Releases·Hardware Release
Kepler Computing Exits Stealth With EUV-Free 3D Memory Chips to Ease HBM Shortage
Kepler Computing, a San Jose startup in stealth since 2018, unveiled a new high-bandwidth memory (HBM) architecture it says targets the chip supply bottlenecks constraining the computing market. Founded by physicists and computer scientists, the company says its approach could ease the global memory-chip shortage if it can manufacture at scale.
0 engagement1 sourceWed, Sep 9, 2026, 01:00 PM
AI Overview
- Kepler Computing, founded in 2018 in San Jose, claims a 3D-stacking approach and proprietary material boost memory density without EUV lithography.
- Kepler raised $468 million from GlobalFoundries, Intel Capital, AMD Ventures, Baillie Gifford, and Bill Gates' Gates Frontier fund.
- The US Commerce Department committed up to $245 million in July to Kepler for 3D and ferroelectric AI memory technology development.
- Kepler produces chips in GlobalFoundries' 28-nanometer Singapore 'mini fabs' and says it converted a fab to next-gen in eight months versus 24.
- Kepler claims SRAM density matching 2nm or 3nm chips without EUV, after 35 composite iterations landed on a low-voltage ferroelectric material.
- Kepler plans first HBM samples this year, Singapore production ramp in 2027, and US chip production in 2028, after testing on about 2,000 wafers.
- GlobalFoundries' Ed Kaste flagged iron in Kepler's composite as a contamination risk requiring dedicated or fully encapsulated equipment.
Related
News·Company News
Coherent keeps all indium phosphide lasers for internal AI data center demand
Coherent, one of the few makers of indium phosphide lasers used for optical data movement in AI data centers, told investors it will not sell these lasers to outside customers for the foreseeable future because internal demand consumes 100% of production. This signals a supply constraint distinct from HBM shortages, potentially impacting AI infrastructure buildouts.
Coherent, one of the few makers of indium phosphide lasers used for optical data movement in AI data centers, told investors it will not sell these lasers to outside customers for the foreseeable future because internal demand consumes 100% of production. This signals a supply constraint distinct from HBM shortages, potentially impacting AI infrastructure buildouts.
119 engagement1 sourceredditTue, Aug 18, 2026, 05:41 AM
Developments·Hardware
Micron says HBM needs three times DDR5's wafer area, and the gap won't close
At Hot Chips 2026, Micron presented data showing HBM consumes roughly three times the wafer area of DDR5 for equal memory capacity. A Micron Fellow said the ratio will not improve with future generations. The comparison matters for anyone modeling AI memory costs and supply.
At Hot Chips 2026, Micron presented data showing HBM consumes roughly three times the wafer area of DDR5 for equal memory capacity. A Micron Fellow said the ratio will not improve with future generations. The comparison matters for anyone modeling AI memory costs and supply.
420 engagement1 sourceredditFri, Aug 28, 2026, 10:19 AM
News·Markets
Memory bottleneck reprices AI boom as HBM, DRAM, flash costs surge
A Medium article argues that memory components—HBM, ordinary DRAM, and flash storage—are becoming the dominant cost and performance bottleneck in AI systems, effectively repricing the entire AI boom. The piece connects chip prices to consumer electronics and industrial policy, suggesting that memory constraints will reshape AI economics.
A Medium article argues that memory components—HBM, ordinary DRAM, and flash storage—are becoming the dominant cost and performance bottleneck in AI systems, effectively repricing the entire AI boom. The piece connects chip prices to consumer electronics and industrial policy, suggesting that memory constraints will reshape AI economics.
0 engagement2 sourcesrssSat, Jul 11, 2026, 11:43 PM
News·Funding
AI chip startup Etched hits $10.3B valuation in Series C led by Sequoia
Etched, an AI chip startup founded by three Harvard dropouts in 2022, closed a $300 million Series C funding round at a $10.3 billion valuation. The round was led by Sequoia, with participation from Andreessen Horowitz, SK Hynix, Jane Street, and others, doubling its valuation from $5 billion in December 2025.
Etched, an AI chip startup founded by three Harvard dropouts in 2022, closed a $300 million Series C funding round at a $10.3 billion valuation. The round was led by Sequoia, with participation from Andreessen Horowitz, SK Hynix, Jane Street, and others, doubling its valuation from $5 billion in December 2025.
0 engagement1 sourcerssThu, Jul 23, 2026, 03:00 PM
News·Company News
OpenAI CFO Sarah Friar explains full-stack advances for abundant intelligence
OpenAI CFO Sarah Friar discussed how compounding advances in chips, compute, models, and products enable more useful intelligence at scale with lower costs. This provides insight into OpenAI's integrated strategy for scaling AI, relevant to practitioners tracking industry direction.
OpenAI CFO Sarah Friar discussed how compounding advances in chips, compute, models, and products enable more useful intelligence at scale with lower costs. This provides insight into OpenAI's integrated strategy for scaling AI, relevant to practitioners tracking industry direction.
0 engagement1 sourcerssTue, Aug 25, 2026, 07:05 AM