Tachyum Secures $220M to Power the Next AI Chip Race
Tachyum, a semiconductor company developing processors for AI and data centers, has secured $220 million in Series C funding from a European investor—alongside a $500 million purchase order for its flagship Prodigy® chips . The investment positions the company to scale production and compete at the highest level of the AI hardware race.
Founded by Dr. Radoslav DanilakDealroom has a profile for this one. Try Dealroom → , Tachyum has raised more than $300 million to date. Its Prodigy Universal Processor is designed to replace multiple types of chips—combining the functions of CPUs, GPUs, and AI accelerators into a single architecture. Each Prodigy chiplet features 256 custom 64-bit cores, delivering what the company claims is three times the performance of top x86 processors and six times that of current GPUs for high-performance computing. By eliminating the need for separate AI hardware, Tachyum says it can dramatically reduce the cost and power consumption of large-scale data centers.
The Series C round will fund the final tape-out of the Prodigy chip, marking the final stage before mass production. Tachyum plans to disclose updated specifications soon, aiming to make Prodigy chips central to next-generation AI data centers capable of supporting models with trillions of parameters. The lead investor behind this round previously took two U.S. semiconductor companies public—one via IPO and another through a multibillion-dollar SPAC deal.
The new funding also arrives amid a global race to expand AI infrastructure. Projects like Stargate in the U.S. and UAE and large-scale initiatives in Saudi Arabia, Asia, and Europe are driving massive demand for efficient chips capable of powering enormous AI workloads. Tachyum’s participation in the EU’s Important Project of Common European Interest (IPCEI) program further ties its technology to Europe’s broader effort to strengthen domestic AI and semiconductor capabilities.
With this latest investment, Tachyum aims to move closer to a potential IPO by 2027 , betting that its universal chip design can redefine how data centers handle AI, cloud, and high-performance computing—reducing costs while pushing the limits of what large AI models can achieve.
Source:
A.M.
