Tachyum

Tachyum offers a versatile processor, “Prodigy,” for data centers, which can run general-purpose processing, high-performance computing (HPC), deep machine learning (ML), and a range of AI disciplines through a single chip.

Its Prodigy processor primarily caters to hyperscale data centers, enabling them to become low-cost, energy-efficient computing centers. Tachyum claims its processors offer improved performance by delivering around 10x higher AI neural network training/inferencing, at a relatively low cost. It believes its chip can reduce the data center cost of ownership by 3x, relative to industry alternatives, by reducing power consumption by 10x. The chips cater to several end markets including hyperscale, OEM, telco, private cloud, and governments.

In 2023, Tachyum completed the development of the software stack for Prodigy, allowing alpha testing to begin. Hardware achievements include successful High-Performance LINPACK testing on the Prodigy FPGA and running various operating systems and applications on the FPGA prototype.

Key customers and partnerships

The company has integration with open source programming environments, Tensorflow, and PyTorch. This increases the usability of Prodigy since programmers need not rely on harder-to-learn programming environments such as CUDA.

Tachyum also expanded its presence in multiple consortiums (GAIA-X and I4DI) and also partnered with several universities around the world to collaborate on supercomputing projects, scientific research, and innovations in AI and Big Data.

The company’s other partners include National Supercomputing Centre, InoCloud, and Gaia-X.

Funding and financials

In June 2023, Tachyum, secured USD 28.5 million in funding as a grant from the European Commission, to accelerate the delivery of its Prodigy 2 Universal Processor, which it plans to mass produce in 2023.