash memory with the fast, low-power, and long life cycle of the volatile DRAM memory technologies. The company estimates that its prototype is 1,000 times faster and uses 1,000 times less power than flash memory, traits which make him perfect candidate for IoT, artificial intelligence, information centers and more.

Lately, Weebit Nano announced first commercial collaborations, both with the Chinese semiconductors companies, XTX and SiEn. Together, they will examine the integration of the Weebit Nano’s memory component into their’ products. “China is the largest chip consumer in the world, and is determined to build an independent semiconductors industry,” said Hanoch.