BofA Just Picked Its Two Favorite AI Chip Stocks — And AMD and Nvidia Made the Cut

TLDR

• Bank of America raised its server CPU market forecast from $110B to $125B by 2030
• AMD and Nvidia named as BofA’s top two AI chip stock picks
• CPUs are becoming the “control plane” of agentic AI inference workloads
• ARM-based custom chips from AWS, Google, and Microsoft expected to be fastest growing segment
• Nvidia reported Q1 FY2027 revenue of $81.6B, up 85% year-over-year

---

Bank of America has lifted its server CPU market forecast to $125 billion by 2030, up from a prior estimate of $110 billion, and named AMD and Nvidia as its two preferred chip stocks to benefit from the shift.

NVIDIA Corporation, NVDA

The updated outlook is driven by the rise of agentic AI — systems that plan tasks, retrieve data, manage memory, and run multi-step workflows. BofA analyst Vivek Arya, who ranks 94th out of more than 12,000 analysts tracked by TipRanks, says this type of AI is more CPU-dependent than previous generations.

The bank projects a compound annual growth rate of 31% for the server CPU market between 2026 and 2030, starting from a base of roughly $43 billion.

Why CPUs Are Becoming More Important in AI

During the early AI boom, GPUs handled most of the heavy lifting in model training. But as AI systems become more complex, CPUs are being called on to coordinate tasks, manage state, and interact with databases and retrieval systems.

BofA describes CPUs as the “control plane” of AI inference, meaning they handle the orchestration layer that keeps agentic AI running smoothly.

The bank says this is not a substitution for GPU demand. Instead, it represents an expansion of the overall data center market. New CPU-only server racks are expected to handle workloads that GPU racks could not serve cost-effectively before.

AMD and Nvidia Lead BofA’s Picks

BofA maintains a Buy rating on both stocks, with a $500 price target on AMD.

AMD is seen holding around 28% of server CPU value share by 2030, with gains expected in both cloud and enterprise segments. Nvidia is favored for its ability to combine CPUs, GPUs, networking, and memory into full AI systems.

Nvidia’s upcoming Vera CPU, set to launch with its Vera Rubin platform in the second half of 2026, is expected to work alongside GPUs at close to a one-to-one ratio in future AI server pods.

ARM-based custom chips, including AWS Graviton, Google Axion, and Microsoft Cobalt, are forecast to be the fastest-growing segment, rising from roughly 15% of server CPU value share today to around 37% by 2030. Intel is expected to lose share in both cloud and enterprise.

Nvidia’s most recent earnings backed up the CPU trend. The company reported Q1 Fiscal 2027 adjusted earnings per share of $1.87, ahead of Wall Street’s estimate of $1.75. Revenue came in at $81.6 billion, up 85% year-over-year.

Despite the strong growth outlook, server CPUs are still projected to represent only 5 to 6% of total data center spending by 2030. AI accelerators are expected to remain the dominant component, in a broader AI data center market that BofA sees reaching $1.7 trillion.