Key Points
-
AMD and Broadcom have reported strong revenue and earnings growth, but investors remain concerned about the durability of AI infrastructure spending.
-
Both companies are positioned to benefit from different parts of the AI market, including GPUs, CPUs, and custom silicon.
-
S&P 500 Global projects that global AI infrastructure spending will exceed $1 trillion next year.
Advanced Micro Devices (NASDAQ: AMD) and Broadcom (NASDAQ: AVGO) have experienced significant share-price volatility over the past three months. The weakness has persisted despite recent quarterly results that surpassed analysts’ expectations.
The uncertainty reflects concern that AI infrastructure investment could slow. However, both companies continue to benefit from expanding demand for advanced computing hardware.
Image source: Getty Images.
What Wall Street may be overlooking with AMD
AMD produces both GPUs and CPUs for data centers, personal computers, and other computing platforms. Although Nvidia dominates the GPU-accelerator market, AMD remains an important supplier to customers seeking alternative semiconductor technology.
In its second-quarter 2026 results, AMD’s data-center revenue increased 107% to $6.7 billion, driven primarily by GPU sales. Total revenue rose 50% to $11.5 billion, exceeding the approximately $11.3 billion consensus estimate. Non-GAAP earnings per share increased 245% to $1.66, also beating the $1.62 consensus forecast.
Despite those results, AMD’s shares declined after the earnings announcement and continued a broader multi-month selloff.
The stock currently trades at a premium, with a trailing price-to-earnings ratio of 117 compared with a technology-sector average of 32. That valuation leaves room for disappointment if growth slows. Nevertheless, investors may be placing too much emphasis on AMD’s GPU business while underestimating its long-term CPU opportunities.
AI companies are increasingly relying on CPUs to handle agentic workloads that require fast and efficient processing. This trend is changing the balance of processors installed in data centers.
AMD Chief Executive Lisa Su has said that servers once contained roughly four GPUs for every CPU, and in some cases as many as eight. She expects that relationship to move closer to one GPU for every CPU as AI agents require more general-purpose processing power. AMD management estimates that this shift could create a $220 billion market by 2030.
Other analysts are also watching the opportunity. A Raymond James analyst recently projected that the CPU market could reach $201 billion over the next four years.
AMD is preparing for higher processor demand as well. The company expects server revenue to increase 70% in 2027 as CPU sales accelerate.
AMD’s elevated valuation warrants caution, but investors focused only on its GPUs may be missing a broader opportunity. The company’s long-term prospects depend not only on AI accelerator spending but also on the growing role of CPUs in modern data centers.
Broadcom’s custom-chip opportunity
Broadcom’s application-specific integrated circuits, or ASICs, are widely used in AI data centers. Unlike general-purpose CPUs and GPUs, these chips can be tailored to run particular AI models, potentially improving performance and efficiency for specialized workloads.
That business has become a major source of growth. Broadcom’s third-quarter revenue increased 86% to $29.6 billion, while non-GAAP earnings per share rose 96% to $3.32. Both figures exceeded consensus expectations.
Broadcom’s shares have recently come under pressure as investors worry that AI spending could decelerate. Management’s outlook suggests that demand for the company’s AI products remains substantial. Broadcom expects AI-related sales to double in 2027, reaching $100 billion.
Counterpoint Research estimates that Broadcom will account for approximately 60% of the ASIC market by the end of next year. The company has also secured custom-chip agreements with OpenAI, Meta, Alphabet, and other major technology firms. Many of these arrangements extend over several years.
S&P 500 Global projects that the world’s largest AI companies will spend more than $1 trillion on infrastructure next year, compared with $750 billion this year. If that spending materializes, Broadcom’s specialized chip business could continue expanding rapidly.
The recent decline in Broadcom’s stock may therefore present an opportunity for investors with a long-term perspective. As with AMD, however, the company’s valuation and the risks associated with concentrated AI demand should remain central considerations.
Should investors buy AMD stock now?
Before purchasing AMD, investors should evaluate the company’s valuation, the pace of data-center growth, and the extent to which CPUs are becoming integral to AI workloads. Strong current results support AMD’s growth story, but they do not eliminate the risk that an expensive stock could struggle if expectations prove too ambitious.
