Within the semiconductor sector, three stocks—NVIDIA (NVDA), AMD (AMD), and Broadcom (AVGO)—deserve attention amid the pullback. All three companies possess significant growth opportunities, with AI data center spending expected to remain robust for many years to come.
Following NVIDIA’s share price pullback, its forward price-to-earnings ratio has dropped to 16 times (based on analysts’ earnings projections for fiscal 2028, ending January 2028), making it one of the more cost-effective names in the chip sector. The moat built by its CUDA software platform enables the company to continue dominating the AI model training market, as the vast majority of foundational AI code is written on CUDA and optimized for its GPUs.
As data center compute usage shifts toward inference and agentic AI workloads, NVIDIA is equally well-positioned in this domain. The company has transformed from a pure GPU manufacturer into a complete AI infrastructure provider, offering end-to-end servers designed for specific AI tasks. Its acquisition of Groq secured specialized inference chips, which have already been integrated into the CUDA ecosystem. Meanwhile, its networking business has become the fastest-growing segment. Given strong future growth prospects, NVIDIA deserves attention at its current discounted valuation.
AMD is currently benefiting from two major trends in AI—inference and agentic AI. Its chip offerings provide a competitive edge in the AI inference market, and it has secured large-scale GPU supply agreements with OpenAI and Meta Platforms.
Inference relies more on fast memory access than on raw compute power, which is precisely where AMD is focusing its efforts. Its chiplet design enables GPUs to pack more memory, while the acquisition of the memory optimization platform MEXT will allow it to virtually scale memory capacity with little sacrifice in performance, helping customers reduce costs.
At the same time, AMD stands to benefit from the emerging wave of agentic AI. Unlike other AI workloads that rely primarily on GPUs, agentic AI workflows require greater CPU involvement. AMD has long maintained a leading position in data center CPUs, and as AI agents become more prevalent, CPU demand is expected to grow rapidly. In fact, the ratio of GPUs to CPUs in new data center builds is projected to shrink from 8:1 in training scenarios to 1:1 for agentic AI. AMD projects that the data center CPU market will double to $120 billion by 2030, and the company is already developing CPUs purpose-built for agentic AI.
Broadcom is one of the primary beneficiaries of the trend among hyperscalers to deploy custom AI accelerators to reduce costs. The company assisted Alphabet in developing its TPU chips, and with the search giant planning to invest up to $190 billion in AI infrastructure this year, Broadcom is poised for rapid growth. Additionally, Alphabet has agreed to sell $21 billion worth of TPU products to Anthropic.
The success of TPU has attracted other hyperscalers seeking Broadcom’s help in developing custom AI chips. The company expects this business to grow to over $100 billion in scale by fiscal 2027, while Citi forecasts that its AI-related revenue could climb to $180 billion in fiscal 2028. Broadcom’s data center networking business is also growing quickly, and after signing a $30 billion agreement with Apple (AAPL), its non-AI chip business could see a turnaround as well. The stock trades at a forward P/E of just 20 times (based on fiscal 2027 projections), which appears relatively reasonable given its potential for explosive growth.
In summary, despite the recent pullback in the AI chip sector, NVIDIA, AMD, and Broadcom each possess clear growth narratives and competitive advantages within their respective niches. NVIDIA dominates the AI training market through its CUDA ecosystem, AMD has made deep inroads in inference and agentic AI, and Broadcom has secured an early lead in custom AI chips. Current valuation levels offer long-term investors a favorable entry point, and the secular growth trend in AI infrastructure spending provides support for these three companies’ performance over the coming years.
