The Chip War's Forgotten Front: Software

When policymakers in Washington discuss the chip war — the escalating contest between the United States and China for supremacy in semiconductor technology — the conversation almost invariably centers on hardware: fabrication plants, lithography machines, and the advanced processors that power everything from smartphones to missile-guidance systems. What receives far less attention, despite arguably mattering more, is the software without which none of those chips could be designed in the first place.

The software in question is called electronic design automation, or EDA — a category of extraordinarily complex tools that allow engineers to design, simulate, and verify integrated circuits containing billions of transistors before a single wafer is ever etched. Without EDA software, chip design as practiced today would be physically impossible. A modern processor contains so many components, interacting in so many ways, that no team of engineers could design or verify it by hand, any more than a team of cartographers could hand-draw a street map of every road on earth.

The EDA market is dominated by three companies: Synopsys and Cadence, both headquartered in California, and Mentor Graphics, now a division of Siemens with significant operations in the United States. Together, they control an estimated eighty-five percent of the global EDA market. Their tools are used by virtually every major semiconductor company on the planet, from the largest foundries to the smallest fabless design houses. The market is not large by the standards of the technology industry — total EDA revenues are roughly $18 billion a year — but its strategic importance is wildly disproportionate to its size.

The U.S. government recognized that importance when it extended export controls to EDA tools in 2022, restricting the sale of certain advanced design software to Chinese companies. The controls were narrowly targeted — aimed at the tools used to design the most advanced chips, particularly those built at process nodes below fourteen nanometers — but they sent a clear signal that Washington views software as a chokepoint of equal strategic value to the lithography machines and fabrication equipment that had been the primary focus of earlier restrictions.

The impact on China's semiconductor ambitions has been significant, though difficult to measure precisely. Chinese chipmakers had been heavy users of Synopsys and Cadence tools, and the restrictions forced them to scramble for alternatives. Domestic EDA companies exist — the most prominent, Empyrean Technology, has been the beneficiary of substantial government investment — but industry experts say the gap between Chinese tools and their American counterparts is measured in years, if not decades. "EDA is not the kind of software you can replicate by throwing money at it," said Dr. Catherine Hsu, a semiconductor industry consultant based in Taipei. "It is the accumulated product of forty years of algorithmic research, customer feedback, and iterative refinement. You cannot leapfrog that. You can only grind through it."

The difficulty of closing the gap reflects the nature of the software itself. EDA tools are not monolithic applications; they are ecosystems of hundreds of interrelated programs, libraries, and databases that must work together with nanometer-level precision. A single bug in a verification tool can cause a chip to fail in ways that may not be discovered until millions of dollars have been spent on fabrication. The reliability requirements are closer to those of aerospace software than to those of a consumer application, and the expertise required to develop and maintain the tools is concentrated in a small community of specialists, many of whom have spent entire careers at one of the three dominant companies.

For the United States, the dominance of American companies in EDA represents a powerful source of leverage — but also a source of risk. Allies have expressed concern about the degree to which their own semiconductor industries depend on tools that are subject to U.S. export-control jurisdiction. A European chipmaker designing a product for a non-restricted market still uses American EDA software, and the theoretical possibility that access to those tools could be restricted — for geopolitical reasons that have nothing to do with the chipmaker's own customers — has prompted conversations in European capitals and in the boardrooms of Asian foundries about the desirability of alternatives.

"The EDA chokepoint is a double-edged sword," said Dr. Hsu. "It gives the U.S. enormous leverage today. But it also creates an incentive for everyone else to invest in alternatives. If you are a European or Japanese or Korean semiconductor company, the lesson of the past two years is that your entire design capability runs on software controlled by a single government. That is a dependency that boards of directors are starting to take very seriously."

For now, the alternatives remain distant. The investment required to build a competitive EDA ecosystem from scratch is enormous, the talent pool is thin, and the incumbents continue to innovate at a pace that makes catching up a moving target. But the chip war has a way of accelerating timelines that peacetime economics would consider unrealistic. The hardware front — fabs, lithography, packaging — has already reshaped global investment patterns and trade relationships. The software front, quieter and less photogenic, may ultimately prove to be the more consequential battleground. The chips that will define the next decade of computing cannot be built without the tools to design them, and those tools, for now, are American.