According to the latest report from Counterpoint Research, the AI ASIC segment—comprising non-GPU server AI Chips—is set to experience rapid growth in the near future. By 2027, shipment volumes are projected to triple compared to 2024 levels. By 2028, AI ASIC shipments are expected to surpass those of GPUs, reaching over 15 million units.

On January 13, the U.S. Department of Commerce announced new regulations stating that applications to export artificial intelligence chips to China will be subject to a case-by-case review, with specific licensing requirements established for companies including NVIDIA and AMD. The key provisions include the following points:

1. Companies must demonstrate that producing chips for Chinese clients does not compromise the capacity originally available for manufacturing chips for U.S. domestic buyers, and ensure these chips are not used for military purposes.

2. The volume of chips exported to China will be capped at 50% of the total quantity produced for the U.S. market.

3. Companies exporting such AI Chips must implement strict customer due diligence procedures to prevent unauthorized use of the technology. Additionally, these chips are required to undergo third-party performance testing within the United States.

NVIDIA and pharmaceutical giant Eli Lilly announced a joint investment of approximately $1 billion to establish a co-innovation AI lab. The investment, to be deployed over five years, will support the construction of a joint research facility in the San Francisco Bay Area. The lab will leverage NVIDIA's latest Vera Rubin architecture AI Chips to advance cutting-edge biomedical research and reinvent drug discovery processes in the age of AI. Kimberly Powell, NVIDIA's Vice President of Healthcare, stated that both companies are committing incremental resources to the new facility. The specific location of the lab is expected to be announced in March 2026, at which point researchers from NVIDIA and Lilly will work side-by-side to generate new datasets for training biotechnology AI models.