Nvidia's Strategic Shift: CFO Forecasts $20B CPU Revenue

What is Nvidia's New Strategy for the Processor Market?

Nvidia has long been recognized as the dominant force in graphics processing units, powering the global surge in artificial intelligence. However, the company is now aggressively pivoting to conquer a different sector of the silicon landscape: central processing units. During its first-quarter earnings call for fiscal year 2027, Chief Financial Officer Colette Kress announced that Nvidia has visibility to nearly $20 billion in total CPU revenue this year. This ambitious target sets the stage for the company to potentially become the world's leading supplier of CPUs.

While Nvidia is not entirely new to processor design, its previous efforts were largely integrated into specialized GPU systems. The company announced its first Arm-based datacenter chip, codenamed Grace, in 2021. For years, these processors were primarily bundled with GPUs and deployed almost exclusively within AI datacenters and supercomputing environments. They served as accelerators for specific high-performance tasks rather than standalone general-purpose processors.

The strategic shift began to materialize earlier this year. In February, Nvidia revealed that Meta was among the first hyperscalers to deploy standalone Grace CPU Superchips in its datacenters. These chips are now powering a variety of workloads beyond traditional AI training, including the social network's complex AI agents. This move signaled a broader intent to decouple the processor from the graphics unit and offer it as a distinct commodity for enterprise infrastructure.

At its GTC conference in March, Nvidia officially expanded its CPU lineup with the introduction of the standalone Vera CPU system. This new architecture represents a significant leap forward in design philosophy and capability. Each chip features 88 custom Olympus Arm cores, engineered to handle intensive computational loads efficiently. The inclusion of simultaneous multi-threading support allows for greater parallel processing capabilities, enhancing performance without requiring proportional increases in power consumption.

Nvidia has also integrated confidential computing capabilities into the Vera architecture, addressing growing security concerns among hyperscalers and enterprise clients who handle sensitive data. By equipping each chip with up to 1.5 terabytes of LPDDR5x SOCAMM memory, Nvidia ensures high memory bandwidth at speeds reaching 1.2 terabytes per second. This memory technology is known for its efficiency and low power usage, traits typically associated with mobile devices but now adapted for massive datacenter scales.

How Does the Vera Architecture Compare to Traditional x86 Processors?

The performance metrics claimed by Nvidia regarding the Vera chip are substantial. Colette Kress stated that Vera will deliver up to 1.5 times faster performance per core compared to existing x86-based alternatives. Furthermore, the architecture offers two times better performance per watt and four times greater density per rack. These figures suggest a compelling value proposition for datacenter operators looking to optimize their infrastructure costs while maximizing computational throughput.

Nvidia's reference designs pack up to two Vera CPUs onto a single board, utilizing high-speed NVLink interconnects to facilitate rapid communication between components. In the company's most powerful rack-scale AI compute platforms, the Vera CPU is paired with Rubin GPUs in a specific 2:1 ratio. This configuration highlights Nvidia's holistic approach to computing, where processors and accelerators are designed to work in tandem rather than as isolated units.

Since the detailed specifications of the Vera chip were released this spring, industry adoption appears to be accelerating rapidly. Kress claims that nearly every major hyperscaler and system builder plans to deploy these chips. This week, several top AI labs and infrastructure providers, including Anthropic, OpenAI, Oracle, and SpaceX, took delivery of Nvidia's first Vera-based systems. The rapid uptake indicates strong confidence in the technology among the most demanding users in the industry.

However, it is important to contextualize these claims. While Nvidia is expanding its addressable market to include standalone CPUs, much like its Ethernet networking products, these chips are designed primarily with AI and high-performance computing applications in mind. They cannot yet replace x86 processors in every application across the broader enterprise landscape. The focus remains on specialized workloads where Arm-based efficiency and density provide a clear advantage over traditional architectures.

Why Does This Expansion Matter for the Total Addressable Market?

The introduction of Vera opens a brand new $200 billion total addressable market for Nvidia, a sector the company has never directly addressed before. By targeting this vast pool of potential revenue, Nvidia aims to diversify its income streams beyond GPU sales. This expansion is critical as the company seeks to sustain growth in an increasingly competitive hardware landscape.

The financial context surrounding this announcement underscores Nvidia's current market dominance. The GPU giant reported $58.3 billion in profits on $81.6 billion in revenue for the first quarter of its 2027 fiscal year. This revenue figure represents an 85 percent growth year-over-year and a 20 percent increase from the prior quarter. Such numbers reflect the intense demand for AI infrastructure that has characterized recent years.

