ARK Invest says AMD chips can outperform Nvidia per dollar
Cathie Wood's ARK Invest released an analysis stating that Advanced Micro Devices Inc. chips can already be more performant than Nvidia Corp. chips per dollar on select workloads. AMD is launching its Helios rack-scale solution in the second half of this year, backed by commitments from OpenAI, Meta Platforms Inc., and Oracle Corp. ARK forecasts the total addressable market for AI systems will reach $1.5 trillion by 2030.

*this image is generated using AI for illustrative purposes only.
Advanced Micro Devices Inc. is positioning itself to challenge Nvidia Corp.'s long-standing dominance in the lucrative AI accelerator market. According to a fresh analysis released by Cathie Wood's ARK Invest, internal research reveals that AMD chips "can already be more performant than Nvidia chips" per dollar on select workloads, highlighting a massive shift in data center cost efficiency.
ARK Invest’s Director of Research, AI & Cloud, Frank Downing, highlighted that AMD is running the same playbook it previously deployed to successfully disrupt Intel Corp. After holding nearly zero server market share in 2017, AMD’s server CPU revenue share skyrocketed to 46% by the first quarter of 2026 due to consistent, power-efficient chip delivery.
Strategic Expansion and Helios Launch
AMD is now utilizing this aggressive execution strategy against Nvidia by launching Helios, a comprehensive rack-scale solution arriving in the second half of this year. ARK remains highly optimistic about Helios, pointing to high-profile customer commitments from OpenAI, Meta Platforms Inc., and Oracle Corp.
Additionally, the emergence of agentic AI is acting as a major tailwind for AMD’s core chip business. As highly capable AI agents command significantly more traditional compute resources than early AI chatbots, CPUs are becoming increasingly critical to the evolving AI ecosystem.
Market Projections and Competitive Landscape
This intense chip rivalry unfolds as global technology capital expenditure as a percent of GDP outpaces spending from the historic tech bubble era. The top four hyperscalers alone are expected to spend over $700 billion in 2026.
| Metric | Value |
|---|---|
| Hyperscaler CapEx (2026) | >$700 billion |
| AI Systems TAM (2030) | $1.5 trillion |
Looking ahead, ARK Invest forecasts that the total addressable market (TAM) for AI systems will triple by 2030, reaching a staggering $1.5 trillion in annual sales. Although Nvidia maintains a firm grip on the market and competitors like Google’s parent Alphabet Inc. and Amazon.com Inc. actively pursue custom in-house hardware projects, ARK concludes that AMD is primed for growth.
Armed with strong chip architectures and a proven design team, AMD is expected to "capture a reasonable share of the future AI compute TAM" from its small current baseline.
Stock Performance
AMD shares were up 155.29% year-to-date, up 14.98% over the last month, and higher by 295% over the year. It closed 5.67% higher at $546.72 per share on Thursday, and was down 0.12% in overnight trading.
How will the launch of Helios impact AMD's market share relative to Nvidia by the end of 2025?
What role will agentic AI play in driving demand for AMD's CPUs in the evolving AI ecosystem?
How might the $1.5 trillion AI systems TAM by 2030 influence competitive dynamics between AMD and Nvidia?
































