Technology

After six years of promises and no shipping silicon, Tachyum revises Prodigy processor specs to 1,024 cores with 1,600W of power consumption — likely another 5-year delay, company claims its chip is 20 times faster than Nvidia's Rubin NVL576 rack

Source

Tom's Hardware

Published

Nov 14, 2025

TL;DR

AI Generated

Tachyum revises the specs for its Prodigy processor, boasting 1,024 cores and 1,600W power consumption, claiming it's faster than Nvidia's Rubin NVL576 rack. The Prodigy processor is said to feature a multi-chiplet design with up to 256 custom cores per chiplet. Tachyum plans to introduce 12 Prodigy SKUs, with the Prodigy Ultimate offering 768 or 1,024 cores, up to 128 PCIe lanes, and a 24-channel memory subsystem. However, the company's history of delays and lack of concrete performance data raise skepticism about the feasibility of these claims. Tachyum's ambitious plans may face challenges in staying competitive due to potential delays, high costs, and evolving market competition.

Read Full Article

Bolt Graphics tapes out its first Zeus GPU test chip on TSMC 12nm — firm touts 17x lower cost of compute

Bolt Graphics has successfully taped out its first Zeus GPU test chip on TSMC's 12nm process, transitioning from FPGA emulation to manufactured silicon. The company claims a significant 17 times lower cost of compute with this move. The test chip will undergo customer benchmarking before a planned production increase in Q4 2027, targeting markets like HPC and graphics rendering. Bolt's Zeus GPU, initially introduced in 2025, aims to offer high performance and efficiency, with plans to utilize more advanced nodes for final production. The company's product pipeline exceeds $500 million, with significant interest from various stakeholders, though specific funding details remain undisclosed.

Tom's Hardware•

1 week ago

TSMC Technology Symposium 2026 Overview

The 2026 TSMC Technology Symposium in Silicon Valley highlighted the semiconductor industry's accelerated growth, driven by AI and HPC. TSMC projects a 30-40% revenue growth, with chip pricing playing a significant role. The industry is moving towards a $1 trillion revenue mark, with AI and HPC expected to contribute over 55% of total demand by 2030. Advanced nodes like TSMC N2U and A13 focus on enhancing power, performance, and area, while technologies like CoWoS and SoIC enable integration for improved compute density. The shift towards system-level optimization and innovations like co-packaged optics are shaping the future of semiconductor technology.

SemiWiki•

1 week ago

Intel launches Wildcat Lake as Core Series 3 for value laptops and edge systems — six consumer SKUs built on 18A promise 'all-day' battery life

Intel has unveiled its Core Series 3 mobile processors, known as Wildcat Lake, featuring six consumer SKUs and one edge-only variant, all built on 18A technology promising extended battery life. The lineup includes various configurations with P-cores, E-cores, NPUs, and Xe3 integrated GPUs, offering up to 40 platform TOPS and hybrid AI capabilities. Memory support tops out at LPDDR5x-7467 or DDR5-6400 in a single-channel setup, with performance improvements compared to previous generations. Initial laptop designs from Acer, HP, MSI, and others are set to launch, targeting students, small businesses, and edge deployments.

Tom's Hardware•

2 weeks ago

SemiEngineering

Chiplet Standards Aim For Plug-n-Play

Chiplet standards are crucial for creating a marketplace where chiplets can be easily interchanged like LEGOs. Various standards are being developed to ensure interoperability and physical composability of chiplets, including die-to-die interconnect standards like Bunch of Wires (BoW) and Universal Chiplet Interconnect Express (UCIe). These standards cover system architecture, security, power delivery, data semantics, physical placement, testing, and more. Organizations like the Open Compute Project (OCP) are leading efforts to standardize chiplet-related aspects, such as packaging descriptions and system architectures. The goal is to pave the way for a plug-and-play chiplet marketplace, although challenges related to practical and economic factors still exist.

SemiEngineering•