Harder Chips, Better Business
Semiconductor design software has a property that most businesses would envy: the harder the underlying work gets, the more of the software gets bought. When chips become more difficult to design, the tool vendors sell more licenses, more simulation time, and more support. Complexity is revenue.
Synopsys (SNPS) is one of the two firms, alongside Cadence Design Systems, that capture most of that spending. The “AI Super Chip” event named Synopsys as its free pick, and the reasoning connects directly to this dynamic. The AI buildout is making chip design dramatically harder, and Synopsys is paid to make it manageable.
The event itself is broken down in the AI Super Chip teardown.
The Three Forces of Complexity
The first force is the shrinking process node. Each new generation packs more transistors into the same area, which multiplies the number of things that can go wrong. Wires get closer, heat gets denser, and timing gets tighter. Every one of those problems has to be solved in software before fabrication.
The second force is the chiplet. Instead of one giant die, designers now assemble several smaller pieces in a single package. Each chiplet is its own design project, and then the whole package has to be verified as a system. That is more tool work, not less.
The third force is the custom accelerator. Companies that want a chip tuned to their own AI workload design something unique instead of buying a standard part. Every custom design is a new customer for the tools.
Why Accelerators Feed the Duopoly
AI accelerators are the purest expression of all three forces at once. They use the newest nodes, they increasingly rely on chiplets, and they are heavily customized. A wafer-scale engine or a next-generation GPU is about as hard a design problem as the industry produces.
That is the key insight behind the free pick. Whether the winning accelerator comes from Cerebras, Nvidia, or a startup no one has heard of yet, it gets designed in EDA software. Synopsys earns revenue from the competition itself, not from picking the winner.
The distinction between training and inference chips is covered in our piece on AI inference vs training, and both categories pass through the same design tools.
The Value Multiplier
There is a second-order effect worth naming. As design gets harder, the cost of a mistake rises. A flawed design that reaches the foundry can waste hundreds of millions of dollars and a year of schedule. That risk makes design teams willing to pay up for the best tools, because the software is cheap relative to the cost of failure.
The result is pricing power that grows with complexity. The tool budget is a small fraction of a chip program’s total cost, but it is the fraction that determines whether the rest of the money is spent well. Vendors sit in that position of outsized importance.
Why New Entrants Struggle
A reasonable question is why a market this valuable does not attract more competition. The answer is the flow. A design team’s tooling is not a collection of standalone programs; it is an integrated pipeline where each stage hands data to the next. Libraries, scripts, and verification harnesses are all built against one vendor’s formats, and those assets accumulate over years and across products.
A challenger can build a strong point tool and still not matter, because customers need the whole flow, not one impressive piece. That is how Synopsys and Cadence Design Systems have held their grip through decades of technological change. The barrier is not a single piece of software; it is the accumulated infrastructure wrapped around it, which no startup can reproduce quickly regardless of how much capital it raises.
The same dynamic rewards the incumbents as complexity grows. The deeper a team is embedded in one flow, the more the flow itself becomes part of the company’s know-how, and the less rational it is to rip it out.
The Flywheel Effect
There is a flywheel hidden in this dynamic. Harder chips generate more tool revenue, which funds better tools, which enable even harder chips. The software vendors and the chipmakers are locked in a loop that raises the ceiling for both, and the toolmakers capture a toll at every turn.
It also means the growth is not dependent on any single customer winning. When one accelerator design cancels, another takes its place, and both spent money on the same tools along the way. That is the structural appeal of selling to the design problem itself rather than to any one chip.
The Verdict
Semiconductor design software is one of the few businesses that gets structurally more valuable as its industry advances. Smaller nodes, chiplets, and custom accelerators all point the same direction: more design work, more risk, and more spending on the tools that manage both.
Synopsys is positioned directly in that flow. The free pick is not a bet on any single chip; it is a bet that designing chips will keep getting harder, which is about as safe a bet as the semiconductor industry offers.
Ready to see the research? Click here to access the Brownstone Research report.
NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.