TREE NEWS update: Meta Platforms plans to begin deploying a new in-house AI chip, MTIA 450 (codename Arke), in its data centers in the first half of next year to cut costs and energy use in running AI models. The company is currently testing the third-generation chip, while the next version, codename Astrid, will finish design in about a month and reach data centers by late 2027. Engineering VP Yee Jiun Song said Meta expects the chips to beat Nvidia processors on performance per watt and per dollar.
Meta to Deploy New In-House AI Chip MTIA 450 in Data Centers Next Year
Meta's in-house silicon push is less about chip design than about vertical integration: owning the compute layer changes the cost structure of running AI at scale, and the per-watt, per-dollar framing targets the exact metric where merchant GPUs are most vulnerable. The claim lands as an assertion, not a benchmark, and Meta's roadmap already runs several generations deep, so the real signal is sustained commitment rather than any single part. Whether internal silicon meaningfully dents reliance on external suppliers — or simply diversifies it — is the open question worth tracking.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.