{"slug": "meta-to-begin-manufacturing-in-house-iris-ai-chip-in-september", "title": "Meta to Begin Manufacturing In-House 'Iris' AI Chip in September", "summary": "Meta Platforms will begin manufacturing its in-house AI chip, code-named Iris, in September, according to an internal memo. The chip, designed with Broadcom and fabricated by TSMC, is part of Meta's MTIA program to reduce dependence on Nvidia GPUs. Meta plans to double computing capacity to 14 gigawatts by 2027 and spend up to $145 billion on AI infrastructure in 2026.", "body_md": "# Meta to Begin Manufacturing In-House 'Iris' AI Chip in September\n\n- Meta will begin manufacturing its in-house AI chip, code-named Iris, in September, according to a company memo obtained by Reuters\n[[1]](https://www.cnbc.com/2026/07/09/meta-to-put-ai-chip-into-production-in-september-report.html) - The chip cleared bug testing in six weeks with no major issues; Broadcom serves as design partner and TSMC handles fabrication\n[[2]](https://thenextweb.com/news/meta-mtia-ai-chip-production-september) - Meta plans to scale computing capacity from 7 gigawatts in 2026 to 14 gigawatts in 2027\n[[1]](https://www.cnbc.com/2026/07/09/meta-to-put-ai-chip-into-production-in-september-report.html) - The company expects to spend as much as $145 billion on AI infrastructure in 2026\n[[2]](https://thenextweb.com/news/meta-mtia-ai-chip-production-september) - Meta has secured long-term supply agreements with Samsung, SanDisk, and Sumitomo Electric for memory, flash storage, and fiber-optic equipment\n[[3]](https://www.933thedrive.com/2026/07/09/exclusive-meta-to-put-ai-chip-into-production-in-september-as-it-looks-to-double-computing-capacity-memo-shows/)\n\nMeta Platforms plans to begin manufacturing its proprietary AI chip, code-named Iris, starting in September, according to an internal company memo obtained by Reuters [1]. The chip is one of four generations under Meta's MTIA (Meta Training and Inference Accelerators) program and represents the company's most significant move yet toward in-house silicon for AI workloads.\n\nBroadcom is serving as Meta's design partner on the chip, while Taiwan Semiconductor Manufacturing Co. will handle fabrication [2]. The chip completed its bug-testing phase in roughly six weeks without surfacing any significant problems, according to the memo\n\n.\n\n[[1]](https://www.cnbc.com/2026/07/09/meta-to-put-ai-chip-into-production-in-september-report.html)The production milestone is part of a broader infrastructure buildout that will see Meta expand its computing capacity from 7 gigawatts in 2026 to 14 gigawatts in 2027 [1]. Meta has guided 2026 capital expenditure to between $125 billion and $145 billion, with nearly all of the increase directed toward data centers, GPUs, and custom silicon\n\n.\n\n[[2]](https://thenextweb.com/news/meta-mtia-ai-chip-production-september)## The Chip\n\nIris is the latest entry in Meta's MTIA lineup, which the company has been developing to handle AI training and inference tasks across Facebook, Instagram, WhatsApp, and its generative AI products. In March 2026, Meta unveiled four MTIA variants — the 300, 400, 450, and 500 — with a planned six-month release cadence, roughly double the typical pace in the chip industry [2].\n\nThe MTIA 300 is already in production handling ranking and recommendation workloads across Meta's apps. The 450 and 500 variants target generative image and video inference through 2027 [2]. The Broadcom partnership on custom silicon extends through 2029 and covers multiple MTIA generations, with newer parts expected to be among the first custom AI chips built on a 2-nanometer process\n\n.