Former Google Staffer Accused of Funneling Supercomputing Secrets to Beijing
DOJ Alleges Software Engineer Quietly Transferred Cutting-Edge AI and Chip Designs to PRC-Based Ventures, Exposing National Security Risks
SAN FRANCISCO — In a groundbreaking indictment unsealed Tuesday—spotlighting the high-stakes race between China and the United States for dominance in artificial intelligence, which experts warn carries critical military implications—federal prosecutors charged Linwei “Leon” Ding, 38, with seven counts of economic espionage for allegedly stealing Google’s proprietary supercomputing technology to benefit Beijing.
Prosecutors say Ding joined Google in 2019 as a software engineer with privileged access to the company’s supercomputing data centers, AI frameworks, and hardware configurations. Beginning in May 2022, he allegedly began secretly uploading Google’s confidential data to his personal Google Cloud account, transferring more than 1,000 files by May 2, 2023. Investigators claim he copied Google source files to Apple Notes on his company-issued laptop, converted them to PDFs, and funneled them into a personal account. Authorities characterize these files as the crown jewels of Google’s AI ecosystem—critical orchestration software capable of harnessing thousands of interconnected chips.
The superseding indictment outlines that the stolen data contained proprietary information about Google's software that allows these chips to communicate and execute tasks, and Google's custom SmartNIC technology—an advanced network interface card that enhances cloud networking and high-performance computing capabilities. This infrastructure is the backbone of Google's AI model training, allowing vast clusters of processors to operate as a single, powerful supercomputer.
For the uninitiated, this type of computing power enables software to process vast amounts of data at speeds approaching real-time decision-making and interaction, mimicking elements of human intelligence.
While Ding was employed by Google, he secretly affiliated himself with two PRC-based technology companies. Around June 2022, he was in discussions to become Chief Technology Officer for an early-stage technology company in China called Beijing Rongshu Lianzhi Technology Co., Ltd. According to prosecutors, Rongshu was developing acceleration software for machine learning on GPU chips, as well as federated learning AI platforms that allow models to be trained on decentralized datasets. The company’s CEO offered Ding the CTO position, a monthly salary of 100,000 RMB (approximately $14,800 in June 2022), an annual bonus, and company stock.
Although Ding never disclosed any PRC affiliations to Google, he allegedly leveraged the confidential information he had been collecting to advance his career at Rongshu. Federal agents assert this marked a key milestone in his alleged plan to commercialize Google’s AI expertise on behalf of Chinese interests.
By May 2023, Ding had founded his own AI and machine learning startup in the PRC—Shanghai Zhisuan Technology Co. Ltd.—and was acting as its CEO. In November, he pitched Zhisuan’s capabilities at a Beijing venture capital conference.
During these trips, he allegedly secured meetings with potential investors, pitched his AI firm’s ambitions, and refined a product roadmap that closely mirrored Google’s high-performance computing blueprint. Investigators say they uncovered communications linking Ding’s endeavor to a PRC “talent program,” which incentivizes overseas experts to transfer cutting-edge technology back to China, including knowledge with military or dual-use potential.
Newly revealed documents indicate that Ding actively promoted China’s AI ambitions within his startup. Prosecutors say he circulated a PowerPoint presentation to employees citing PRC national policies encouraging the expansion of domestic AI infrastructure. Another PowerPoint, submitted as part of an application to a PRC talent program in Shanghai, explicitly stated that his company’s product “will help China to have computing power infrastructure capabilities that are on par with the international level.”
The superseding indictment underscores how PRC-sponsored talent programs offer lucrative salaries, research funding, and lab space in exchange for the transmission of cutting-edge research and industrial secrets. Court filings suggest that Ding’s alignment with these initiatives provides clear evidence that he was not merely engaging in private entrepreneurship but was actively working to strengthen China’s technological standing, potentially at the expense of U.S. national security.
The Ding indictment takes on added weight in light of the DeepSeek controversy, in which U.S. officials have raised concerns about another Chinese AI firm’s alleged misuse of American-made semiconductors in violation of export control restrictions. DeepSeek’s rapid rise—punctuated by its AI assistant becoming the top download on Apple’s App Store—shook U.S. tech markets and prompted Microsoft, an investor in OpenAI, to accuse DeepSeek of misappropriating proprietary technology.
These parallel concerns underscore the urgency with which authorities are scrutinizing China’s accelerating efforts to acquire advanced computing capabilities. Experts warn that such technology transfers, whether through insider threats or supply chain loopholes, could bolster the PRC’s commercial competitiveness while enhancing its military and surveillance capabilities.
American officials have repeatedly emphasized that economic espionage cases involving AI and supercomputing extend beyond corporate theft and touch upon national security interests. By allegedly siphoning off the intricate blueprints for Google’s supercomputer-scale AI systems, Ding may have significantly widened China’s pathway to developing or refining similar platforms.
If the Department of Justice succeeds in proving economic espionage—a charge that surpasses simple trade secret theft by encompassing actions intended to benefit a foreign government—Ding’s prosecution could set a crucial precedent for how the United States confronts clandestine efforts to extract Silicon Valley’s most valuable innovations.
https://nationalpost.com/news/canada/canada-doesnt-screen-china-affiliated-researchers