The distinction between original equipment manufacturer (OEM) and original design manufacturer (ODM) relationships is primarily one of IP ownership: in an OEM arrangement, the customer provides the design specifications and the manufacturer produces to those specifications, while in an ODM arrangement, the manufacturer develops and owns the underlying design that the customer brands and sells. This distinction has profound implications for the IP provisions in the manufacturing agreement. In an OEM context, the customer typically owns all IP in the design, and the manufacturer's rights are limited to manufacturing for that customer; in an ODM context, the manufacturer retains design ownership and grants the customer a license — often non-exclusive — to sell products based on that design. Gurpreet Bal advises customers that exclusivity terms in ODM agreements require careful drafting: a commitment not to sell the same design to specified competitors, or within a specific geographic market, is fundamentally different from a commitment to produce designs that are uniquely developed for the customer. Tooling cost ownership — molds, fixtures, and jigs that may cost hundreds of thousands of dollars — should be addressed explicitly, including the customer's right to take possession of tooling on termination and the manufacturer's lien rights against tooling for unpaid invoices.
Semiconductor vendors — including Qualcomm, MediaTek, and NXP — often provide reference designs to their customers as a means of accelerating product development and ensuring that their chips are designed into end products in configurations that deliver the vendor's intended performance characteristics. Reference designs include schematics, bill-of-materials, PCB layouts, firmware, and software drivers, all of which are typically licensed under a reference design license agreement that imposes restrictions on modification, redistribution, and use with competing components. Gurpreet S. Bal advises hardware companies that the scope of permitted modifications in a reference design license determines how much of the customer's end product is independently owned versus subject to the vendor's IP rights, and that restrictions on derivative works can limit the customer's ability to sublicense design elements to ODM manufacturers, share designs with contract electronics manufacturers (CEMs), or incorporate reference design elements into products that also contain a competitor's components. The interaction between reference design license terms and the customer's obligations to its own downstream customers — including the right to provide design files for warranty service or regulatory certification — should be addressed before signing rather than litigated after a manufacturing dispute.
A complex hardware bill of materials aggregates patent exposure from dozens of component suppliers, standard-essential patent holders, and technology licensors, creating a royalty stacking problem that is difficult to quantify and manage. For connected hardware products — smartphones, IoT devices, automotive telematics units, industrial wireless sensors — the SEP exposure from Wi-Fi (IEEE 802.11), Bluetooth, and cellular (3G/4G/5G) standards is substantial, and the FRAND licensing terms for those SEPs are often contested. Gurpreet Bal advises hardware companies that the structure of SEP licensing in the cellular industry — where major holders including Qualcomm, Ericsson, Nokia, and Huawei each assert independent licensing programs at the device level — means that the component-level royalties paid by the modem manufacturer do not exhaust the SEP holder's rights against the device maker. This is a consequential and contested legal question: the exhaustion doctrine, as applied to FRAND commitments and multilevel supply chains, is the subject of ongoing litigation in the U.S., EU, and Asia. Patent pool licenses through Avanci (for cellular IoT at the device level) offer a mechanism for obtaining a single license covering major cellular SEP portfolios, reducing transaction and litigation costs for device manufacturers willing to accept pool pricing.
Firmware — software stored in non-volatile memory that is integral to the hardware product's operation — occupies a legally ambiguous position between hardware and software that has significant implications for licensing, warranty, and export control. From a licensing perspective, firmware bundled with hardware at the point of sale is typically treated as part of the hardware product, transferred under the hardware purchase agreement rather than a separate software license; but firmware updates distributed post-sale, or firmware provided under a separate software development agreement, may be subject to distinct software licensing terms including source code escrow, update obligations, and open source compliance requirements. Gurpreet S. Bal advises hardware companies that the hardware-software distinction also affects export control analysis: hardware and software components of the same product may be controlled under different Export Control Classification Numbers (ECCNs), and the product-level export classification must account for both. For hardware products incorporating advanced encryption, wireless communication technology, or dual-use functionality, the export compliance analysis should be conducted at the design phase — before manufacturing and distribution arrangements are finalized — to avoid the significant costs of retroactive reclassification, voluntary self-disclosure, and re-engineering to achieve a more favorable classification.
Gurpreet S. Bal is a Partner at Foley and Lardner LLP in Silicon Valley, where he advises technology companies on licensing, venture financings, M&A, and corporate transactions. He has represented clients in hundreds of transactions with aggregate deal value exceeding $60 billion across AI, semiconductors, fintech, and emerging technology.