For years, Nvidia’s dominance in artificial intelligence has been built on one thing: GPUs. The company’s graphics processors became the backbone of the AI boom, powering everything from ChatGPT-style models to massive data centers around the world. But in China, where export restrictions have severely disrupted Nvidia’s traditional business, the company appears to be changing its playbook.

According to Reuters, Nvidia has begun informing Chinese customers that its new Vera central processing units (CPUs) could be available as early as August. More importantly, Chinese firms have reportedly been told they can start placing orders now.

The move isn’t just about launching another product. It signals a strategic pivot at a time when Nvidia is trying to reclaim relevance in one of the world’s largest technology markets while simultaneously opening a new front in its battle against Intel and AMD.

Nvidia’s China Strategy Is Shifting Beyond GPUs

China was once one of Nvidia’s most important growth markets. However, escalating U.S. export controls on advanced semiconductors and Beijing’s push to develop domestic alternatives have dramatically altered the landscape.

The impact has been severe.

Nvidia CEO Jensen Huang admitted last year that the company’s market share in China had effectively fallen to zero. Shipments of Nvidia’s H200 AI chips, among the company’s most powerful processors approved for limited export, have reportedly been stalled for months.

Now, Vera may offer a path back.

Unlike Nvidia’s flagship GPUs, CPUs currently face fewer restrictions, making them a potentially more viable option for Chinese clients navigating a complicated regulatory environment. Reuters reports that some Chinese companies have already expressed interest in Vera, particularly for testing deployments outside mainland China.

The approach reflects Nvidia’s ability to adapt quickly when one door closes.

Rather than waiting for geopolitical tensions to ease, the company is building opportunities around products less vulnerable to export complications.

What Exactly Is Vera?

While Nvidia built its reputation on graphics processing, Vera represents something different.

Introduced by Jensen Huang in March, Vera is Nvidia’s first standalone CPU specifically designed for the era of agentic artificial intelligence. These AI systems don’t simply generate responses when prompted. Instead, they’re designed to perform tasks autonomously, make decisions based on objectives, and manage increasingly sophisticated workflows with minimal human intervention.

That shift matters.

The AI industry is evolving from training giant models toward inference — the process of actually putting those models to work. Whether it’s AI assistants handling customer service, enterprise software automating business processes, or intelligent systems managing operations, inference workloads are becoming the next major battleground.

Nvidia claims Vera can run up to 1.8 times faster than comparable processors from competing products.

If those performance figures hold up in real-world deployments, Vera could become one of Nvidia’s most important products outside its traditional GPU business.

At its unveiling, Huang suggested the CPU had the potential to become Nvidia’s next multibillion-dollar opportunity.

Chinese Companies Are Already Showing Interest

According to Reuters, Nvidia’s outreach has received an encouraging response.

One major Chinese cloud provider is reportedly preparing to test more than 300 servers equipped with two Vera processors each. The company intends to evaluate the systems before deciding whether to commit to larger-scale purchases.

The cautious approach isn’t surprising.

Chinese firms have spent years building software ecosystems and infrastructure around existing hardware. Migrating workloads to a completely new platform introduces questions about compatibility, integration costs, and operational efficiency.

Another source familiar with the discussions indicated that widespread adoption remains uncertain for precisely those reasons.

Testing first allows customers to assess whether Vera can deliver meaningful advantages without disrupting their existing systems.

Even so, the willingness to explore Nvidia’s latest offering suggests the company’s relationships in China haven’t disappeared entirely.

Vera Also Puts Nvidia in Direct Competition With Intel and AMD

The Chinese market isn’t the only challenge shaping Vera’s future.

Historically, Intel and AMD have dominated server CPUs through x86 architecture. Nvidia’s entry into the CPU market intensifies competition at a time when demand for AI infrastructure continues to explode.

Unlike its rivals, Vera is built using Arm-based technology.

That distinction reflects broader industry trends. Arm processors have increasingly gained traction because of their energy efficiency and scalability, particularly in cloud environments where performance-per-watt matters.

Demand is already stretching supply chains.

Reuters previously reported that Intel warned Chinese customers of server CPU delivery times extending up to six months. AMD has similarly acknowledged that demand for processors has outpaced expectations, creating constraints across the market.

The result is a rare opportunity.

As enterprises rush to expand AI capabilities, Nvidia doesn’t necessarily need to steal every customer from Intel and AMD. It simply needs to secure a meaningful share of an expanding market.

The Price Tag Shows Nvidia Is Thinking Big

Vera won’t come cheap.

According to SemiAnalysis, a single Vera processor is expected to cost well above $20,000 before volume discounts are applied. Fully configured racks containing 256 chips could reach approximately $10 million, depending on memory configurations.

Those figures underscore who Nvidia is targeting.

These aren’t products designed for small businesses experimenting with AI.

Instead, Vera is aimed squarely at hyperscalers, major cloud providers, and enterprise operators building the infrastructure expected to power the next generation of artificial intelligence.

SemiAnalysis noted that most initial production will focus on large, ready-to-deploy rack systems preferred by major customers, while simpler dual-processor server configurations are expected to scale later.

Nvidia reportedly expects Vera to generate around $20 billion in revenue by the end of its current fiscal year, which concludes in January.

That’s an ambitious target.

But then again, ambitious targets have become part of Nvidia’s identity.

Why Vera Could Define Nvidia’s Next Chapter

For much of the AI revolution, Nvidia’s story has been synonymous with GPUs.

Vera suggests the company is preparing for a future where success depends on offering entire computing ecosystems rather than individual components.

The transition won’t be seamless.

Questions remain about adoption rates, geopolitical tensions, software compatibility, and how aggressively Intel and AMD respond. Chinese firms are also under pressure to support domestic semiconductor champions wherever possible.

Yet Nvidia has repeatedly shown an ability to anticipate where the industry is heading before competitors fully adjust.

The company transformed gaming chips into AI powerhouses. It turned GPUs into some of the most sought-after hardware on the planet. Now it’s betting that CPUs designed for autonomous AI agents could become the next critical layer of computing infrastructure.

For China, Vera may represent one of the few remaining avenues to maintain ties with Nvidia despite tightening restrictions.

For Nvidia, it’s far more than that.

It’s a chance to prove that even without unrestricted access to one of its biggest markets, the company can still reinvent itself, expand into new categories, and stay at the center of the global AI race.