Senators question Meta's tax credit use

Seven Democratic senators have written to Mark Zuckerberg, questioning how Meta is using $6.6 billion in federal tax credits intended for research. The inquiry follows a New York Times investigation that suggests the company may be misusing these credits for its AI data centers.

Meta classifies data centers as pilot models

According to the New York Times, Meta has labeled its AI-focused data centers as "pilot models" for tax purposes. This classification allows the company to claim research tax credits, which are meant to support experimental and innovative activities, not routine commercial operations.

Meta's tax savings from the credits

The New York Times reports that Meta is using these credits to reduce its tax burden significantly. The company is projected to save $3.9 billion in 2025, $700 million in 2023, and $2 billion in 2024 by routing purchases of AI chips, mainly from Nvidia, through this tax relief mechanism.

Meta acknowledges potential tax disputes

Meta has acknowledged the risk that the IRS might challenge its use of these tax credits. The company has disclosed an increase in its "unrecognized tax benefits" (UTB) from $12.9 billion in 2024 to $18.74 billion in 2026, indicating growing uncertainty around its tax strategy.

Senators warn against misuse of tax credits

The senators emphasized that research tax credits are not meant for technologies already available on the market or for routine commercial operations. They expressed concern that Meta’s use of the credits for AI data centers does not align with the intended purpose of the tax incentive.

Meta defends its approach as experimental

Meta argues that its data centers are still in a testing phase, allowing the company to experiment with different server rack configurations and equipment. However, the IRS has previously rejected similar claims for proven technologies, stating that the tax credits should support innovation, not the construction of AI infrastructure.