SANTA CLARA, Calif., April 07, 2026 (GLOBE NEWSWIRE) -- As AI infrastructure drives an unprecedented surge in memory demand, DRAM has become the scarcest and most expensive resource in computing. Today, MEXT announced Predictive Memory™, a breakthrough software-only solution that cuts infrastructure costs by 50% and expands usable memory capacity by 2-4x.
“Memory is an increasingly critical component of modern compute infrastructure,” said Robert Hormuth, Corporate Vice President, Architecture and Strategy, AMD. “As AI and data-intensive workloads continue to grow, finding new approaches to memory scalability is becoming an industry imperative. We have been working closely with MEXT and see their technology as a strong answer to this imperative.”
“MEXT was founded three years ago around a simple conviction: DRAM costs would become the defining constraint on computing, already accounting for half the cost of a server,” said Gary Smerdon, Founder and CEO of MEXT. “Today, with prices surging 3.5x in the past two quarters alone, that prediction has become the reality — and the industry is hitting a breaking point. MEXT’s Predictive Memory™ software is redefining how memory is deployed, slashing costs and unlocking new capacity.”
The Memory Crisis: 50 Years in the Making
While processors, networking, and storage have undergone revolutionary transformations over the past half century, DRAM has remained architecturally unchanged since 1970, when Intel introduced the first device to the industry. Despite billions invested by industry leaders in next-generation memory technologies, every attempt to displace DRAM's decades-old architecture has failed. Studies from Meta, Microsoft, and Google have all highlighted that the cost of this stagnation is staggering. Now, memory, not compute, is the bottleneck limiting how fast organizations can scale.
Predictive Memory™: AI Solving the Problem AI Created
MEXT Predictive Memory™ attacks the DRAM problem by transparently bringing flash into the memory domain. Much like how large language models predict the next logical word in a natural language pattern, MEXT’s AI engine predicts the next logical memory page in a workload´s behavioral pattern.
The process works in three steps: MEXT identifies memory pages that are not actively in use (cold) and offloads them to flash, which costs 50x less than DRAM. MEXT’s patent-pending AI engine predicts which offloaded memory pages will be needed soon. The engine proactively pushes those pages back to DRAM before they are required, so the application experiences little to no performance impact.
In this way, MEXT creates a new price-performance tier: DRAM-class performance with flash-level economics and capacity.
Key Specifications:
- Both training and inference (predictions) run on a single CPU core — no GPU required
- Makes predictions in microseconds and continually self-optimizes
- Works on-premises, on new or existing servers, or in the cloud (AWS, GCP, and others)
- Requires no changes to existing hardware, OS, or applications
- Installs in under five minutes
“As DRAM supply remains constrained and memory costs stay elevated, enterprises are paying closer attention to software-driven approaches that can improve utilization and reduce overprovisioning,” said Mast Eastwood, Senior Vice President of Enterprise Infrastructure, Cloud, Telecom, Security, Developers, Channels and Enabling Technology at IDC. “MEXT has entered the market at a favorable time. Enterprises should consider the technology and validate performance and operational fit through proof-of-concept testing.”
Delivering Results Where Memory Matters Most
MEXT is already delivering measurable impact across a range of memory-intensive workloads where performance, cost, and scalability are critical.
Media and Entertainment: At a premier Hollywood Movie Studio, MEXT is transforming animation and special effects rendering by cutting the required DRAM footprint in half while also increasing available memory. Artists can iterate more and production teams can scale faster without needing to purchase DRAM, yielding lower infrastructure costs — all without changes to existing workflows or systems.
Semiconductor / EDA: One of the World’s Leading Chip Manufacturers is accelerating time-to-results across advanced electronic design automation (EDA) workloads. By expanding memory capacity, MEXT allows engineers to run larger, more complex simulations — reducing bottlenecks, shortening design cycles, and helping bring new chips to market faster.
Gaming: A Gaming Company is using MEXT to effectively double computing and memory capacity without purchasing additional DRAM. This allows the company to support richer, more immersive environments while maintaining tight control over infrastructure spend — critical for scaling in a highly competitive and fast-paced market.
Availability
MEXT Predictive Memory™ is available today for deployment on-premises or in the cloud. Organizations can begin a proof-of-concept in under five minutes, with no infrastructure changes required. MEXT is licensed as a subscription with an annual list price of $3.99 / GB. Visit www.mext.ai/contact to get started.
Additional Resources
- MEXT Overview Video (3 min)
- Quick Product Demo (1 min)
- Technology Overview
- DRAM Market Update Webinar
- MEXT Blog
For media inquiries, contact Juhi Muthal at juhi.muthal@mext.ai.
About MEXT
Founded in 2023, MEXT is transforming the economics of computing with AI-Powered Predictive Memory™ technology. The company was built by a team with deep roots in systems innovation, including pioneers of CPU virtualization, data center flash, and the creation of the Internet itself. MEXT’s patent-pending, software-only solution makes flash storage perform as DRAM-speed memory — empowering customers to run applications within a far smaller DRAM footprint for 50% lower costs or expand memory capacity by 2-4x. MEXT works transparently on-premises or in the cloud, requiring no changes to the underlying hardware, OS, or applications. Learn more at https://www.mext.ai.
