In late 2025 Nvidia announced a strategic licensing agreement with Groq valued at $20 billion over five years, marking one of the first major moves in what the company describes as a four-front fight to define the future AI hardware and software stack. Under the deal, Groq gains access to Nvidia’s CUDA architecture and compiler toolchain for its upcoming accelerators, while Nvidia secures rights to integrate Groq’s low-latency tensor execution cores into its own data-center GPUs. Management projects that licensing revenue will ramp from $800 million in fiscal 2026 to more than $3 billion by 2028, representing roughly 10% of Nvidia’s total revenue by that time. This partnership is designed to lock in enterprise builders—especially in hyperscale and telecom sectors—well ahead of the industry’s anticipated pivot to heterogeneous compute environments in 2026.

Nvidia’s exposure to China showed renewed strength in the third quarter, with revenue from Greater China climbing 58% year-over-year to $7.3 billion, driven by AI datacenter orders from leading hyperscalers. The company’s backlog in the region rose 22% sequentially to $4.8 billion, as firms scrambled to secure A100 and H100 GPU allocations before new export restrictions take effect. Technical analysts note that Nvidia shares have formed a consolidation base with a traditional buy point at 212.19. Management indicated that streamlined export licensing approvals have cut average shipping lead times by 15 days since Q2, and that Chinese AI startups now account for over 40% of new data-center GPU bookings. Nvidia will report fourth-quarter results on February 25, when investors will watch for further color on supply-chain resilience and regulatory headwinds in Asia.