Lian Weihan of Tenstorrent: Adhere to the strategy of full openness and launch the fastest RISC-V CPU in 2027

Publisher:EEWorld资讯Latest update time:2025-07-17 Source: EEWorldAuthor: 付斌Keywords:Tenstorrent  RISC-V  AI Reading articles on mobile phones Scan QR code
Read articles on your mobile phone anytime, anywhere

"Open source is innovation, and sharing is the driving force of progress." On July 16, 2025, at the main forum of the "5th RISC-V China Summit", Tenstorrent CEO Jim Keller and Tenstorrent Chief Architect Wei-Han Lien said this in a speech entitled "Integrating Computing into Openness, Enabling the Future with Wisdom".


Lian Weihan explained that the theme of this speech has three connotations: first, "integration" represents the deep integration of AI and CPU technology; second, "open" highlights Tenstorrent's adherence to the spirit of open source; and finally, "wisdom for the future" indicates Tenstorrent's vision of going beyond computing power improvement and moving towards the intelligent era. "We firmly believe that the future computing ecosystem will belong to those pioneers who maintain an open mind and dare to integrate innovation."


The golden age of AI chips , Tenstorrent embraces open source


At present, Nvidia's market value has exceeded 4 trillion US dollars, indicating that basic computing power has become a key threshold for the commercialization of AI algorithms. However, the diversified characteristics of AI application scenarios determine that chip architecture must move towards differentiated development.


Lian Weihan emphasized that this marks the arrival of the golden age of AI chip design. Only through collaborative innovation in the industry chain and the development of a chip portfolio with diverse architectures can the diversified needs of future AI be met.


DeepSeek is a good example, it reduced the price of AI by 30 times. Therefore, the price of future chip design is not just the chip, but also includes software, software open source development system, motherboard, and server. These make the chip price very expensive, reaching billions of dollars.



The key to achieving AI democratization lies in building an open and optimized infrastructure. To this end, Tenstorrent fully embraces open source.


As we all know, Tenstorrent is a company co-founded by legendary chip designer Jim Keller, who also serves as CEO. Lian Weihan also emphasized Jim Keller's attitude towards open source: "Our CEO has always maintained a firm belief in RISC-V and open source technology. At Tenstorrent, open source is not only a technical strategy, but also an innovation philosophy - we firmly believe that open sharing is the core driving force for the progress of the industry. All the company's technological achievements are open source, and this attitude of open competition stems from our confidence in our technological innovation capabilities."


Open Chip Architecture (OCA) , released at the end of the year


"We estimate that by 2028, 60% to 80% of high-performance AI chips will adopt chiplet architecture." Lian Weihan emphasized the RISC-V-based CPU design, open AI accelerator architecture, open source software stack, and the ongoing chiplet technology open source plan. Faced with the R&D cost of $1 billion to $2 billion for a single chip, only open collaboration can lower the threshold for innovation.



After two and a half years of research and development, Tenstorrent is about to release a revolutionary open chip architecture (OCA) at the end of this year. Compared with traditional solutions, OCA has three major breakthroughs: first, full stack open source, complete open source implementation from the physical layer to the software layer; second, systematic design, including a complete abstract system of the physical layer, transport layer, protocol layer, system layer, and software layer; third, ecological interoperability, ensuring plug-and-play of chips from different manufacturers.



OCA will bring significant value to the industry: first, it will reduce R&D costs by more than 40%; second, it will support manufacturers to focus on core value design; and third, it will achieve seamless collaboration of chipsets across manufacturers.


Taking automotive electronics as an example, OCA can flexibly support the computing power requirements of autonomous driving systems at all levels from L1 to L4 while maintaining a diversified supply chain.



Tenstorrent 's product plans


Tenstorrent is committed to creating the world's top CPU products, which stems from Tenstorrent's deep understanding of open source systems - excellent hardware practices must be accompanied by an equally excellent software development system. The success of the Linux system has proven this, and its excellent cost-effectiveness demonstrated by its perfect combination with the Pentium processor is the best example.


Lian Weihan said that Tenstorrent meets different market demands through diversified designs such as the Athena 8 core particles, and has formulated a clear product roadmap in the field of open hardware architecture.


According to the plan, the Ascalon series processors will maintain an annual performance improvement of about 10% in the next two to three years, and it is hoped that they will become the world's leading CPU product by 2027.



The Callandor processor, which is expected to be released in 2027, is expected to become the world's fastest processor and will also have excellent compilation capabilities and operating efficiency.



