> ## Content Index
> Fetch the complete content index at: https://aiadoption.org/llms.txt
> Use this file to discover other available public pages before exploring further.

# The most efficient AI chips now do 2.5 trillion operations per second per watt
- URL: https://aiadoption.org/data/ai-chip-energy-efficiency/
- Published: 2026-10-19T14:40:49.000Z
- Updated: 2026-10-21T06:01:16.000Z
- Description: The most energy-efficient AI chips deliver about 2.5 trillion operations per second per watt, roughly 33 times the leading chip of 2016.
- Author: Manu
- Tags: Data, AI Infrastructure, #chart, #ai-index-2026, #hub-infrastructure

2.5T

FLOP/s per watt: the most efficient leading chip in the data (late 2024), against 74.8 billion for the leading chip of 2016

Most efficient leading chip to date

20162025

## What the chart shows

Energy efficiency, the computation a chip delivers per watt of power, has risen steadily. The most efficient leading chip in the data, released in late 2024, delivers about 2.5 trillion operations per second (FLOP/s) per watt, roughly 33 times the 74.8 billion of the leading chip of April 2016.

The AI Index names Nvidia’s B200 and Google’s TPU v5e among the most efficient. But models have grown faster than chips have become more efficient, so the total power needed to train frontier models keeps rising ([related chart](https://aiadoption.org/data/power-draw-to-train-frontier-models/)).

## Key numbers

| Measure                                | Value              |
| -------------------------------------- | ------------------ |
| Most efficient leading chip (Nov 2024) | 2.5T FLOP/s per W  |
| Leading chip, Apr 2016                 | 74.8B FLOP/s per W |

## How to read it

- Leading hardware means the chips used for the most compute-intensive models of their time.
- Efficiency is based on manufacturers’ peak performance and rated power, not measured workloads.
- The vertical axis is logarithmic.

**Source:** [Epoch AI](https://aiadoption.org/sources/#epoch-ai), 2026, via the AI Index Report 2026 (Stanford HAI), figure 1.4.1: *Energy efficiency of leading machine learning hardware, 2016–25*. Chart and text by AI Adoption.  
Numbers checked against the published data on 19 October 2026.  
[Download the data (CSV)](https://charts.aiadoption.org/ai-index/csv/ai-chip-energy-efficiency.csv?ref=aiadoption.org) · [Cite this chart](#source)