Tenstorrent partners with Japan’s LSTC to create 2nm AI accelerator chips

Key Points:

  • Tenstorrent and LSTC partnership to build cutting-edge AI Accelerator
  • Collaboration with Rapidus Corp. for logic semiconductor technologies
  • Formation of high-performance compute design center in Japan by Tenstorrent


Tenstorrent, an AI chip company led by industry veteran Jim Keller, has entered into a significant partnership with Japan’s Leading-edge Semiconductor Technology Center (LSTC) to develop a cutting-edge two-nanometer AI Accelerator. This collaboration aims to revolutionize AI performance by combining Tenstorrent’s expertise in RISC-V CPU technology and chiplet technology with LSTC’s edge AI accelerator vision. Through a multi-tiered partnership, Tenstorrent will not only provide intellectual property licensing but also engage in co-designing the chip, emphasizing the future of silicon and heterogeneous compute integration.

The partnership expands to include Rapidus Corp., a new Japanese semiconductor company specializing in logic semiconductor technologies. Together, Tenstorrent and Rapidus will collaborate on wafer processing and advanced packaging, aligning with a shared goal of optimizing manufacturing processes. Tenstorrent will leverage its advanced RISC-V CPU core technology to develop a RISC-V CPU chiplet for LSTC’s AI accelerator.

Jim Keller’s leadership at Tenstorrent, known for its successful silicon products, was a key factor in securing this partnership. Keller, who has assembled a team of high-performance CPU engineers from prominent tech companies, including Arm, AMD, and Tesla, focuses on delivering cost-effective solutions for AI models. The collaboration marks a pivotal move in Japan’s pursuit of leadership in high-performance compute design, with Tenstorrent’s commitment, LSTC’s vision, and Rapidus Corporation’s expertise coming together to shape the future of AI technology in Japan.

Tenstorrent’s strategic approach includes a strong emphasis on open standards, the integration of compute and acceleration technologies, and a flexible business model that monetizes different stages of chip development. As a competitor to Nvidia, Tenstorrent’s AI acceleration technology, powered by RISC-V compute cores, aims to offer flexible and high-performance solutions for various AI applications, from inference to training. The company’s partnerships with prominent tech giants and its ongoing development of next-generation AI tech position it as a significant player in the evolving AI hardware market.

With an expanding team, successful fundraising, and strategic collaborations, Tenstorrent is set to continue pushing the boundaries of AI technology with its innovative chip designs and commitment to open standards, advancing the landscape of AI hardware solutions for diverse applications globally.



Prompt Engineering Guides



©2024 The Horizon