Stealth startup Kepler Computing emerges to tackle memory shortage with 3D stacking
nordicinst · x · 2026-09-09
Kepler Computing, a San Jose startup founded in 2018, has exited stealth mode claiming a new HBM architecture that could ease the global memory shortage. Key points:
- Approach: Boosts density via "3D stacking" and a proprietary material — no EUV lithography needed, and it works with existing fabs.
- SRAM too: Claims similar gains for high-speed cache memory in CPUs, GPUs and XPUs.
- Funding: $468M raised from GlobalFoundries, Intel Capital, AMD Ventures, Baillie Gifford and Bill Gates.
Scaling the technology remains the big open question.
More from Infra
- Broadcom ASICs podcast: GPU supply means little when substrates and MLCCs bottleneck output — BenBajarin · 2026-09-09
- 27B 1-bit model runs in the browser at 25-30 tok/s on a 6GB RTX 3060 laptop (WebGPU) — mentria-ai · 2026-09-09
- The expensive part of coding agents isn't the agents—it's the silent retries ($900 for one task) — mrtrly · 2026-09-09
- AI infra is unbundling: model, harness, inference and compute become four separate choices — _changxu · 2026-09-09
- Running a 27B Q4 Model at 131K Context on a Single RTX 3090: ~25.6 tok/s Measured — bjivanovich · 2026-09-09
- Polymarket puts 73% odds on a US state enacting a data center moratorium by 2026 — Polymarket · 2026-09-09