Nvidia is accelerating production of its Blackwell Ultra graphics processing units after reporting a near-complete sellout of its current GPU allocation through mid-2026, according to sources familiar with the company's supply chain planning. The chipmaker is working with TSMC to increase wafer starts as demand from cloud hyperscalers, enterprise AI adopters, and government-backed sovereign AI initiatives continues to overwhelm supply.
Chief Executive Jensen Huang, speaking at an industry conference this week, indicated that the company's data center segment is experiencing demand unlike anything in its history. Nvidia's Blackwell platform, which includes the B200, B100, and GB200 NVLink rack systems, has become the de facto standard for large language model training and inference workloads globally.
Nvidia shares climbed in premarket trading Thursday, extending what has already been a strong year for the stock. Analysts at multiple firms reiterated buy ratings, with several raising price targets, citing upward revisions to fiscal year 2027 revenue estimates driven by accelerating enterprise AI adoption and new sovereign AI spending from the Middle East, Asia, and Europe.
The Nvidia supply expansion ripples through the broader semiconductor ecosystem. Companies including TSMC, SK Hynix, Micron, and ASML are expected to see increased orders as Nvidia builds out its production capacity. HBM memory demand remains a critical bottleneck, and SK Hynix and Micron are both investing in additional capacity to capture the opportunity.
Beyond hardware, the AI infrastructure boom is creating tailwinds for power generation, data center construction, and cooling technology companies. Utilities with significant data center exposure and real estate investment trusts specializing in digital infrastructure have also benefited from the Nvidia-driven buildout narrative.
Competitors AMD and Intel are watching closely. AMD's MI300X accelerators have gained traction in certain inference workloads, though the company continues to trail Nvidia significantly in training workloads. Intel's Gaudi 3 accelerator has found limited adoption but remains a wildcard as enterprises seek to diversify suppliers and manage costs.