At the end of the introduction, Lian Weihan emphasized that Tenstorrent has always adhered to a comprehensive open strategy, adopting an open model from software development systems to hardware IP. Tenstorrent firmly believes that open collaboration can effectively reduce design costs, give rise to more innovative architectures, and ultimately bring better quality and more affordable AI products to the market.


Keywords:Tenstorrent  RISC-V  AI Reference address:Lian Weihan of Tenstorrent: Adhere to the strategy of full openness and launch the fastest RISC-V CPU in 2027

Previous article:AMD: AI accelerator card MI308 will resume shipment, and license application will be advanced to the review process
Next article:Intel roadmap exposed: Titan Lake CPU abandons P cores in 2028, up to 100 E cores

Recommended ReadingLatest update time:2026-05-12 10:48

Tutorial on setting up a Spacengine AI model deployment environment on the RISC-V development board MUSE Pi Pro
Spacengine is an AI algorithm model deployment tool developed by SpaceTime, which can help users easily deploy their own models on the edge. The official documentation provides two methods for environment deployment: Docker container Local deployment To isolate the local environment, we will
[Embedded]
Industry leaders talk about the prospects of RISC-V: AI, autonomy and innovation
On July 17, 2025, at the roundtable forum of the 5th RISC-V China Summit, Dai Weimin, Chairman of the Shanghai Open Processor Industry Innovation Center, Chairman of the RISC-V Summit, and Chairman of VeriSilicon Microelectronics, discussed a series of forward-looking topics about RISC-V with other RISC-V industry l
[Semiconductor design/manufacturing]
The world's first RISC-V tablet that can talk: Deepin Digital DC-ROMA unveiled, with built-in AI assistant
According to news on November 10, the three-day RISC-V North America Summit 2023 came to an end. At this RISC-V Summit, the organizer RISC-V Foundation set up a "LaunchPad" session for the first time, inviting companies to showcase their latest products. or solution. Domestic Deepin Intelligence showed its latest achi
[Home Electronics]
Esperanto demonstrates AI acceleration demo based on RISC-V chip
This article is translated from Forbes Esperanto has been talking about their edge AI chips for years, and now the company showed off a demo of AI acceleration for image, language, and recommendation processing. I had a chance to watch a demo of the platform and was impressed by the performance and energy efficiency
[Embedded]
RISC-V AI chips will be everywhere
Adoption of RISC-V, a free and open-source computer instruction set architecture first introduced in 2010, is taking off like a rocket, fueled in large part by demand for artificial intelligence and machine learning. The number of chips incorporating at least some RISC-V technology will grow 73.6% annually through 202
[Embedded]
An AI chip with more than 1,000 RISC-V cores
Esperanto Technologies, a developer of high-performance, energy-efficient machine learning (ML) inference accelerators based on the RISC-V instruction set, today announced that Dave Ditzel, the company's founder and executive chairman, detailed new details about the company's new ET-SoC-1 "supercomputer on a chip" at
[Embedded]
An AI chip with more than 1,000 RISC-V cores
Alibaba DAMO Academy releases XuanTie C950, breaking the global RISC-V CPU performance record.
On March 24th, at the 2026 XuanTie RISC-V Ecosystem Conference held in Shanghai, Alibaba DAMO Academy released its new-generation flagship CPU product, the XuanTie C950. Adopting the open-source RISC-V architecture, its single-core general-purpose performance exceeded 70 points in the SPECint2006 benchmark test, set
[Embedded]
Alibaba DAMO Academy releases XuanTie C950, breaking the global RISC-V CPU performance record.
RISC-V CPU and Imagination GPU Collaboration: A Perspective from K3: How to Build High-Performance SoC Capabilities
With the rapid growth in demand for edge AI and high-performance computing, the division of labor in the processor industry is changing. Recently, Arm released its self-developed AI chip, a move that has drawn more attention from the industry to the openness and ecosystem neutrality of IP models. Agains
[Embedded]
RISC-V CPU and Imagination GPU Collaboration: A Perspective from K3: How to Build High-Performance SoC Capabilities
Latest Embedded Articles
Change More Related Popular Components

EEWorld
subscription
account

EEWorld
service
account

Automotive
development
community

Robot
development
community

About Us Customer Service Contact Information Datasheet Sitemap LatestNews


Room 1530, 15th Floor, Building B, No.18 Zhongguancun Street, Haidian District, Beijing, Postal Code: 100190 China Telephone: 008610 8235 0740

Copyright © 2005-2026 EEWORLD.com.cn, Inc. All rights reserved 京ICP证060456号 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号