Meta Platforms AI Cost Breakthrough Lifts Stock After BofA Report

Bank of America analysts reviewed an internal Meta memo that suggests the company’s AI infrastructure costs are far lower than Wall Street expected. The memo puts the cost of building one gigawatt of AI compute capacity at about $22 billion, compared with earlier estimates of $45 billion per gigawatt. Meta plans to deploy roughly 6.5 gigawatts of AI compute by 2026, with most of that capacity added in the second half of the year.

The cost‑savings news prompted Meta shares to rise about 3 percent in early trading. BofA reaffirmed a buy rating and an $835 price target for the stock. In parallel, Reuters reported that Meta will begin manufacturing its custom AI accelerator chip, code‑named "Iris," in partnership with Broadcom and Taiwan Semiconductor Manufacturing later this year, a move that could further strengthen its AI roadmap through 2027.

Investors see the lower‑cost outlook as a potential relief for Meta’s $125‑$145 billion AI‑related capital spending plan, which has weighed on the stock amid fears of profit compression.