\n\n[[2]](https://thenextweb.com/news/meta-mtia-ai-chip-production-september)## Why It Matters\n\nMeta's custom chip program is designed to reduce the company's dependence on third-party GPU vendors, principally Nvidia and AMD. The internal memo noted that adopting the latest GPUs at a company of Meta's scale has been difficult and time-consuming [3].\n\nHowever, analysts frame MTIA as absorbing growth rather than replacing Nvidia entirely. Meta continues to expand GPU purchases alongside its custom silicon deployment [2]. Nvidia shares fell 1.2% on the day to $201.63, while Broadcom rose 2.6% to $398.63\n\n. Meta shares were roughly flat at $602.86\n\n[[4]](https://financialmodelingprep.com).\n\n[[4]](https://financialmodelingprep.com)## The Supply Chain\n\nBeyond the Broadcom and TSMC partnerships, Meta has locked in long-term supply agreements with Samsung for memory chips, SanDisk for flash storage, and Sumitomo Electric for fiber-optic equipment [3]. The agreements reflect the scale of Meta's infrastructure ambitions and the company's effort to secure components ahead of industrywide demand.\n\nThe infrastructure expansion plan calls for 7 gigawatts of computing capacity coming online in 2026, doubling to 14 gigawatts by 2027 [1]. For context, 14 gigawatts exceeds the total electricity consumption of many small countries and underscores the enormous energy demands of training and running large AI models at scale.\n\n## What's Next\n\nMeta's chip production timeline puts it on a similar path to other hyperscale cloud companies building custom silicon, including Google with its TPU line and Amazon with its Trainium and Inferentia chips. The September start date gives Meta roughly three months to ramp production before the end of the year.\n\nThe company's $125 billion to $145 billion capex guidance for 2026 makes it one of the largest single-year infrastructure spenders in corporate history [2]. With chip launches planned approximately every six months through 2027, Meta is betting that vertically integrated AI hardware will deliver cost and performance advantages over relying solely on merchant silicon from Nvidia and AMD\n\n.\n\n[[2]](https://thenextweb.com/news/meta-mtia-ai-chip-production-september)## Further sources\n\n[[1] Reuters exclusive via CNBC: Meta to put AI chip into production in September as… ↗](https://www.cnbc.com/2026/07/09/meta-to-put-ai-chip-into-production-in-september-report.html)\n\n[[2] The Next Web: Meta to put its own AI chip into production in September, aiming … ↗](https://thenextweb.com/news/meta-mtia-ai-chip-production-september)\n\n[[3] Reuters exclusive via 93.3 The Drive: Meta to put AI chip into production in Se… ↗](https://www.933thedrive.com/2026/07/09/exclusive-meta-to-put-ai-chip-into-production-in-september-as-it-looks-to-double-computing-capacity-memo-shows/)\n\n[[4] Financial data from FMP (Meta, Nvidia, Broadcom real-time quotes, July 9, 2026) ↗](https://financialmodelingprep.com)\n\nThe stories that matter, in one email. Free — unsubscribe anytime.", "url": "https://wpnews.pro/news/meta-to-begin-manufacturing-in-house-iris-ai-chip-in-september", "canonical_source": "https://mlq.ai/news/meta-to-begin-manufacturing-in-house-iris-ai-chip-in-september/", "published_at": "2026-07-09 19:35:03+00:00", "updated_at": "2026-07-12 17:36:33.739124+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-chips", "ai-infrastructure", "ai-products", "ai-startups"], "entities": ["Meta", "Broadcom", "TSMC", "Nvidia", "Samsung", "SanDisk", "Sumitomo Electric", "AMD"], "alternates": {"html": "https://wpnews.pro/news/meta-to-begin-manufacturing-in-house-iris-ai-chip-in-september", "markdown": "https://wpnews.pro/news/meta-to-begin-manufacturing-in-house-iris-ai-chip-in-september.md", "text": "https://wpnews.pro/news/meta-to-begin-manufacturing-in-house-iris-ai-chip-in-september.txt", "jsonld": "https://wpnews.pro/news/meta-to-begin-manufacturing-in-house-iris-ai-chip-in-september.jsonld"}}