Efficient Computer and GlobalFoundries announced a partnership on August 29, 2024, to bring a processor based on Efficient’s Fabric architecture to market using GF’s 22FDX platform. Embedded MRAM is a capability of that platform; the announcement does not establish that Efficient’s particular chip has shipped. Efficient targeted customer sampling by summer 2025, but the sources available here do not verify whether that milestone was met.
What the agreement covers
The companies described a planned processor combining Efficient’s Fabric processor architecture with GlobalFoundries’ manufacturing capabilities on 22FDX. The announcement was framed as a processor partnership, not confirmation that a finished MRAM product was already being manufactured or sold. Efficient’s August 29, 2024 announcement says the company was working with select customers and aimed to begin customer sampling by summer 2025.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
1M1-M000127DVA Development Board TUL PYNQ-Z2 Zynq-7000 XC7Z020 PYNQ-Z2 Development Board FPGA | $304.98 | Buy on Amazon |
That target date has passed. The cited material does not establish whether samples were delivered on schedule, whether the chip entered commercial production, or whether customers deployed it. The deal therefore signals a development and manufacturing plan, not a confirmed product launch.
What embedded MRAM means here
MRAM, or magnetoresistive random-access memory, is non-volatile memory: it retains data without continuous power. In this announcement, it is best understood as a memory capability associated with GF’s 22FDX process platform, rather than as proof of a stand-alone memory chip or of the exact memory configuration in Efficient’s processor.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
#1 Best Overall
- 1M1-M000127DVA Development Board TUL PYNQ-Z2 Zynq-7000 XC7Z020 PYNQ-Z2 Development Board FPGA
GlobalFoundries said embedded MRAM on its 22nm FD-SOI 22FDX platform entered production in 2020. GF described it as a possible alternative to embedded NOR flash for applications including IoT, microcontrollers, automotive systems, and edge AI. That platform-level production status does not establish the production status of Efficient’s processor. GF’s 2020 announcement is about the process technology generally.
In a 2016 release, GF said its 22FDX eMRAM could serve both code-storage roles typically associated with flash and working-memory roles associated with SRAM. It also claimed 1,000-times-faster write speeds and 1,000-times-greater endurance than non-volatile-memory offerings available at that time. Those are historical company comparisons, not independent current benchmarks or specifications for Efficient’s chip. GF’s 2016 release provides the original context.
GF announced Auto Grade 1-ready eMRAM availability on an enhanced FDX platform in March 2026. That update indicates continued development of embedded memory across GF’s FDX offerings; it does not verify Efficient’s processor schedule, implementation, or automotive qualification. The 2026 announcement concerns GF’s broader automotive platform.
Which markets the partners named
Efficient identified the Internet of Things, wearable and implantable health devices, space systems, and security and defense as intended markets. It also cited machine-learning-enabled extreme-edge machine vision, continuous audio intelligence, and sensory and signal-intelligence workloads. These are target applications given by the company, not confirmed customer deployments or evidence that the processor is available in those sectors. The partnership announcement describes those aims.
How to interpret the efficiency claims
Efficient claimed “up to 166x” greater energy efficiency than industry-standard embedded CPUs and “up to 99% lower DC power” for its architecture. These are vendor claims, not independent findings, and the headline efficiency figure should not be generalized to other programs or workloads.
The release’s benchmark note specifies a 5×5 convolution and says the low-power SoC and Efficient E0 ran the same C program. It identifies direct lab-bench measurement for those two measurements, while the accelerator comparison uses advertised MLPerf Tiny 1.1 figures. The different measurement bases limit what the figures can show about performance across products or real-world applications. Efficient’s announcement contains the claim and its benchmark qualification.
What the announcement does—and does not—show
- Established: In August 2024, Efficient and GlobalFoundries announced a plan to bring a Fabric-based processor to market on GF’s 22FDX platform.
- Established at the platform level: GF said 22FDX embedded MRAM entered production in 2020.
- Not established for Efficient’s processor: Whether customer sampling happened by summer 2025, whether the processor entered production, its commercial availability, named customers, or deployment volumes.
- Not established by the cited material: Independent validation of Efficient’s headline energy-efficiency comparison.
For readers following the business significance, the key distinction is between a foundry process with a history of embedded-MRAM production and a particular partner chip whose sampling and commercial status have not been confirmed in the cited announcements.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
