in-depth: A Stealth Startup Thinks It Just Hacked the Memory
Kepler Computing has emerged from stealth with a new chip architecture using 3D stacking and proprietary materials to solve the global memory shortage, bypassing expensive EUV lithography. Backed by $468M and a US Commerce Department commitment, the startup aims to boost HBM and SRAM density, essential for the AI boom.
San Jose, California-based startup Kepler Computing, after seven years in stealth mode, has unveiled a groundbreaking approach to computer memory chip design that it claims can alleviate the persistent global memory shortage, particularly impacting the burgeoning artificial intelligence market. Founded in 2018 by a team of physicists and computer scientists, Kepler is introducing a novel architecture for high-bandwidth memory (HBM) and high-speed cache memory (SRAM) that bypasses traditional, expensive chip manufacturing processes. This announcement comes as the company secured substantial funding and government backing, signaling a significant potential shift in semiconductor production.
Rethinking Chip Architecture
Kepler Computing’s core innovation lies in its ability to dramatically increase memory density without relying on extreme ultraviolet lithography (EUV), the costly process typically used to shrink transistors. Instead, the startup employs a "3D stacking" method combined with a proprietary new material. This unique combination allows for greater chip density within existing semiconductor fabrication plants, offering a more accessible and potentially faster path to increased supply.
For high-bandwidth memory (HBM), which is crucial for data centers and AI applications, Kepler has developed a novel 3D-manufacturing technique. This method allows more memory chips to fit within a fixed footprint, bringing the core compute closer to the memory. The ultimate goal is to achieve data transfer energy consumption in HBM comparable to that of SRAM, while retaining HBM's large capacity.
Simultaneously, Kepler has enhanced the density of SRAM using ferroelectrics—materials known for their ability to read and write data at lower voltages. This advancement is powered by a newly developed low-voltage composite material, a result of 35 iterations by the Kepler team to optimize for easier and cheaper memory chip production. This allows them to achieve densities comparable to 2-nanometer or 3-nanometer chips without the need for advanced EUV equipment.
Significant Investment and Strategic Partnerships
The company has garnered significant financial support, totaling $468 million from prominent investors including GlobalFoundries, Intel Capital, AMD Ventures, British investment fund Baillie Gifford, and Bill Gates’ private Gates Frontier fund. Further bolstering its efforts, the US Department of Commerce pledged up to $245 million in July to support Kepler’s development of high-performance AI memory technology utilizing 3D and ferroelectric innovations within the United States.
This influx of capital underscores the urgent need for solutions to the memory chip shortage, exacerbated by the booming demand for AI. Traditional chipmakers, like SK Hynix and Micron, are investing billions in new fabs, a process that is both expensive and time-consuming. Kepler's strategy offers a potential alternative by optimizing existing infrastructure rather than solely relying on new, colossal facilities.
GlobalFoundries, a manufacturing partner and investor, is already working with Kepler, converting existing fabs into “next-generation” facilities in as little as eight months. This rapid conversion stands in stark contrast to the typical 24-month timeframe required for such upgrades, highlighting the efficiency gains Kepler’s approach promises.
The Road to Scale Production
While the technological breakthroughs are significant, Kepler Computing still faces the formidable challenge of scaling production to meet global demand. To date, their technology has been tested on approximately 2,000 wafers. The company aims to ship its first HBM chip samples later this year (2026), ramp up production in Singapore in 2027, and commence US chip production in 2028.
One notable hurdle highlighted by Ed Kaste of GlobalFoundries is the composite material used by Kepler, which includes iron. Iron is a tough contaminant in semiconductor manufacturing, requiring dedicated equipment or full encapsulation to prevent its escape into the production facility. Overcoming such manufacturing complexities at scale will be crucial for Kepler's success.
Industry analysts, like Austin Lyons of Creative Strategies, emphasize that proving a new memory method is one thing, but meeting historic demand is an entirely different feat. This sentiment echoes skepticism faced by other innovative semiconductor startups attempting to disrupt established norms. Kepler Computing’s emergence from stealth offers a hopeful, albeit ambitious, vision for addressing critical bottlenecks in the chip industry. If successful, their innovations could profoundly impact the future of AI and advanced computing.
FAQ
Q: What is Kepler Computing's main innovation for memory chips?
A: Kepler Computing introduces a novel approach using "3D stacking" and a proprietary material, along with ferroelectrics, to increase memory density for both high-bandwidth memory (HBM) and SRAM. This method allows them to achieve higher densities comparable to advanced nodes (e.g., 2-3nm) without relying on expensive extreme ultraviolet lithography (EUV) equipment, working within existing fabrication plants.
Q: When does Kepler Computing expect to begin mass production?
A: The company plans to ship its first samples of HBM chips later this year (2026). They intend to ramp up production out of Singapore next year (2027) and begin chip production in the US in 2028, but acknowledge a long road to full-scale production.
Q: What are the primary challenges Kepler Computing faces in scaling its technology?
A: The main challenges include scaling production from thousands of wafers to millions of devices, managing potential contamination from their proprietary iron-containing composite material, which requires specialized handling or encapsulation, and ensuring the new method can meet incredibly stringent specifications on time and on budget to fulfill historic demand.
Related articles
Xi pitches open-source AI to BRICS amid domestic curb debates
Chinese President Xi Jinping proposed a China-led open-source AI community and invited BRICS nations to join the World AI Cooperation Organization (WAICO) at the recent BRICS summit. This push for global collaboration contrasts sharply with Beijing's ongoing internal debates about restricting its own advanced AI models. Meanwhile, the EU's comprehensive AI Act, with its clear, enforceable rules for open-source AI, highlights a significant divergence in global AI governance approaches.
Tesla Set to Finally Unveil Second-Generation Roadster on October 1
The much-anticipated second generation of the Tesla Roadster, a halo vehicle promising revolutionary performance, is finally slated for a public unveiling on October 1. After years of delays and a protracted development
Seattle Warned on Big Tech Reliance; Microsoft/OpenAI Sued; Apple's
A new City of Seattle study warns of the city's risky economic over-reliance on a few dominant tech companies. Simultaneously, the Seattle Times and Newsday are suing Microsoft and OpenAI for alleged AI training data theft, while Apple's new foldable iPhone Duo evokes memories of Microsoft's defunct Surface Duo.
in-depth: The Best 3-in-1 Apple Charging Stations After Testing 30
Wired has released its top picks for 3-in-1 Apple charging stations, extensively tested for iPhone, Apple Watch, and AirPods. The guide highlights six leading models, from premium speedy options to budget-friendly and compact designs, all focused on decluttering and optimizing charging for Apple users.
Nscale Adds Former OpenAI Exec Fidji Simo to Board Ahead of IPO
Nscale, the U.K.-based AI data center startup, has appointed former OpenAI, Meta, and Instacart executive Fidji Simo to its board of directors. This high-profile addition comes as Nscale prepares for a potential IPO this fall, leveraging Simo's extensive experience in scaling major tech platforms and guiding a company through a successful public offering.
Microsoft comms chief Frank Shaw to exit after nearly three decades
Frank X. Shaw, Microsoft's long-serving chief communications officer, will exit at year-end after nearly three decades shaping the company's message through pivotal periods. Shaw, 64, is not retiring but plans a break before his next move, leaving behind a legacy of adapting communications for a digital age and embracing AI tools. Microsoft is now searching for his successor.