Nvidia (NVDA) Stock Slides as Meta Prepares In-House AI Chip Launch This Fall

Key Takeaways

Table of Contents

Nvidia shares decreased 0.3% to $202.14 during Friday’s premarket session
Meta is launching its proprietary AI chip, dubbed “Iris,” with production beginning in September
The new chip aims to complement rather than eliminate GPU acquisitions from Nvidia and AMD
Google and Amazon are similarly positioning their custom AI chips for external market sales
Industry experts suggest Nvidia’s total revenue from major cloud providers may continue rising despite declining market share percentages

Nvidia shares retreated 0.3% to $202.14 during Friday’s premarket session after Meta Platforms revealed intentions to commence production of its proprietary AI chip starting September.

NVIDIA Corporation, NVDA

Dubbed “Iris,” Meta’s chip represents one phase of a comprehensive four-generation initiative within its Meta Training and Inference Accelerator (MTIA) framework. The semiconductor is being developed in collaboration with Broadcom and will roll off production lines at Taiwan Semiconductor Manufacturing Co (TSMC).

According to an internal communication obtained by Reuters, debugging procedures required merely six weeks and uncovered no critical defects — an impressively expedited timeline for an initiative that has encountered obstacles since its inception over five years ago.

Meta intends to unveil successive chip iterations approximately every six months extending through 2027, a considerably accelerated pace compared to the standard industry cycle of annual releases or longer.

The underlying objective is clear: minimize computational expenses and decrease dependence on third-party chip manufacturers including Nvidia and AMD.

However, “Iris” isn’t positioned to completely supplant Nvidia GPUs. Instead, the chip is intended to supplement the substantial GPU volumes Meta continues purchasing. Meta’s internal documentation conceded that implementing cutting-edge GPUs at the company’s scale “has been a heavy lift, and it has cost us time.”

Expert Analysis

Benchmark Research analyst Cody Acree challenged the notion that this development signals disaster for Nvidia. He observed that although Nvidia might surrender relative market share within Meta’s expanding infrastructure plans, aggregate hyperscaler expenditures are projected to more than double, suggesting Nvidia’s total revenue from these clients could still expand.

Meta anticipates allocating up to $145 billion toward AI infrastructure throughout this year alone, forming part of a broader technology sector investment wave projected to surpass $700 billion.

Broader Implications for Nvidia

The Meta announcement represents just one element of an evolving competitive landscape. Google and Amazon are each positioning their proprietary AI chips — Google’s Tensor Processing Units and Amazon’s Trainium processors — for commercial availability to outside clients.

Both solutions demonstrate capability in operating contemporary AI models and potentially deliver superior cost-efficiency for specific computing tasks, introducing additional competitive challenges to Nvidia’s market leadership.

Nvidia’s stock performance has trailed the semiconductor sector lately, with AMD climbing 5.67% and Meta advancing 4.70% during the session, while Nvidia remained negative.

Nasdaq 100 futures declined 0.2%, indicating Nvidia’s movement aligned with broader market trends, though company-specific developments contributed additional downward pressure.

Meta plans infrastructure deployment reaching seven gigawatts of computing capacity in 2026, with intentions to double that volume in 2027, having already secured long-term supply contracts with Samsung, Sandisk, and Sumitomo Electric to facilitate that expansion.

✨ Limited Time Offer