A six-month-old microchip startup with no product yet on the market is raising hundreds of millions of dollars at a valuation of roughly $2.5 billion, underscoring how intense investor appetite has become for the processors that keep artificial intelligence systems running, according to reports.
The company, Nuvacore, is developing a new central processor designed for data centres. CPUs perform vital functions as traffic controllers for AI graphics chips and increasingly run autonomous AI software known as agents. As demand for data-crunching capacity to power chatbots and related tools has swollen, interest in the once-unfashionable CPU has revived.
Intel and Advanced Micro Devices, which use x86 technology, have long dominated the data centre CPU market. Companies designing chips based on Arm technology, found inside custom processors at major cloud firms, have also attempted to make inroads. Nuvacore last month described an unconventional design strategy: build the core functionality of the chip before committing to a specific architecture, allowing its engineers to optimise the design for the computations most common in data centres and AI infrastructure, according to reports.
The roughly $2.5 billion valuation has not been previously reported, and the funding round has not closed, meaning the valuation and the size of the capital raise could still change. The company declined to comment.
One reason CPUs are back in fashion is the rise of AI agents: software that browses, writes code, books meetings and chains tasks together. Those workloads lean heavily on general-purpose processors rather than graphics chips alone, and cloud operators have responded by designing their own Arm-based CPUs. A startup that can offer data centre operators an alternative, without forcing them to choose an architecture first, is pitching flexibility as its edge.
For U.S. readers, the story is less about one startup than about where the AI build-out is heading. Graphics processors grab the headlines, but every large AI deployment also needs CPUs to direct traffic, run agents and keep data moving. Investors are betting that whoever designs a better traffic controller for the AI era could claim a valuable place in the data centre, even starting from a standing start.
