Meta Launches Meta One Subscription Service and Advances Custom MTIA AI Chip Program
Summary
- • Meta launches Meta One AI subscription at $2.99–$14.99/month across Instagram, Facebook, WhatsApp
- • Meta One records 15 million subscriptions and trials at launch with 50+ premium features
- • Meta MTIA 450 chip (Arke) set for H1 2027 data center deployment to cut costs and energy use
- • Meta partners with Broadcom and TSMC on custom chips, reducing dependence on NVIDIA
Details
Meta One subscription launches at $2.99–$14.99/month across Meta's full app ecosystem
Meta launched Meta One on September 15, 2026, offering premium AI features across Instagram, Facebook, WhatsApp, and Meta AI. Single-product plans start at $2.99/month, individual bundles at $7.99, and creator/business bundles at $14.99/month.
Meta One records 15 million subscriptions and trials at launch with 50+ features
At launch, Meta One has recorded 15 million subscriptions and trials. The service offers 50+ features including higher AI usage limits, image and video generation, and creator and business tools. The core experience remains free.
Meta MTIA 450 chip (Arke) to deploy in H1 2027 data centers to cut cost and energy use
Meta plans to deploy its third-generation custom AI chip, MTIA 450 code-named Arke, in data centers during the first half of 2027. The chip is designed to cut costs and reduce energy consumption as Meta's AI infrastructure scales.
MTIA 500 (Astrid) completes design in ~1 month, targets deployment by end of 2027
Meta's next-generation chip MTIA 500, code-named Astrid, is expected to complete design work within approximately one month, with deployment targeted for end of 2027, indicating an accelerating custom silicon roadmap.
Meta partners with Broadcom on chip design and TSMC on chip production
Meta is working with Broadcom on chip designs and Taiwan Semiconductor Manufacturing Company (TSMC) on production — a partnership structure similar to Apple's chip development approach.
Custom chip program aims to reduce Meta's reliance on NVIDIA processors
Meta's custom silicon push is explicitly framed as a path to reducing reliance on NVIDIA's processors — a strategic priority as AI compute costs soar and supply constraints have affected major hyperscalers.
Meta VP Yee Jiun Song: chips target better performance per watt and per dollar
Meta VP of Engineering Yee Jiun Song stated the custom chips are designed to deliver better performance per watt and per dollar compared to off-the-shelf alternatives, underpinning the efficiency rationale for the custom silicon program.
Meta's dual September 2026 announcements: Meta One subscription service launch and MTIA custom AI chip program advancement
What This Means
Meta is making two significant moves simultaneously: a direct consumer monetization push via Meta One, which monetizes its AI capabilities at accessible price points across its entire app ecosystem, and a long-term infrastructure bid to reduce NVIDIA dependency through custom silicon. The 15 million trials at Meta One's launch signals strong initial demand for premium AI features at the consumer level. The MTIA chip roadmap — with two generations in flight and TSMC/Broadcom partnerships — mirrors Apple's and Google's chip strategies, signaling Meta's intention to control its AI compute stack rather than remain dependent on external GPU suppliers.