Nvidia preps to sell its Vera CPUs into China as its GPU sales stay frozen — customers encouraged to place orders for CPU shipments as early as August
Nvidia preps to sell Vera CPUs into China as early as August, while H200 GPU shipments remain frozen amid dual U.S.-China regulatory gridlock.
Nvidia Pushes Vera CPUs Into China While GPU Sales Remain Blocked
Nvidia has informed Chinese clients that its Arm-based Vera server CPUs could ship as early as August 2026, with orders open now, according to Reuters sources. The move comes as H200 AI GPU shipments to China remain completely frozen — months after CEO Jensen Huang acknowledged Nvidia's market share in the country had effectively fallen to zero.
The August timeline aligns with Nvidia's previously stated roadmap. At GTC Taipei during Computex, the company confirmed Vera systems would reach customers through system builders and cloud partners starting in fall 2026. Fast-tracking Chinese buyers during a global server CPU shortage suggests they are receiving allocation priority for a product line Nvidia expects to generate $20 billion in revenue by the end of its fiscal year in late January 2027.
Chinese cloud companies are already testing more than 300 Vera servers, according to the report. At least one major cloud provider plans to place an order, though initial deployments will be restricted to those companies' overseas data centers due to regulatory considerations.
A Tale of Two Markets: CPUs vs. GPUs
If this goes ahead, Vera will reach Chinese buyers where Nvidia's GPUs cannot. Server CPUs face significantly lighter U.S. export restrictions than the high-performance AI accelerators that underpin Nvidia's data center business. However, Washington is no longer the only obstacle. The U.S. government licensed roughly ten Chinese firms to purchase the H200 GPU, yet not a single unit has been delivered — because Chinese officials, intent on nurturing domestic chipmakers like Huawei and SMIC, withheld approval on their side.
This dual regulatory pressure helps explain why Vera CPU deployments will initially be restricted to overseas data centers: Chinese cloud providers want the hardware, but placing U.S. silicon into domestic data centers invites scrutiny from Beijing officials. The strategy mirrors how some Chinese tech giants have deployed AI infrastructure in Singapore and Southeast Asia to sidestep both sets of restrictions.
Vera: From Superchip Component to Standalone Powerhouse
Vera began life as the CPU half of the Vera Rubin superchip, first shown at GTC 2025 alongside the Rubin GPU architecture. Nvidia broke it out as a standalone product at GTC San Jose in March 2026, launching it alongside a rack-scale design that packs 256 liquid-cooled Vera CPUs sustaining more than 22,500 concurrent CPU environments. At Computex weeks later, Nvidia confirmed the chip had entered full production, claiming 1.8x faster task completion than x86 processors on agentic AI workloads.
The chip features 88 custom Arm cores designed specifically for server workloads, with a focus on the high-throughput, low-latency processing demands of modern AI inference pipelines. Its predecessor, the Grace CPU, has shipped nearly 2.5 million units to date, establishing Nvidia's credibility in the CPU space — a market traditionally dominated by Intel and AMD. Nvidia's ability to leverage its GPU ecosystem and CUDA software stack gives Vera a unique integration advantage in AI-optimized data centers.
Global CPU Crunch Intensifies
The broader server CPU market is under severe strain. The shift of AI workloads from training toward inference and agentic execution is driving unprecedented demand for host processors. While GPUs dominated the training phase, agentic AI — systems that autonomously plan and execute tasks — relies heavily on CPUs for tool calls, code execution, and data handling. This architectural demand shift has caught the industry flat-footed.
Intel has quoted Chinese customers lead times of up to six months, while AMD has warned that the global CPU market is tight, with demand outpacing forecasts and supply constraints expected to persist well into 2027. Nvidia's decision to prioritize Vera allocations against this backdrop signals confidence in its in-house production capacity and its strategic bet on the Chinese market as a long-term growth engine.
Geopolitical Implications
The Vera CPU story marks a critical inflection point in the ongoing semiconductor cold war — where product capability, export controls, and geopolitical maneuvering converge. China is the world's largest semiconductor market, and any avenue for U.S. chipmakers to access it carries enormous financial and strategic weight.
Nvidia's approach with Vera represents a nuanced strategy: ship products that comply with U.S. export frameworks while being valuable enough that Chinese regulators cannot easily reject them. The restriction to overseas data centers for initial deployments is a compromise that allows both sides to save face while keeping commerce flowing.
Whether Nvidia can navigate both U.S. and Chinese regulatory hurdles will determine whether Vera becomes a genuine bridgehead into China's data center infrastructure — or simply another chapter in the chip industry's growing fragmentation along geopolitical lines.
