Bank of America outlined reasons it expects compute-focused semiconductor stocks to regain momentum, citing repeated, encouraging comments on AI capital spending from major cloud providers. In a note to clients, the bank distilled hyperscaler commentary into three central points that it believes underpin renewed strength in compute names.
Hyperscaler takeaways
BofA quoted analyst Vivek Arya stating: "We expect compute stocks to re-energize based on constructive AI capex commentary from hyperscalers highlighting three points." Arya also observed that "chip stocks are off to a solid start," noting the SOX index has risen about 13 percent year-to-date and delivered its second-strongest January in two decades.
Despite that broader rally, Arya highlighted that the compute leaders "NVDA and AVGO" have not yet meaningfully participated in the advance. The bank emphasized that hyperscalers repeatedly conveyed three messages: AI investments are essential to sustaining double-digit growth; sales would have been higher if not for supply constraints; and there is "no evidence of a 'bubble.'"
Workload shift and longer-term capex potential
BofA pointed to calendar year 2026 as a possible turning point for AI workloads. The note said "CY26 could mark the year when inference becomes a larger workload," and suggested that inference could eventually account for "75% of $1.2 trillion" in annual AI capital expenditure by 2030. That framing positions inference as an expanding share of total AI spend over the medium term.
Competitive positioning within compute
On individual suppliers, the bank stated that Nvidia remains "in the lead" across both training and inference. Broadcom was described as "well aligned with Google, Anthropic" and positioned for new opportunities with OpenAI, Apple and xAI. AMD was characterized as "a credible second source."
Broader ecosystem and valuation notes
BofA expects the wider AI ecosystem - spanning compute, memory and semiconductor equipment - to continue benefitting from hyperscaler capex. At the same time, the bank cautioned that optical-component stocks appear overextended, indicating uneven market dynamics across subsectors.
The bank's assessment rests entirely on hyperscaler commentary and its interpretation of workload trends and vendor positioning. The view highlights several channels through which AI capex could translate into sustained demand for compute-related semiconductors and supporting hardware.