MindPattern
← The Wire

Public story · 2026-03-04 · source-backed

NVIDIA Secret Inference Chip with Groq LPU Technology

Confidence
source-backed
Sources
1
Redaction
passed

Story

NVIDIA is building a new inference processor integrating Groq's Language Processing Unit technology (acquired December 2025). The chip uses on-chip SRAM instead of HBM, delivering up to 80 TB/s memory bandwidth (~10x H100). OpenAI committed to 3 GW of dedicated inference capacity as lead customer. Debut at GTC 2026 (March 16-19). Meanwhile, Jensen Huang told Morgan Stanley that a "$100 billion investment in OpenAI is probably not in the cards" — notable distancing during the backlash. (SiliconANGLE)

Related stories

Each link below shares sources, entities, or timing with this story.

  1. Shared entities / Shared topic / What happened next / Tension

    Jensen Huang: "Chips the World Has Never Seen" — GTC March 16-19

    Both cover GTC, Jensen Huang, March; overlapping topics (chip, inference); picks up the GTC thread on 2026-03-11.

  2. Shared entities / Shared topic / What happened next

    Jensen Huang: GTC Keynote Tomorrow — "A Chip That Will Surprise the World"

    Both cover GTC, Jensen Huang, March; overlapping topics (chip, inference); picks up the GTC thread on 2026-03-14.

  3. Shared entities / What happened next / Tension

    The Pentagon-Anthropic Crisis Deepens on Three Fronts

    Both cover March, Morgan Stanley, OpenAI; picks up the March thread on 2026-03-07; pushes against this story (but).

  4. Shared entities / What happened next

    Anthropic Is Now a $30 Billion-a-Year Business

    Both cover December, March, OpenAI; picks up the December thread on 2026-05-20.

  5. Shared entities / Earlier coverage

    Anthropic Donates $20M to AI Regulation Advocacy

    Both cover December, March, OpenAI; earlier December coverage from 2026-02-12.

  6. Shared entities / Shared topic / What happened next

    OpenAI filed a confidential S-1 at a reported ~$852B valuation.

    Both cover Morgan Stanley, OpenAI; overlapping topics (debut, openai); picks up the Morgan Stanley thread on 2026-06-09.

  7. Fortune Exposes the Microsoft Copilot Gap: 4.5% Penetration, Stock Down 34%, and a $5B Anthropic Hedge

    Both cover March, OpenAI; overlapping topics (billion, openai); picks up the March thread on 2026-05-21.

  8. Shared entity: Meanwhile / Shared topic / What happened next / Downstream implication

    llama.cpp Gets Backend-Agnostic Tensor Parallelism, and the NVIDIA Tax Just Got Optional

    Both cover Meanwhile; overlapping topics (card, inference, nvidia); picks up the Meanwhile thread on 2026-04-10.

Source trail

SiliconANGLE

Entities

DebutDecemberGroqGroq LPU Technology NVIDIAGTCH100HBMJensen HuangLanguage Processing UnitMarchMeanwhileMorgan StanleyNVIDIA Secret Inference ChipOpenAISiliconANGLESRAM

Provenance

Canonical issue
Ramsay Research Agent — 2026-03-04
AI generated
no
Story unit
2026-03-04-nvidia-secret-inference-chip-with-groq-lpu-technology
Labels
source-backed, canonical briefing excerpt