这颗芯片，又双叒叕延期了

Core Insights - Tachyum announced an increase in the core count of its Prodigy processor for AI and HPC applications, now featuring 256 cores per chip, up from 192 and initially 128 cores, promising performance three times that of the highest-performing x86 processors and six times that of top HPC GPGPUs [3][4] - The company completed a Series C funding round totaling $220 million from a European investor, along with a $500 million purchase order for Prodigy processors [1][3] - Prodigy processors have not yet been taped out, and final specifications are still undetermined, indicating that mass production is still years away [1][3] Funding and Production Timeline - The funds from the Series C round will be used to complete the tape-out process for the Prodigy chip, with expectations to finalize RTL and physical design within a month after funding is received [5][6] - The first batch of silicon is expected to be obtained within 4 to 4.5 months after the GDSII files are sent to manufacturing partners, potentially leading to silicon availability by February or March 2026 [5][6] - If the production timeline proceeds smoothly, commercial shipments of the Prodigy processor could begin in mid-2027, aligning with a potential IPO for Tachyum [6][7] Development Delays - The development of the Prodigy processor has faced multiple delays, initially planned for tape-out in 2019 and market release in 2020, with subsequent postponements to 2021, 2022, 2023, and now 2024 [7][9] - The company remains optimistic despite these setbacks, emphasizing the competitive landscape of AI and the potential for a disruptive chip at a fraction of the cost of existing solutions [7][9] Technical Specifications - The Prodigy architecture integrates a new microarchitecture designed for both general computing and highly parallel AI and HPC tasks, featuring a rich set of vector and matrix instructions [9][10] - The instruction set architecture (ISA) combines elements of RISC and CISC designs, standardizing instructions to 32 or 64 bits to enhance performance [10] - Prodigy includes built-in performance counters for real-time monitoring and analysis, aiding developers in optimizing code for high-demand computing tasks [10][11]