CUDA and the Walled Garden: How the Relationship Between Nvidia’s Software and Hardware Violates the Sherman Act
ABSTRACT
This Note considers the role of antitrust law in the semiconductor
industry––one marked by extreme levels of corporate concentration and
power. This Note assesses the liability of Nvidia, a leading chip designer,
under Section 2 of the Sherman Act. Nvidia has anticompetitively sought to
maintain its monopoly position in the market for graphics processing units,
or GPUs, through software licensing agreements. The licensing agreements
for CUDA, Nvidia’s software product, leave software developers with two
options: incur switching costs to migrate their source code onto another
firm’s GPU hardware, or remain within the Nvidia ecosystem. As source
code translators threatened to splinter the flywheel between the firm’s
software and hardware, the licensing agreement emerged as a legal tactic to
protect the firm’s market power. A monopolization suit under Section 2 of
the Sherman Act, while costly, is likely the only means for the government to
rejuvenate competition for GPUs. Nvidia’s anticompetitive and illegal
maintenance of this monopoly harms consumers and stalls innovation in a
key industry of the future.