Nvidia (NVDA) shares edged lower on Thursday, extending a recent pullback.
The stock started the day in the green and even went on to reclaim the $200 mark, but fell shortly after.
Shares of Nvidia fell about 1% in early trading after finishing Wednesday below the $200 level, a level it has struggled to hold in recent weeks.
While Nvidia remains one of the central beneficiaries of rising AI spending, the stock has underperformed many semiconductor peers in 2026 as investor enthusiasm broadens across the industry.
Investors rotate across the AI supply chain
The weakness follows an extraordinary first half for semiconductor stocks.
The VanEck Semiconductor ETF gained more than 70% during the first six months of 2026, marking the strongest first-half performance since the fund launched in 2000.
However, some of the sector’s biggest winners have recently pulled back as investors locked in profits following the historic rally.
Nvidia has notably lagged much of the broader semiconductor advance despite maintaining its leadership position in graphics processing units used for artificial intelligence workloads.
Investor attention has increasingly shifted toward other segments of the AI supply chain.
Memory-chip makers have benefited from supply constraints and rising demand, while companies focused on central processing units have attracted growing interest as investors bet that next-generation agentic AI systems will require substantially greater computing resources beyond GPUs alone.
Micron has emerged as one of the biggest winners from the memory cycle, while Advanced Micro Devices and Intel have benefited from expectations that CPU demand could accelerate alongside the expansion of AI infrastructure.
The trend has left Nvidia facing a more competitive investment landscape even as demand for its products remains strong.
Nvidia launches revenue-sharing program
Separately, Nvidia announced a new initiative designed to deepen its relationships with fast-growing artificial intelligence startups.
Under the program, Nvidia will enter revenue-sharing arrangements with selected companies, allowing them to access computing resources powered by Nvidia hardware in exchange for a portion of future revenue.
The company said participating startups will receive token credits that can be used to support development and deployment of AI products.
Cloud-based AI companies, model developers, and other technology firms will share portions of their product and cloud-generated revenue with Nvidia as part of the arrangement.
The initiative further expands Nvidia’s role beyond hardware supplier and positions the company as a more active participant in the economics of the AI ecosystem.
Nvidia also identified two initial partners participating in the program.
Australia-based Sharon AI plans to deploy as many as 40,000 Nvidia graphics processors under the arrangement.
Meanwhile, Singapore-based AI infrastructure company Firmus Technologies is developing a data center in Batam, Indonesia, that is expected to scale to 360 megawatts and eventually house up to 170,000 Nvidia GPUs.
The initiative reflects the growing importance of access to computing power across the artificial intelligence industry.
As demand for advanced AI infrastructure continues to outpace supply in many areas, graphics processors have become one of the most sought-after resources for startups and model developers.
The scarcity of computing capacity has encouraged a growing number of AI companies to pursue revenue-sharing and equity-based arrangements with infrastructure providers and chipmakers as an alternative to traditional financing.
For Nvidia, the strategy creates another avenue to participate in the growth of emerging AI businesses while reinforcing demand for its hardware platform.
The post Nvidia stock continues decline: what's hurting the AI darling? appeared first on Invezz
