AI chip design is coming with great momentum, and EDA giants are experiencing revenue growth along the way

New idea: AI EDA tools become a powerful tool for grabbing orders

Xinsi released its Q2 financial report on May 22, with quarterly revenue reaching $1.455 billion, a year-on-year increase of 15%. The non GAAP operating profit margin was 37.3%, a year-on-year increase of approximately 3%. Therefore, Xinsi has once again raised its guidance on annual revenue and non GAAP earnings per share.
In this era of widespread adoption of artificial intelligence chips and software defined systems, Xinsi believes that these technological trends have driven the complexity of research and development systems, but have also brought many opportunities. Nowadays, its chip customers are competing to design and manufacture complex specialized chips, and these customers are no longer limited to chip design companies, but also include many system solution companies.
The rapid adoption of AI tools has also brought unprecedented competitiveness to Xinsi, with its design automation business revenue increasing by 14% year-on-year, thanks to the strong performance of the overall business and Synopsys The rapid adoption of AI within customers includes a range of AI suites such as DSO.ai and ASO.ai. For example, in the second quarter, multiple Asian design service providers broke through the highest frequency target with the help of DSO.ai, while a leading American GPU manufacturer increased productivity by deploying DSO.ai.
Xinsi also mentioned that in analog and mixed signal design, customers hope to use Xinsi's solutions to achieve process modernization as soon as possible and migrate to more advanced process nodes. Xinsi won 10 alternative designs this quarter, including a leading American system solutions company switching to Xinsi's full process solution, and a leading Asian storage company switching to Xinsi's simulated design environment for its next-generation storage product design. In terms of verification tools, VSO.ai has also been widely used by more than 30 customers, reducing turnaround time by up to 10 times.
However, Xinsi is cautious about its performance in the Chinese market this year. Although the revenue performance in the first half of the year was good, due to macroeconomic challenges and some restrictions, Xinsi expects that the revenue from the Chinese market will still show a growth trend this year, but the proportion may be slightly lower than last year.

Cadence: The popularization of AI automation is just beginning

Cadence released its Q1 financial report at the end of April, with a total revenue of $1.009 billion, almost unchanged from $1.022 billion in the same period last year. Cadence expects revenue in the second quarter to be between $1.03 billion and $1.05 billion, with a total revenue of $4.56 billion to $4.62 billion for fiscal year 2024, higher than last year's total revenue of $4.09 billion. Cadence attributed its impressive Q1 performance and full year revenue expectations to the release of multiple new products last year and this year, as well as the tremendous success of its AI design engine.
At the recent Cadence Live conference, Qualcomm announced that they have shortened verification time by 20 times using Cadence's Verisium AI verification platform. And Cadence's other major AI tool, Cerebrus, has also been widely used in digital full process design. While providing first-class PPA and efficiency, it has become an indispensable part of the design process for major clients and the DTCO process for multiple wafer foundries. Currently, it has been used in more than 350 design projects, and more than 50 clients have deployed Cadence's solutions in the design of 3nm and above processes.
Cadence pointed out that in addition to the popularization of AI in the field of digital design, there will also be more AI automation in the PCB and packaging fields in the future, and Cadence's Allegro X AI will become a leader among them, which has been used in more than 300 customers. For example, Intel mentioned that using Allegro X AI in PCB design resulted in a 4x to 10x efficiency improvement.

Siemens: EDA Supports DI Business

Compared to the other two EDA giants, Siemens' EDA business is classified as part of its digital industrial group, alongside other industrial automation, electrical automation, and other businesses. According to Siemens' recent Q2 financial report, its entire digital industry group's business revenue was 4.505 billion euros, a year-on-year decrease of 13%.
This is due to the significant challenges in the market environment, high customer inventory, and overall decline in automation business orders, especially in the Chinese market. However, its software business increased from 1.165 billion euros in Q2 last year to 1.366 billion euros, and there was also a double-digit increase in the number of orders. Siemens stated that this was mainly driven by the growth of its EDA business, thanks to the high demand for EDA from semiconductor customers in the United States. Siemens achieved over 50% revenue growth in its EDA business. For the second half of this year, Siemens stated that it will see stronger customer activity in its software business, with a focus on EDA and licensing.
At present, Siemens has imported many AI functions into its EDA tools, such as Solido DE, which is a fully AI driven SPICE level design environment that integrates AI and cloud deployment technology. Solido DE can use AI to help users identify optimization paths, improve circuit PPA, and provide yield analysis, which is several orders of magnitude faster than traditional Monte Carlo analysis methods.
Currently, many major clients have introduced Siemens AI EDA tools, such as SK Hynix. In order to further improve validation accuracy and turnaround time, SK Hynix has utilized Solido DE in the design of next-generation memory technology, significantly reducing the time required from initial design to production. In addition, Forza Silicon, a CMOS image sensor manufacturer, also uses Solido DE as its simulation environment to design high-resolution high-speed CMOS image sensors for applications such as machine vision, automotive, and XR.
Recently, Siemens has released a new Catapult AI NN software for advanced synthesis of neural network accelerators on ASIC and SoC. It can convert neural network descriptions in AI frameworks into C++and integrate them into Verilo or VHDL for implementation on ASIC, FPGA or SoC. This can enable AI developers to achieve the best PPA performance in chip design during software development.

Write at the end

Compared to EDA giants, domestic EDA manufacturers are slightly behind in the progress of AI import. Although this does not affect its rapid revenue growth during the frequent design activities of last year and this year, the equally rapid growth in R&D investment is further compressing its profit margin. At present, mainstream chip design manufacturers have introduced AI EDA tools into the most advanced chip design process, and have shown unprecedented advantages in PPA and design efficiency. I believe this is also the reason why domestic EDA manufacturers are increasing their investment in AI engine research and development while striving to build full process tools.