Morgan Lewis has published a legal analysis outlining the principal US export control considerations facing companies building data centers outside the United States, addressing a compliance environment that has tightened materially for chip-heavy infrastructure projects since 2022.

The piece, titled "Key US Export Controls Considerations for Global Data Center Projects," sits on the firm's publication platform. Morgan Lewis is a US-headquartered international law firm with a substantial trade-controls practice whose semiconductor and technology clients include foundries, fabless design houses, and hyperscale operators.

Data center projects carry elevated export-control exposure because their core compute, networking, and storage systems depend on advanced semiconductors, high-bandwidth memory, and lithography-related manufacturing tools — categories the Bureau of Industry and Security (BIS) has progressively restricted under the Export Administration Regulations (EAR).

The analysis organizes its guidance around the practical decision points project sponsors encounter: technology classification, license requirements, end-use and end-user review, and the extraterritorial reach of US rules over foreign-made products that incorporate US-origin technology.

What categories of equipment trigger the most scrutiny?

Advanced computing chips, AI accelerators, and high-end networking semiconductors sit at the top of the BIS priority list. EAR controlled-items entries — including ECCN 3A090 and 4A090, added in successive rulemakings — capture the silicon that powers modern training and inference clusters, including the GPUs and custom ASICs deployed by hyperscalers.

Manufacturing capital equipment falls under separate but related controls. Lithography systems, deposition tools, etch equipment, and metrology instruments produced by ASML, Applied Materials, Lam Research, and KLA all carry ECCN designations that restrict shipment to China and other destinations without a license.

Memory devices, SSDs, and high-bandwidth memory stacks used in AI servers are also captured, particularly when classified above defined thresholds for throughput, density, or interconnect bandwidth.

How do extraterritorial rules shape overseas builds?

The Foreign Direct Product (FDP) rule extends EAR jurisdiction to foreign-made items that are "direct products" of US technology or software. For data centers sited abroad, equipment assembled outside the US can still require a US export license if US-origin intellectual property sits anywhere in the production chain.

The Entity List, Unverified List, and Military End User List further constrain which counterparties can receive controlled items. Project sponsors must screen lessees, contractors, and ultimate beneficiaries against these lists before deploying hardware.

The analysis also flags the EAR's de minimis rules, which determine when a foreign-made item containing US-controlled components crosses the threshold requiring US authorization for re-export.

What does this mean for project economics?

License timelines, technology substitution costs, and alternative-architecture engineering all affect total project cost. A data center designed around a specific accelerator family may face licensing delays that push commissioning dates; in some cases, operators have re-architected deployments around lower-classification silicon to preserve schedule.

Supply-chain documentation becomes a recurring operational burden. Export compliance teams must maintain classification records, end-use statements, and screening logs across the project lifecycle, not just at initial shipment.

The Morgan Lewis piece comes as hyperscale and colocation operators expand capacity in the Middle East, Southeast Asia, India, and Latin America — markets where US export controls can apply to even modest compute densities when the underlying silicon crosses ECCN thresholds.

The firm's analysis frames export controls as a gating design parameter for global data center projects rather than a transactional paperwork step, and signals that compliance diligence will continue to influence where new capacity comes online and which silicon architectures hyperscalers specify for cross-border deployments.