Prediction: These AI Chip Stocks Could Soar (Hint: It's Not Nvidia or Broadcom)

While Nvidia and Broadcom currently dominate the AI chip market, AMD and Marvell Technology are strategically positioned for significant growth in the evolving AI infrastructure landscape. AMD is targeting the high-volume AI inference segment, leveraging its improved ROCm software and the UALink open standard to potentially capture market share from Nvidia, despite its significantly smaller current data center revenue. Marvell is securing multi-generational custom AI chip design wins with hyperscalers, focusing on the high-growth XPU Attach market (projected $15 billion by 2028) and aiming for 20% of its $94 billion data center TAM, with custom AI chip growth expected to reaccelerate in Q4.

Analysis

While Nvidia and Broadcom currently dominate the AI infrastructure market, Advanced Micro Devices (AMD) and Marvell Technology (MRVL) are positioned as strategic beneficiaries of the market's next phase. AMD is targeting the AI inference segment, which is projected to eventually dwarf the training market in volume. The company's competitive angle relies on offering a lower total cost of ownership, supported by its improved ROCm 7 software platform and the UALink Consortium, an open-standard alternative to Nvidia's proprietary NVLink. Given AMD's data center revenue was approximately $3 billion last quarter versus Nvidia's $40 billion, even modest market share gains in inference would be highly significant. Meanwhile, Marvell is establishing itself in the custom AI chip space, securing 18 custom compute sockets, including 12 with major U.S. hyperscalers. Despite recent stock performance headwinds due to lumpiness in its custom AI business impacting Q3, Marvell anticipates a reacceleration in Q4. The company's long-term strategy is focused on high-growth areas like the XPU Attach market, which it forecasts to grow at a 90% CAGR to $15 billion by 2028, and a broader goal of capturing 20% of its stated $94 billion data center total addressable market.

Key Decisions for Investors

• Investors seeking exposure to the next phase of AI growth beyond the current market leaders should evaluate AMD and Marvell as distinct plays on inference and custom silicon, respectively.
• For AMD, monitor adoption rates of its GPUs for inference workloads and progress from the UALink Consortium, as any market share captured from Nvidia could have a disproportionately large impact on AMD's smaller revenue base.
• For Marvell, it may be prudent to look past the reported Q3 lumpiness and focus on the expected reacceleration of custom AI chip growth in Q4 and the company's execution on its multi-generational design wins with hyperscalers.
• Consider that both companies represent a higher-risk, higher-reward scenario compared to the incumbents, with success contingent on displacing established leaders and capitalizing on specific, evolving segments of the AI market.