Manufacturing is an age-old cycle of using past work to advance the precision and efficiency of the next generation of products. This circular iteration is on grand display right now in the semiconductor industry. At Nvidia's (NVDA 6.18%) technology conference in March, it unveiled new ways its most advanced chips are now being put to work to accelerate the development and manufacture of even more advanced future semiconductors.
The semiconductor industry is beginning to run up against the limitations of physics as it shrinks the dimensions of chips down to the atomic level. Nevertheless, as the once-parabolic advance of chip performance begins to flatten, Nvidia is becoming a key partner in keeping its peers working at peak efficiency. One of these important partners is software leader Synopsys (SNPS 2.56%). Here's why Nvidia's recent advances are fantastic news for this software-based play on computing technology's basic building blocks.
Moore's Law is finished, long live Moore's Law
Moore's Law -- the prescient prediction made by the late great Intel co-founder Gordon Moore decades ago -- is getting more difficult to keep advancing. Nvidia CEO Jensen Huang has been saying for years that Moore's Law is over as the ability to double chip performance every year and a half gets harder to achieve.
But now Nvidia is helping push the boundaries of Moore's Law, thanks to its work in artificial intelligence (AI). Today, the most advanced chips are fabricated using a process called extreme ultraviolet (EUV) lithography. ASML Holding is currently the only company that produces the complex and expensive EUV equipment, used by top chipmakers like Taiwan Semiconductor Manufacturing.
Designing these tiny little powerhouses requires the use of specialized software called electronic design automation (EDA) -- sort of like a drafting tool, but for chips. Synopsys is the leader in EDA software. Chip design goes to the nanometer (1 million nanometers per millimeter) scale to draw up and later sculpt intricate patterns of silicon and copper wiring. To speed up the process, AI is used -- that's where the "automation" comes from in EDA.
This is where Nvidia swoops in. Its new chips (utilizing the company's next-gen "Hopper architecture") are running an AI-based software library Nvidia developed called cuLitho (short for computational lithography), which is being adopted by ASML, TSMC, and Synopsys. Nvidia says that using cuLitho provides a 40x leap in designing advanced patterns for the EUV manufacturing process.
In practical terms, Nvidia says that means just 500 of its new Hopper chip systems (called the DGX H100 system) can replace up to 40,000 legacy CPUs (like the kind Intel specializes in) currently being used to handle the same computational workload.
Why Synopsys could be the biggest winner from cuLitho
Nvidia's pioneering work will be great news for chipmakers like TSMC, as it will speed up production times and increase yield (the number of chips produced from a single wafer). It could also help further adoption of ASML's very expensive EUV equipment.
But Synopsys could be the biggest winner here. Adopting cuLitho using new Nvidia chips could make its cloud software run far more efficiently (Synopsys also cited the 500 Nvidia systems that can replace 40,000 CPUs). That means cost savings, not to mention spurring on more use of its EDA software suite across the whole chip industry -- from design to manufacture.
Synopsys is a steady growth business, as can be expected from a primarily subscription-based software outfit. But a key tenet to my Synopsys bull thesis is management's stated objective to increase operating profit margin toward the 30% mark in the coming years -- an achievement already attained by smaller EDA peer Cadence Design Systems.
Data by YCharts.
Operating costs have been rising for cloud computing companies, since running all that equipment is getting more expensive due to higher energy prices. Increasing computing efficiency could be a key way Synopsys gets itself more fit as it steadily grows its revenue.
Of course, Synopsys is not a cheap stock. Shares currently trade for 37 times trailing-12-month free cash flow. But the company has a clear path to sustaining a low-teens-percentage revenue growth for years to come, all the while increasing that profit margin. It all could add up to a top cloud software stock over the next decade. Nvidia's announcement looks like great news for owners of Synopsys.
