Meta’s Infrastructure Pivot Gains Momentum: Iris Chip, Cloud Plans, and a Discrimination Lawsuit

Meta faces class-action lawsuit alleging AI tools unfairly targeted workers in layoffs, even as it invests up to $145 billion in AI chips and cloud services.

Meta Platforms is hurtling toward a future that looks nothing like its past. Over the past week, the social-media giant has laid out plans to manufacture its own artificial-intelligence chips, rent excess computing capacity to outside clients, and pour up to $145 billion into energy and hardware by 2027. But even as the stock rallies on those ambitions, a group of 26 current and former employees has filed a class-action lawsuit alleging that Meta used its own AI tools to single out workers for dismissal during the May 2026 layoffs that cut 8,000 roles.

The complaint, lodged July 14 in the U.S. District Court for the Northern District of California, accuses Meta of deploying internal systems such as the virtual assistant “Metamate” and algorithms that scored employees on productivity metrics derived from keystrokes and browser histories. Those scores, the plaintiffs contend, unfairly penalized workers who took legitimate sick leave or protected parental leave, making them more likely to be selected for termination. Meta denies the allegations, insisting that all personnel decisions are made by human managers, not automated systems.

Investors have so far shrugged off the legal threat. Meta’s shares closed at 590.10 euros on Wednesday, up 1.88 percent for the day. The stock has gained 11.80 percent over the past week and 15.30 percent over the past 30 days, pushing its 14-day Relative Strength Index to 67.3 — approaching the overbought threshold. The current price sits 13.21 percent above the 50-day moving average of 521.25 euros and 7.80 percent above the 200-day average of 547.39 euros. Still, the stock remains 12.94 percent below its 52-week high of 677.80 euros, reached in July 2025, and annualized 30-day volatility stands at 49.61 percent, underscoring the shares’ sharp swings.

Behind the rally lies a strategic recalibration that extends well beyond the core advertising business. Meta confirmed that mass production of its proprietary AI chip, codenamed “Iris,” will begin in September 2026. Developed in collaboration with Broadcom and fabricated by TSMC, Iris is part of the multiyear “Meta Training and Inference Accelerator” program. The chip is designed to make the recommendation algorithms powering Facebook and Instagram more efficient, while reducing the company’s dependence on external GPU suppliers. By the end of 2026, Meta aims to have 7 gigawatts of total computing capacity online; it added 1 gigawatt in the first half and plans another 2.5 gigawatts by December.

Alongside the hardware push, Meta is launching “Meta Compute,” an initiative to lease surplus AI computing capacity to external customers. That move pits the company directly against established cloud providers Amazon Web Services and Microsoft Azure. The longer-term roadmap calls for 14 gigawatts of capacity by 2027, supported by total capital expenditures of $125 billion to $145 billion for the full year 2026 alone, according to the primary source; a separate report puts the energy-and-hardware investment at up to $145 billion by 2027.

Institutional investors appear to be buying into the vision. Resona Asset Management, for instance, increased its stake in Meta during the recent period. Analysts also point to cooling inflation and robust demand for AI-powered recommendation systems as tailwinds. The next major test arrives on July 29, 2026, when Meta reports second-quarter earnings after the U.S. market close, followed by a conference call at 10:30 p.m. German time. Investors will be listening for how management plans to demonstrate a return on the billions being poured into chips, data centers, and cloud infrastructure — and whether the current optimism can survive the scrutiny of actual numbers.