Meta Cuts AI Infrastructure Costs by Half to $22B Per GW; BofA Raises Price Target to $835

According to an internal Meta memo reviewed by Reuters, Bank of America Securities analyst Justin Post on July 10 maintained a Buy rating and $835 price target for Meta Platforms (META), citing significantly reduced AI infrastructure construction costs.

BofA originally estimated Meta's AI computing capacity costs at $45 billion per GW, but based on Meta's disclosed 2026 deployment plans and $145 billion capital expenditure budget, actual costs have declined to approximately $22 billion per GW—roughly half the initial estimate. Meta plans to add 6.5GW of computing capacity in 2026, substantially exceeding BofA's prior forecast of 2.6GW. Additionally, Meta will begin mass production of its self-designed AI chip Iris in fall 2026, partnering with Broadcom and TSMC to reduce GPU dependence and further enhance infrastructure efficiency.