Editor's note: Chen Tianshi is included in the 'Pattern' section not because Cambricon's market value once surpassed Moutai, but because he spent two decades validating a proposition about industrial patience. In the AI chip arena dominated by international giants, he started from a 30-square-meter office, endured Huawei's supply cut, US sanctions, and eight consecutive years of losses, yet never changed his technical path. His pattern lies not in short-term market fluctuations, but in his clear recognition of the 'long race'—there are no shortcuts in the chip industry, only sustained investment can lead to breakthroughs. This long-termism based on industrial laws is the archival value this column aims to preserve.
Chen Tianshi's entrepreneurial journey began in a 30-square-meter office at the Institute of Computing Technology, CAS. In 2010, he and his brother Chen Yunji proposed the concept of dedicated deep learning processors. Despite industry skepticism, the team persisted, publishing an ASPLOS Best Paper in 2014 and taping out the world's first deep learning processor prototype in 2015. [2] In 2016, Chen left ICT to found Cambricon with a registered capital of only 0.9 million RMB. [1] In 2017, Huawei's Kirin 970 integrated the Cambricon 1A processor, boosting Cambricon's valuation from $100 million to $1 billion, making it a star AI chip company. [1] However, in 2019, Huawei announced its own Da Vinci architecture and gradually ended cooperation, causing Cambricon's revenue to plummet; Huawei-related revenue accounted for 98.95% to 99.69% from 2017 to 2019. [3] In July 2020, Cambricon listed on the STAR Market, becoming the first AI chip stock with a market cap exceeding 100 billion on the first day. [4] After listing, the company faced challenges such as stock price declines, expanding losses, and being added to the US Entity List, but maintained high R&D investment, finally achieving quarterly profitability in Q4 2024 and annual profitability in 2025. [8]
Chen Tianshi has repeatedly emphasized that the chip industry is a long-distance race with no shortcuts. He once said, "Intel is 52 years old, AMD is 51, NVIDIA is 27. Cambricon is only 4, the long race has just begun." [8] This long-termism permeates his business decisions. Facing Huawei's supply cut, he quickly diversified customers, bringing in internet giants like ByteDance, Alibaba, and Baidu. [7] Despite eight consecutive years of losses totaling over 5.4 billion RMB, Cambricon maintained R&D expenses at over 50% of revenue, insisting on technological investment. [8] After being added to the Entity List in 2022, the company accelerated domestic substitution, launching the Siyuan 590 chip with performance comparable to NVIDIA A100. [7] Chen believes that chip development has no shortcuts, and only persistent deep cultivation can achieve breakthroughs. [8]
Chen Tianshi has led Cambricon to become a landmark enterprise in China's AI chip field. Cambricon's listing filled a gap in the domestic AI chip track, and its technological accumulation provided important support for domestic chips in cloud and terminal markets. [4] In 2025, Cambricon achieved revenue of 6.497 billion RMB and net profit of 2.059 billion RMB, with market value once surpassing Kweichow Moutai to become the top A-share company, marking market recognition of the value of domestic AI chips. [8] [9] Chen himself holds 28.63% of the company, with a book market value of about 160.2 billion RMB, but he still emphasizes the long-distance nature of the industry, stating that the journey is far from over. [8] His experience also inspires many entrepreneurs: from confusion in the junior class to choosing a major by drawing lots, to the comeback after 20 years on the cold bench, demonstrating the power of persistence and long-termism. [1]
Chen Tianshi leaves a business proposition: how to balance short-term capital market pressure with long-term technological investment in an industry with long technology cycles. Cambricon's comeback proves that even after huge losses and external blockades, persistent technological cultivation can eventually realize value. His story also reminds future generations that true pattern is not chasing trends, but after judging trends, having the patience to sit on the cold bench for ten or even twenty years. This ability to embed personal choices into the logic of the times is what this archive truly seeks to preserve.