Kepler Computing Secures $468M From Intel Capital, AMD Ventures, Bill Gates to Deliver EUV-Free HBM for AI

Kepler Computing, a San Jose-based chip startup, has emerged from stealth mode to address the AI memory shortage with a novel high-bandwidth memory (HBM) architecture. This new approach increases memory density through 3D stacking and proprietary materials, crucially avoiding the need for expensive extreme ultraviolet (EUV) lithography equipment. For broader context, explore our AI News. For broader context, explore our AI Tools Pricing.

This strategic pivot towards HBM, driven by the immense demand spurred by the ChatGPT era, positions Kepler Computing to address the tightest supply chain component in modern AI systems. The company has garnered substantial financial backing, raising $468 million from prominent investors including GlobalFoundries, Intel Capital, AMD Ventures, Baillie Gifford, and Bill Gates.

Addressing the AI Memory Bottleneck Without EUV

The core of Kepler Computing's innovation lies in its ability to achieve high-density memory without relying on EUV lithography, a complex and costly manufacturing process essential for producing the smallest features on advanced chips. By utilizing a unique 3D stacking technique and proprietary materials, Kepler claims its architecture can deliver SRAM density comparable to chips manufactured with 2- or 3-nanometer-class processes, but without the associated EUV requirements.

This development is particularly significant as the demand for HBM in AI accelerators continues to surge. Traditional HBM production often faces limitations due to the intricate manufacturing processes and the high capital expenditure associated with EUV technology. Kepler's alternative method could potentially streamline production and reduce costs, offering a more accessible path to high-performance memory.

Strategic Investments and US Development

Kepler Computing's emergence is bolstered by significant financial commitments. Beyond the $468 million raised from private investors, the US Department of Commerce announced in July a commitment of up to $245 million to support the development of Kepler's technology within the United States. This federal backing underscores the strategic importance of domestic semiconductor innovation, particularly in critical areas like AI memory.

The company's production efforts are currently focused on

Sources

• A Stealth Startup Thinks It Just Hacked the Memory Shortage | WIRED
• Department of Commerce Announces Letters of Intent With 7 Companies for $874 Million to Accelerate Semiconductor R&D for the Compute Supply Chain | NIST
• Intel Capital on X: "Today, Kepler Computing (@keplercompute) emerged from stealth after over 7 years in development. Their new #memorychip architecture boosts HBM and SRAM density without relying on costly #EUV lithography, addressing a real bottleneck in the memory supply chain. Read more from @… / X
• Kepler Computing Emerges to Build HBM Alternative Using FeRAM | TechPowerUp