Colette Kress attributed this sequential jump in revenue to an inflection in inference demand. As AI models move from training phases to deployment, the need for efficient inference processing grows exponentially. The Vera CPU is positioned to meet this specific need by offering high density and power efficiency, making it suitable for the massive scale of inference workloads.

The company has also reorganized its business units to reflect these shifting priorities. Nvidia now groups its revenues into a datacenter group, which includes cloud, hyperscale, neocloud, and enterprise sales. An edge group serves as a catchall for gaming, robotics, automotive, and vRAN products. This structural change highlights the centrality of datacenter operations in Nvidia's business model.

Datacenter revenues accounted for the vast majority of the quarter's income, totaling $75.2 billion. Of this amount, $38 billion came from hyperscaler and public cloud customers. The remaining $37 billion was generated by neocloud, industrial, and enterprise customers. Edge sales contributed a mere $6.4 billion, with demand for Blackwell-based workstation gear cited as a key driver in that segment.

What Are the Implications for Global Supply Chains?

Nvidia's forward-looking guidance provides insight into its expectations for the coming quarter. The company forecasts revenue to hit $91 billion plus or minus two percent for the second quarter. This prediction assumes no datacenter sales in China, reflecting ongoing geopolitical tensions and regulatory hurdles that impact semiconductor exports.

Nvidia has been trying for months to reignite its GPU business in the Middle Kingdom since receiving approval from the Trump administration in December. The company was granted permission to sell its aging H200 processors to Chinese customers for the first time ever. Despite receiving billions of dollars worth of orders, shipments remain stuck in Beijing's red tape.

This delay highlights the fragility of global supply chains and the impact of political decisions on technology markets. While Nvidia continues to expand its product portfolio with innovations like Vera, external factors can significantly influence revenue realization. The company must navigate these complexities while maintaining its technological edge and meeting the demands of its global customer base.

The expansion into CPUs also positions Nvidia against established competitors in the processor market. Companies like Intel and AMD have long dominated x86 architectures, but Nvidia's Arm-based approach offers a different path to efficiency. As hyperscalers seek to optimize their energy consumption and computational density, Nvidia's Vera chip presents a viable alternative that could reshape industry standards.

For investors and industry observers, the $20 billion CPU revenue target serves as a benchmark for Nvidia's success in this new domain. Achieving this goal would require not only technological superiority but also widespread adoption across diverse computing environments. The early deliveries to major AI labs suggest that the foundation is being laid, but the scale of execution remains to be seen.

How Will Nvidia Balance GPU and CPU Development?

Nvidia's dual focus on GPUs and CPUs requires careful resource allocation and strategic planning. The company must ensure that its CPU innovations complement rather than cannibalize its core GPU business. By designing Vera to work in tandem with Rubin GPUs, Nvidia creates a synergistic ecosystem that enhances overall system performance.

This integrated approach allows Nvidia to offer complete solutions to hyperscalers and enterprises. Instead of sourcing processors from one vendor and accelerators from another, customers can rely on Nvidia for both components. This simplification of the supply chain can reduce integration costs and improve reliability, making Nvidia's platforms more attractive to large-scale operators.

As the company continues to refine its Vera architecture, it will likely face challenges in scaling production and managing costs. The complexity of designing 88 custom cores with high-density memory support requires significant engineering resources. Nvidia must maintain its pace of innovation while ensuring that manufacturing capabilities keep up with demand.

The broader implications of Nvidia's CPU expansion extend beyond the company itself. It signals a shift in the semiconductor industry toward specialized architectures tailored for specific workloads. As AI and HPC applications become more prevalent, the demand for efficient, high-density processors will continue to grow. Nvidia's move into this space positions it at the forefront of this evolving landscape.

For now, the focus remains on delivering Vera-based systems to early adopters and gathering feedback to refine future iterations. The success of this initiative will depend on Nvidia's ability to demonstrate tangible benefits over existing solutions in real-world deployments. As more hyperscalers integrate these chips into their infrastructure, the industry will watch closely to see if the projected revenue targets are met.

The journey toward becoming the world's leading CPU supplier is a long-term endeavor. It requires sustained investment in research and development, strategic partnerships with key customers, and navigation of complex regulatory environments. Nvidia has laid the groundwork with its Vera architecture and early adoption milestones, but the ultimate outcome remains uncertain. The company's ability to execute on its vision will determine whether it can successfully conquer this new market segment.

As Nvidia continues to push boundaries in both graphics and central processing, the lines between traditional computing roles may blur further. The future of datacenter infrastructure could see a convergence of specialized processors working together seamlessly, driven by companies like Nvidia that prioritize efficiency and performance above all else. The $20 billion revenue target is just the beginning of this ambitious transformation.