🎤 Our co-founder & CEO martin snelgrove will be speaking this Wednesday at the IEEE AI Hardware & Infrastructure Working Group webinar. Topic: "Everything Old is New Again" - Energy-Efficient Tile-based At-Memory AI Compute HW Architecture. Martin will share lessons from years of building next-generation AI inference engines—from architectural tradeoffs to production-scale deployment challenges. 📅 Sept 17, 12PM ET 🔗 Register: https://lnkd.in/gqBBe9Eb This is part of IEEE’s new technical webinar series focused on the AI compute stack—from chip to cloud. #HepzibahAI #AIHardware #InferenceAtScale #EdgeAI #AIInfrastructure #Webinar #Semiconductors #IEEE #Founders #AIAcceleration
Hepzibah AI
Semiconductor Manufacturing
Toronto, ON 211 followers
Energy-efficient Full-stack AI Compute Architecture - from Models to Metals
About us
Toronto - Semiconductor IP - AI Compute Hepzibah designs drop-in AI-compute IP for the next era of ubiquitous intelligence. Just as an ARM® CPU core became indispensable to every modern SoC, our memory-centric, precision-adaptive AI engine is built to sit at the heart of any chip that must think—cloud accelerator, edge sensor, or anything in between.
- Website
-
www.hepzibah.ai
External link for Hepzibah AI
- Industry
- Semiconductor Manufacturing
- Company size
- 2-10 employees
- Headquarters
- Toronto, ON
- Type
- Privately Held
- Founded
- 2024
Locations
-
Primary
Toronto, ON, CA
Employees at Hepzibah AI
Updates
-
🛬 Hepzibah AI has officially landed in Taipei. We’re proud to be part of the Ontario Technology Mission to #SEMICONTaiwan 2025, and we’re kicking off the week with company visits in Hsinchu alongside some incredible Canadian innovators. Later this week: 📍 Ontario Pavilion — Booth J3140-3 🎤 UPNEXT Stage: Our co-founder & CEO martin snelgrove will share how we’re building next-gen AI compute from the transistor level up. Let’s connect — and talk about where inference, memory, light-training and efficiency intersect. #SEMICONTaiwan2025 #OntarioTechMission #UPNEXTStage #DeepTech #AIhardware #HepzibahAI #ComputeInMemory
-
-
🧠 AI needs a new kind of silicon. We’re thrilled to be part of #SEMICONTaiwan 2025 — showcasing our ultra energy-efficient, SRAM-based “at-memory” compute architecture, purpose-built for next-gen AI inference. Catch our CEO martin snelgrove on the #UPNEXT Stage, where he’ll share how Hepzibah is rethinking AI hardware from the transistor level up. 📍 Booth J3140 – Ontario Pavilion, Booth #J3140 TaiNEX2 Hall 1 🎤 UPNEXT Stage Presentation: 2:30pm UPNEXT Stage, Silicon Startups Zone, Booth #R8120, TaiNEX 2 (Taipei Nangang Exhibition Center, Hall 2) Whether you’re building the next-gen datacenter or pushing AI to the edge — come talk to us. 🧠 Built for inference. Designed for efficiency. Capable of light-training 🇨🇦 Proud to represent Canada’s hardtech/deeptech sector on the global stage. #SEMICONTaiwan2025 #UPNEXTStage #AIhardware #ComputeInMemory #DeepTech #Semiconductors #OntarioPavilion #CanadaInnovation
-
-
We’re excited to share this interview with our #Co-Founder & #CEO, martin snelgrove, featured by Judy Lin 林昭儀 on TechSoda. At Hepzibah AI, we believe the next wave of AI isn’t only about making chips bigger — it’s also about making them smarter and dramatically more efficient. In this TechSoda interview, Martin explains how our “models to metals” approach helps SoC teams embed ultra-efficient AI into devices from edge sensors to cloud servers. We’re heading to #SemiconTaiwan2025 this September to connect with chipmakers, foundries, and ecosystem partners who share our vision for a cleaner, more efficient future of AI computing. #Semiconductors #AIHardware #EdgeAI #ChipDesign #PowerEfficiency #SoC #ArtificialIntelligence
Hepzibah AI's guiding principle, “models to metals,” means mastering everything from the high-level #AI model to the #semiconductors’ physical routing density. Hepzibah’s early work involves delivering IP to IC designers through highly tailored, “white glove” engagements. Over time, the aim is to build a systematic IP library and become “the ARM of AI”—offering standardized, widely adopted architectures that simplify development and interoperability. Read the #interview: https://lnkd.in/dct6-Eef