Gaudi 3
Gaudi 3 is an Intel custom AI chip (ASIC) based on the Heterogeneous MME + TPC architecture, first released in 2024.
Built from BenchGecko data as of April 14, 2026 · updates when the data changes
Gaudi 3 is an Intel custom AI chip (ASIC) based on the Heterogeneous MME + TPC architecture, first released in 2024.
Basic
The Intel Gaudi 3 is a custom AI chip (ASIC) made by Intel. Peak throughput is 1,835 TFLOPS at FP16/BF16 and 1,835 TFLOPS at FP8. It carries 128 GB of HBM2e with 3.7 TB/s of memory bandwidth. Rated power (TDP) is 900 W. It is built on N5 at TSMC.
Deep
In the BenchGecko dataset it sits in the mainstream tier: current-generation non-flagship parts or last-generation flagships still widely deployed. Disclosed buyers include IBM. Specs come from manufacturer datasheets; the spec card on this page shows the dataset values and their as-of date.
Expert
That is 2.04 FP16 TFLOPS per watt. Memory bandwidth per unit of compute is 2.02 GB/s per FP16 TFLOP: a higher ratio helps memory-bound inference, while peak TFLOPS matter more for compute-bound training. Fabrication: Taiwan. 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
- ·Gaudi 3: 1,835 TFLOPS FP16/BF16 · 1,835 TFLOPS FP8
- ·Process: N5 at TSMC
- ·Tier: mainstream
- ·128 GB of HBM2e per chip
- ·Cloud availability in the dataset: Intel Tiber and IBM Cloud
- ·Compare it with other chips on /hardware/gaudi-3
- ·Disclosed buyers: IBM
- ·Intel · fabbed at TSMC
- ·Supply tightness is tracked on /hardware
- ·Gaudi 3 is a chip built to run AI, made by Intel
- ·It has 128 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
- Gaudi 3 announcementwww.intel.com