Nvidia shares were broadly flat on Thursday, but the chipmaker continues to face questions from investors about its ability to maintain its dominance in artificial intelligence hardware as the market evolves beyond graphics processing units (GPUs).
Nvidia (NVDA) shares fell 0.12% on the session to $212.35, underperforming the broader S&P 500 and Nasdaq Composite indexes, which both rose 1.5%.
Despite its status as one of the most closely watched AI companies on Wall Street, the stock has significantly underperformed the broader semiconductor sector this year.
As of Wednesday’s close, Nvidia shares had gained about 14% in 2026, lagging the Philadelphia Semiconductor Index’s 79% rise.
Part of this slower performance reflects Nvidia’s enormous size and widespread ownership by institutional and retail investors, making it increasingly difficult for strong earnings results to generate outsized stock gains.
Nvidia faces growing competition beyond GPUs
The central debate around Nvidia is whether the company’s long-standing advantage in AI hardware can endure as artificial intelligence workloads evolve.
Nvidia has built its market leadership thanks to GPUs, which have become the chips of choice for training advanced AI models.
However, investors are increasingly focusing on the possibility that future demand for AI computing may shift to a broader mix of processors, including central processing units (CPUs) and custom-designed AI chips.
To respond to this change, Nvidia recently introduced its Vera standalone CPUs.
The company estimates these products could generate up to $20 billion in revenue this year and is targeting a total addressable market of approximately $200 billion.
Investors are now closely watching how Nvidia’s CPUs will measure up against competing offerings from Intel and Advanced Micro Devices.
Early benchmark results released this week by benchmarking company Phoronix showed encouraging signs for Nvidia’s new processors, although broader commercial comparisons remain pending.
“Although the testing was hosted by Nvidia at its Santa Clara headquarters, early data from these trials indicates that Vera is highly competitive with AMD’s EPYC and Intel’s Xeon offerings, at least for the workloads Nvidia is targeting with the chip,” he said. writes Oppenheimer analyst Rick Schafer in a research note.
Full performance comparisons are not expected until Vera units are shipped to trading partners.
Taiwan investment highlights AI supply chain strategy
Separately, Nvidia also announced a significant expansion of its investment plans related to the Taiwanese semiconductor ecosystem.
Chief Executive Jensen Huang said Wednesday at an event in Taipei that Nvidia plans to spend about $150 billion a year with Taiwanese suppliers.
The announcement came during celebrations surrounding the launch of Nvidia’s new campus in Taiwan.
Huang noted that Nvidia already spends around $100 billion annually in Taiwan, home to key manufacturing partner Taiwan Semiconductor Manufacturing.
The projected increase represents a dramatic rise from the estimates of $10 billion to $15 billion that Nvidia was spending annually in Taiwan about five years ago.
Taiwan remains at the heart of the global semiconductor supply chain for AI due to its dominance in advanced chip manufacturing.
Nvidia’s increased spending commitments also signal continued confidence in demand for AI infrastructure despite broader investor concerns over supply chains and geopolitical risks involving Taiwan.
As computing for AI expands across cloud services, enterprise software and consumer applications, Nvidia remains one of the most influential companies in the market, even as investors increasingly examine how long its competitive advantage can persist.
