---
format: "aidr-story-markdown/v1"
id: "89e223166653bc02c42896ad62bf2d0d2698214ca65d9c6e554924f94648bca9"
canonical_url: "https://aidr.today/89e22316?lang=en"
title: "Meta Commits to 1 GW Custom AI Chip Capacity For 2027 Rollout"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-15T15:06:23.000Z"
category: "Chips"
topics: ["meta","broadcom","tsmc","chips","infra"]
source_urls: ["https://huggingnews.com/ai/meta-commits-to-1-gw-custom-ai-chip-capacity-for-2027-rollout-6cf3e88a","https://x.com/wallstengine/status/2099863041534025739","https://x.com/financialjuice/status/2099861024744591545","https://x.com/financialjuice/status/2099860926249795810","https://x.com/financialjuice/status/2099861142935912585","https://marketbrief.now/ai/meta-commits-to-1-gw-custom-ai-chip-capacity-for-2027-rollout-6cf3e88a"]
summary: "The MTIA 450 chip, code-named Arke, will begin deployment across Meta data centers in the first half of 2027 to lower the energy and financial cost of running AI models. A second next-generation chip, the MTIA 500 code-named Astrid, will enter data centers by the end of 2027. Meta has committed to more than 1 gigawatt of custom silicon capacity over a 12 month period to provide better performance per watt than current Nvidia systems. Meta partnered with Broadcom for design and TSMC for manufacturing, receiving the first 12 Arke units on Sept. 1 with testing results within 2% to 3% of simulations. The company canceled a separate chip project called Olympus after determining that a design for both training and inference would cost 30% more. Later generations of custom silicon will focus on fiber optic technologies to increase throughput and speed."
---

# Meta Commits to 1 GW Custom AI Chip Capacity For 2027 Rollout

> [Open the canonical story](<https://aidr.today/89e22316?lang=en>)

**Published:** 2026-09-15T15:06:23.000Z
**Category:** Chips
**Topics:** meta, broadcom, tsmc, chips, infra

## Summary

The MTIA 450 chip, code\-named Arke, will begin deployment across Meta data centers in the first half of 2027 to lower the energy and financial cost of running AI models\. A second next\-generation chip, the MTIA 500 code\-named Astrid, will enter data centers by the end of 2027\. Meta has committed to more than 1 gigawatt of custom silicon capacity over a 12 month period to provide better performance per watt than current Nvidia systems\. Meta partnered with Broadcom for design and TSMC for manufacturing, receiving the first 12 Arke units on Sept\. 1 with testing results within 2% to 3% of simulations\. The company canceled a separate chip project called Olympus after determining that a design for both training and inference would cost 30% more\. Later generations of custom silicon will focus on fiber optic technologies to increase throughput and speed\.

## Sources

- [Story source](<https://huggingnews.com/ai/meta-commits-to-1-gw-custom-ai-chip-capacity-for-2027-rollout-6cf3e88a>)
- [Story source](<https://x.com/wallstengine/status/2099863041534025739>)
- [Supporting source](<https://x.com/financialjuice/status/2099861024744591545>)
- [Supporting source](<https://x.com/financialjuice/status/2099860926249795810>)
- [Supporting source](<https://x.com/financialjuice/status/2099861142935912585>)
- [Story source](<https://marketbrief.now/ai/meta-commits-to-1-gw-custom-ai-chip-capacity-for-2027-rollout-6cf3e88a>)