California-based semiconductor startup Nuvacore is seeking hundreds of millions of dollars to develop a central processing unit (CPU) designed for data centers, according to a Reuters report citing people familiar with the matter. The fundraising effort could value the company at approximately $2.5 billion, marking a substantial valuation for a business that has yet to bring a product to market. The round remains open, and its final size and valuation could change as discussions progress. The Information previously reported that Nuvacore was seeking to raise $200 million or more.

Nuvacore Targets Growing Demand for AI Processors

Nuvacore is entering the market as AI infrastructure expands beyond graphics processing units (GPUs) into the broader computing systems that support data-intensive applications. CPUs coordinate data movement across servers running Nvidia’s AI chips and execute autonomous AI software, commonly known as agents. The growing use of chatbots, including Anthropic’s Claude, also underscores the importance of the processing infrastructure behind increasingly sophisticated AI services. Consequently, Nuvacore is positioning its proposed processor around the computing requirements that emerge as AI workloads become more complex.

The data center CPU market has traditionally relied on Intel and Advanced Micro Devices (AMD), whose processors use the x86 architecture. Arm-based designs have also entered the market through custom processors developed for companies such as Amazon and Microsoft, although they have secured a smaller share. Nuvacore is pursuing a different design approach by developing the essential features of its processor before committing to an established architecture such as x86 or Arm. The strategy aims to give engineers greater flexibility to build a chip around the demands of modern data center and AI workloads rather than adapting those requirements to existing architectural constraints.

Startup Pursues Workload-Specific Chip Design

Last month, Nuvacore outlined its approach as an effort to help engineers bypass the conventional limitations associated with existing processor architectures. The company said this could produce a design optimized for the computations most frequently required by data center infrastructure and AI applications. Such a focus could matter as operators balance processing performance, server efficiency and the demands of increasingly heterogeneous computing environments. However, the commercial value of the approach will ultimately depend on whether Nuvacore can translate its design philosophy into a competitive product and establish demand among infrastructure customers.

Nuvacore was founded by engineers John Bruno, Gerard Williams and Ram Srinivasan, bringing experienced technical leadership to its attempt to enter a market dominated by established chipmakers. The company completed an initial seed funding round earlier this year, led by Sequoia Capital. Its latest fundraising effort reflects renewed investor interest in semiconductor startups as demand for AI computing infrastructure reshapes capital allocation across the technology sector. Large venture capital investments are returning to hardware businesses after years in which software companies attracted much of the industry’s attention.

Intel, AMD and Nvidia Signal CPU Market Momentum

Established chipmakers have reported stronger demand for CPUs as AI infrastructure expands and operators require more processing capacity alongside specialized accelerators. Intel said this year that it could not manufacture enough chips to meet customer demand, highlighting supply constraints within the existing market. AMD has also cited rising CPU sales, while Nvidia introduced its Vera CPU and expects $20 billion in related revenue for the fiscal year ending in January. These developments provide a significant backdrop for Nuvacore’s fundraising ambitions, although they do not guarantee that a new entrant can capture meaningful market share.

Nuvacore’s proposed $2.5 billion valuation reflects investor interest in the possibility of a new approach to data center computing, but the company still needs to move from chip design to commercial execution. Its architecture-first flexibility could offer a route to differentiation if engineers deliver measurable advantages for widely used workloads. Yet established competitors benefit from existing products, customer relationships and mature semiconductor ecosystems, creating substantial barriers for an unproven startup. The next test will be whether Nuvacore can turn its technical strategy and fresh capital into a processor that data center operators have compelling reasons to adopt.