Key Highlights
- Nvidia (NVDA) shares declined 0.3% to $202.14 during Friday’s premarket session
- Meta is set to launch its proprietary AI chip, dubbed “Iris,” starting in September
- The custom chip aims to complement rather than completely substitute GPU acquisitions from Nvidia and AMD
- Both Google and Amazon are advancing plans to commercialize their proprietary AI processors to third-party clients
- Industry experts believe Nvidia’s total revenue from cloud giants will continue expanding despite declining market share percentages
Nvidia shares experienced a modest 0.3% decline to $202.14 during Friday’s premarket hours, reacting to revelations that Meta Platforms intends to launch production of its proprietary AI processing chip by September.
The chip, internally referenced as “Iris,” represents one component of a comprehensive four-phase initiative within Meta’s Training and Inference Accelerator (MTIA) framework. Development is occurring in partnership with Broadcom, while Taiwan Semiconductor Manufacturing Co (TSMC) will handle fabrication.
According to an internal document obtained by Reuters, the debugging phase required merely six weeks and revealed no critical defects — representing a remarkably compressed development cycle for an initiative that has encountered obstacles since its inception over half a decade ago.
Meta has outlined an aggressive roadmap to introduce updated chip versions approximately every six months extending through 2027, significantly outpacing the standard industry release pattern of annual or longer intervals.
The strategic objective is clear: minimize computational expenses and decrease dependence on third-party semiconductor providers such as Nvidia and AMD.
However, “Iris” isn’t intended to completely eliminate Nvidia GPU purchases. Instead, the processor is meant to supplement the substantial GPU volumes Meta continues to procure. Meta’s internal documentation conceded that implementing cutting-edge GPUs across its massive infrastructure “has been a heavy lift, and it has cost us time.”
Expert Analysis and Market Perspective
Benchmark Research analyst Cody Acree dismissed concerns that this development signals significant problems for Nvidia. He emphasized that although Nvidia might experience declining relative market share within Meta’s expanding infrastructure projects, total hyperscaler investment is anticipated to more than double, suggesting Nvidia’s actual dollar revenue from these clients could still increase substantially.
Meta projects infrastructure spending reaching $145 billion dedicated to AI capabilities this year alone, contributing to an industry-wide technology investment surge expected to surpass $700 billion.
Broader Competitive Landscape for Nvidia
The Meta announcement represents just one element of an evolving competitive environment. Google and Amazon are simultaneously advancing initiatives to market their proprietary AI processors — specifically Google’s Tensor Processing Units and Amazon’s Trainium chips — to external enterprise customers.
Both alternatives demonstrate capability in executing contemporary AI workloads and may deliver superior cost-performance ratios for specific applications, intensifying competitive challenges to Nvidia’s market leadership.
Nvidia’s equity performance has recently underperformed the semiconductor sector broadly, with AMD gaining 5.67% and Meta advancing 4.70% during the trading session, while Nvidia posted negative returns.
Nasdaq 100 futures contracts showed a 0.2% decline, indicating Nvidia’s movement aligned generally with broader market trends, though company-specific developments contributed additional downward momentum.
Meta has committed to deploying seven gigawatts of computational capacity in 2026, with plans to double that infrastructure in 2027, supported by long-term supply contracts already secured with Samsung, Sandisk, and Sumitomo Electric to facilitate that ambitious expansion.
