Investing.com -- Bank of America revised its estimate for aggregate AI capital expenditure by the six largest hyperscalers upward by 17% month-over-month to $3.6 trillion through 2028, according to an industry overview released Tuesday.

The bank's semiconductor analyst projected in May that hyperscaler capex would exceed $800 billion for fiscal year 2026, with a potential path to surpass $1 trillion in fiscal year 2027.

BofA economists characterized AI capex as a growth driver through manufacturing output, AI-related employment, data-center construction and equity wealth effects.

The bank identified four channels through which banks may gain exposure to AI capex: market footprint in data center projects, loan exposure, deposit capture, and capital markets and advisory services. The latter is expected to present greater opportunities for larger banks.

Bank of America said mid-cap banks should benefit through credit formation, particularly by supporting the physical infrastructure component of the capex expansion, including manufacturing production and data-center construction.

The National Conference of State Legislatures reported that 15 states were considering legislation to ban or impose moratoriums on new data center development as of July 2026.

The bank mapped industries from an Anthropic theoretical AI usage report against Bureau of Labor Statistics industry data to identify lending opportunities. FHN, TCBI, and ZION appeared best positioned to support customers' AI financing needs based on their commercial and industrial credit exposures to data-heavy industries.