
According to TrendForce, NVIDIA is expected to see a notable shift in its high-end AI GPU shipment mix in 2026, driven by geopolitical uncertainties and ongoing supply chain adjustments. The share of Hopper and Rubin architectures is projected to decline, while the newer Blackwell platform is forecast to expand its dominance, increasing its shipment share from 61% to 71%.
TrendForce notes that robust demand for AI computing, combined with NVIDIA’s strategic push for highly integrated GB/VR rack-scale solutions with higher chip density, will continue to drive overall shipment growth. High-end GPU shipments are still expected to grow significantly in 2026, although the annual growth rate has been slightly revised downward from 26.8% to approximately 26%.
The adjustment is primarily attributed to potential delays in the Rubin series. Challenges include extended validation timelines for next-generation HBM4 memory, as well as the need to address system-level complexities such as network upgrades from CX8 to CX9, increased power consumption, advanced power management requirements, and the integration of more sophisticated liquid cooling solutions. As a result, Rubin’s share of high-end GPU shipments is expected to decrease from 29% to 22%.
Meanwhile, geopolitical factors are also impacting the Hopper series, with its shipment share projected to decline from 10% to 7%.
In contrast, the more mature Blackwell architecture is set to surpass a 70% shipment share, led by the GB300/B300 series. Although GB200/B200 shipments are expected to be relatively limited, continued fulfillment of 2025 orders and cost-sensitive demand segments may support production through the second half of 2026.
Beyond AI training, NVIDIA is also accelerating its expansion into AI inference applications. Demand for its new LPU-based solutions is projected to reach several hundred thousand units in 2026, with expectations to double in 2027.
To address mid-range, entry-level, and edge AI markets, NVIDIA is advancing products such as the RTX PRO 4500 and 6000 series. These offerings are expected to increase the share of mid- to low-end products to over 32% of total shipments in 2026. Amid generational transitions and a complex global landscape, NVIDIA continues to strengthen its leadership in AI computing through diversified product strategies and expanding application ecosystems.