Meta is reportedly exploring a major partnership with Google to power its fast-growing AI ambitions. According to multiple reports, the company is in advanced discussions to use Google’s custom AI chips—Tensor Processing Units (TPUs)—with the first phase potentially kicking off as early as next year.

If the deal moves forward, Meta would begin renting TPUs through Google Cloud in 2026, before deploying the chips in its own data centres by 2027. Such a move would mark a notable shift in Meta’s AI infrastructure and could challenge Nvidia’s dominance in the market, a space it has led for years under CEO Jensen Huang.

Also Read| Accel, Google Launch Atoms AI Cohort 2026 to Back India’s Next Wave of AI Startups

The timing has caught industry attention. AI developers around the world are facing escalating costs and long wait times for Nvidia hardware, pushing major companies to consider alternatives. Google has been positioning its TPUs as a more cost-effective and readily available option that can support the training of large-scale AI systems without supply constraints.

For Meta, this flexibility matters. The company trains vast AI models across its ecosystem—Facebook, Instagram, WhatsApp and its expanding line of AI assistants. Access to TPUs could help lower expenses and diversify its hardware strategy, reducing reliance on a single chip supplier.

Reports suggest Meta would use the rental period to rigorously test TPU performance before committing to large-scale deployment in 2027. Within Google, the potential deal is being viewed as an opportunity to meaningfully grow the TPU footprint and capture part of the spending that now flows largely to Nvidia.

Also Read| India–EU Payments Boost: RBI and ECB Begin Work on UPI–TIPS Cross-Border Link

The market reaction was immediate. Nvidia shares fell about 2.7% following the reports, as investors weighed the possibility of Meta redirecting future chip orders. Alphabet stock, meanwhile, climbed by a similar margin, reflecting optimism around Google’s cloud and AI trajectory.

None of the companies involved—Meta, Google or Nvidia—have issued public statements so far. But the talks alone underscore how rapidly the competitive landscape for AI chips is evolving, and how the next wave of partnerships could reshape where the industry places its biggest bets.