LPU
LPU is a Groq custom AI chip (ASIC) based on the TSP deterministic dataflow architecture, first released in 2023.
Built from BenchGecko data as of April 14, 2026 · updates when the data changes
LPU is a Groq custom AI chip (ASIC) based on the TSP deterministic dataflow architecture, first released in 2023.
Basic
The Groq Language Processing Unit is a custom AI chip (ASIC) made by Groq, codenamed GroqChip 1. Peak throughput is 188 TFLOPS at FP16/BF16. It carries 0 GB of SRAM on-die with 80 TB/s of memory bandwidth. Rated power (TDP) is 375 W. It is built on N14 at GlobalFoundries.
Deep
In the BenchGecko dataset it sits in the specialized tier: wafer-scale, LPU and other custom designs. Disclosed buyers include Aramco Digital. Specs come from manufacturer datasheets; the spec card on this page shows the dataset values and their as-of date.
Expert
That is 0.5 FP16 TFLOPS per watt. Memory bandwidth per unit of compute is 425.53 GB/s per FP16 TFLOP: a higher ratio helps memory-bound inference, while peak TFLOPS matter more for compute-bound training. Fabrication: USA and Germany. Real throughput depends on the model, precision, batch size and software stack, so compare tokens per second per dollar for your own workload.
Depending on why you're here
- ·LPU: 188 TFLOPS FP16/BF16
- ·Process: N14 at GlobalFoundries
- ·Tier: specialized
- ·0 GB of SRAM on-die per chip
- ·Cloud availability in the dataset: Groq
- ·Compare it with other chips on /hardware/groq-lpu
- ·Disclosed buyers: Aramco Digital
- ·Groq · fabbed at GlobalFoundries
- ·Supply tightness is tracked on /hardware
- ·LPU is a chip built to run AI, made by Groq
- ·It has 0 GB of fast memory so large models fit
- ·Companies buy or rent these to train and serve AI models
Frequently Asked Questions
Read the primary sources
- Groq LPU architecturewow.groq.com