Smugglers solder HBM onto low-performance processors to bypass US export rules
Smugglers are evading US export restrictions on high-bandwidth memory (HBM) by soldering it onto low-performance processors, allowing the packages to be legally shipped to China. Once there, workshops in Shenzhen extract the HBM3E stacks and sell them for approximately $4,000 each, primarily to Huawei customers. The scheme exploits a December 2024 BIS exclusion for co-packaged chips where the dominant function is processing.
Editor, Lazyfounder

TABLE OF CONTENTS
How the smuggling scheme works
According to TechRadar, smugglers are exploiting a loophole in US export regulations by soldering high-bandwidth memory (HBM) from Samsung and SK hynix onto low-performance processors. The combined package is classified as a processor, which allows it to be legally exported to China under a December 2024 exclusion by the US Bureau of Industry and Security (BIS) for co-packaged chips where the dominant function is processing.
Packaging houses in Taiwan, South Korea, and Southeast Asia bond the memory and processor onto a single substrate. The packages are then shipped to Hong Kong, declared as processors, and smuggled into mainland China.
In Shenzhen, workshops use heat to separate the HBM3E stacks from the processors, which are discarded. The extracted HBM3E is sold for approximately $4,000 per stack, primarily to Huawei customers.
Why the method avoids export restrictions
The scheme targets a specific exclusion in the BIS rules introduced in December 2024. The exclusion applies only if the dominant function of the co-packaged integrated circuit is processing, rather than memory. By pairing HBM with a low-performance or deliberately ineffective processor, smugglers ensure the package meets this criterion.
This method was previously documented by SemiAnalysis more than 18 months ago, indicating it has been in use for some time.
Risks and market impact
The extraction process is technically challenging. Excessive or rapid heating can warp or crack the HBM3E stacks, leading to high failure rates. Despite these risks, the financial incentives remain strong due to the acute shortage of HBM in China.
Huawei has reportedly increased the price of its Ascend 950DT AI chip to over RMB 250,000 (~$37,000), citing memory costs and tight supply. The black-market trade has contributed to this scarcity, driving up prices for end customers.
Responses and denials
SK hynix has denied supplying Huawei since US restrictions took effect in 2020, stating it complies with US and South Korean export controls. Samsung has not publicly commented on the reports.
Faraday Technology, a Taiwan-based company, denied breaching export rules and claimed it has no ties to Huawei.
The scheme mirrors a separate case reported in April 2025, involving CoAsia Electronics, Samsung, and Faraday Technology.
What this reveals about enforcement gaps
The scheme underscores the limitations of export controls when faced with creative circumvention tactics. While the BIS rule targets co-packaged chips, it does not account for cases where the processor is functionally irrelevant, allowing smugglers to exploit the loophole.
TechRadar’s report also points to a pattern of enforcement challenges, including a 2024 US investigation into TSMC after one of its chips was found in a Huawei product. These cases suggest that supply chain tracing and enforcement remain inconsistent.
Broader context in chip smuggling
This story connects to earlier reporting on Huawei’s efforts to secure HBM through black-market channels. In a related development, Huawei directed customers to independently procure HBM for its AI chips, bypassing US restrictions and inflating prices to $4,000 per unit due to shortages.
Other cases highlight the broader trend of smuggling restricted semiconductor technology. In March 2025, Super Micro’s co-founder and two others were charged in a $2.5 billion scheme to route servers built with restricted Nvidia chips to China. In October 2024, the CEO of Earthmade Computer was arrested for allegedly shipping over $300 million of export-controlled servers via Malaysia and Singapore.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
This scheme highlights a critical vulnerability in export control enforcement. While the US and its allies have tightened restrictions on advanced semiconductor technology, smugglers continue to find creative workarounds. For founders and operators in the AI hardware space, this story underscores two key risks:
First, the unpredictability of supply chains. Companies relying on HBM or other restricted components may face sudden shortages or price spikes due to black-market demand. Second, the evolving nature of compliance. Regulators may close this loophole, but new ones could emerge, requiring businesses to continuously adapt their procurement and due diligence strategies.
For startups in emerging markets, this also signals the potential for regulatory arbitrage—but with significant legal risks. The arrests and investigations linked to similar schemes serve as a reminder that enforcement can escalate rapidly, even for tactics that exploit technicalities.
Related on Lazyfounder
Sources
- TechRadar · 2026-11-03
Smugglers are reportedly soldering top-tier HBM onto deliberately useless processors to sneak it into China, and the trick isn't new
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.


