Is cutting AI spending more beneficial? Bank of America offers a counterintuitive judgment: though "unpopular," it may extend the boom cycle, and chip stocks could enjoy a longer earnings runway.

Title context: Is cutting AI spending more beneficial? Bank of America offers a counterintuitive judgment: though "unpopular," it may extend the boom cycle, and chip stocks could enjoy a longer earnings runway. Text: Concerns have risen recently, stemming from the U.S. midterm elections, worries about rising AI infrastructure costs, and doubts about the sustainability of this spending boom. Bank of America Corp said that while any form of spending cut could be an "unpopular" signal for investors, it may actually extend the spending cycle. The bank's analysts offered a contrarian argument: if monetizable AI demand remains strong while capacity growth slows, this could actually increase compute supply capacity and delay the peak of the AI cycle. "Slower supply growth combined with resilient demand could push up utilization, support pricing and reduce the risk of overbuilding," the bank's analysts wrote in a note to clients. "The result could be lower peak earnings but a longer earnings runway, and potentially higher valuation multiples for chipmakers that have already been priced for 'the cycle is about to end.'" The bank remains bullish on NVIDIA Corporation (NVDA.US), AMD (AMD.US), Broadcom Inc. (AVGO.US), Marvell Technology, Inc. (MRVL.US), Credo Technology (CRDO.US), Lam Research (LRCX.US), KLA (KLAC.US), Micron Technology, Inc. (MU.US), Analog Devices, Inc. (ADI.US) and Texas Instruments Incorporated (TXN.US). Going further, the bank said recent earnings reports from hyperscalers highlighted the "resilience" of AI demand their combined long-term contract backlog has reached US$2.5 trillion. In addition, Amazon.com, Inc. (AMZN.US) AWS, Microsoft Corporation (MSFT.US) Azure and Alphabet Inc. Class C (GOOGL.US) cloud platform grew 39%, 49% and 100%, respectively. With capacity constrained, token processing volume rose more than 10-fold year over year, and rental prices for Hopper and Blackwell rose 33% and 51%, respectively, during the year (SpaceX recently said current spot pricing supports a payback period of about one year). Other notable data points include: Fortune 500 companies' spending on Amazon.com, Inc. Bedrock rose more than 7-fold year over year; Microsoft Corporation's number of customers with more than 50,000 Copilot seats grew 8-fold (Copilot daily active users rose 50% quarter over quarter). Investors have recently expressed concerns about cash flow conditions involving the free cash flow of eight U.S. hyperscalers and emerging cloud vendors. Free cash flow consumption is expected to widen from US$128 billion in 2026 to US$335 billion in 2027. But analysts believe investors may be "missing the forest for the trees." "Calendar 2027 may mark the peak of cash burn, and by calendar 2028, the deficit will narrow to US$77 billion, equivalent to 2% of sales, an improvement of US$258 billion," the analysts added. "Financing remains the core debate, but the ecosystem's free cash flow offers another potential release valve: NVIDIA Corporation, Broadcom Inc. and Micron are expected to generate combined free cash flow of about US$577 billion in calendar 2027 and more than US$800 billion in calendar 2028 (though still further exposed to vendor financing and circular transaction risks)." "As a result, the market appears to have priced in capex risk before cracks appear in demand data," the analysts continued. "In our view, if paid AI demand remains resilient, lower capex can reduce overbuilding, preserve scarcity, extend the EPS cycle and support higher valuation multiples."