AI Simulation, the Next Frontier in Chip Design Software Tools (EDA) – Part-IV

In Part-I, Part-II?& Part-III of this series of articles, we discussed Triopoly in EDA Market, EDA TAM & Risks in the segment respectively. In this article, we will discuss how Chip Designers are using AI tools and the huge impact of this innovation over the next few years.

Why AI in Chip Design?

???????????The term AI is used very loosely and so let’s first define how AI is used in Chip design. Let’s say TSLA is designing an Automotive Chip, the designer has 20 variables he/she can change and then say another 10 EDA tool options. Now based on the Chip designer’s intuition and experience he/she would select a few options and run the EDA design tool for say 1-2 days and then tweak some variables to optimize and then rerun for another 1-2 days and so on and so forth. This process could take 6 months, is highly iterative and painful; could attest to it from my software engineering days. Now, with the AI simulation software combined with the EDA toolset, this design process could be completed in a week. EDA tool providers are moving up the value chain by providing this AI-powered simulation software as a part of the EDA tool set. This is a major productivity improvement for chip design engineers as it reduces the time to optimize a chip from six months to a week; secondly, it also makes the process less dependent on the chip designer’s intuition and experience. These AI-powered chip design simulation tools practically take one of the major functions of chip designers away from them but also help them design many more chips at the same time by boosting their productivity manifold. Finally, as Moore’s law slows down the chip industry is moving towards chiplets (multiple chips on a package) architecture; AI-powered simulation tools also help in chiplet design & analysis since chiplets add exponential complexity to the chip design process. Chip design simulation is developing as one of the niche segments within the general simulation market.

AI-powered Simulation: Total Addressable Market (TAM)

???????????As of now, EDA is a $10B market and is estimated to grow in the high single digits over the next five years. AI-powered simulation is another $10B market estimated to grow at low teens CAGR for the next five years; so, Simulation doubles the TAM available to EDA tool providers. This $10B TAM includes the entire Simulation industry (not only Chip design simulation) that EDA vendors are actively targeting.?The simulation segment could be broadly classified into two types:

A) Finite Element Analysis (FEA) Simulation: FEA is a simulation of any given product to analyze reaction to forces such as vibration, heat, fluid flow, and other physical effects. For example, undersea cables for wind farms are tested for heat, soil, pressure, etc. using FEA. Here EDA companies will compete with existing FEA tool providers such as Aspen Technology, Siemens PLM, Altair, Ansys, Dassault, MSC Software, Autodesk & ESI Group.

B) Computation Fluid Dynamics (CFD) Simulation: AI-powered simulation is also used in Computational Fluid Dynamics (CFD) by EDA companies. CFD simulation goes beyond chip design into system-level CFD simulation and optimization such as simulation to design and optimize the wings of an airplane or the shape of a car. This is an adjacent space since if Tesla is using EDA simulation tools for automotive chip design, they are also trying CFD system simulation software from EDA companies.

State of AI-Powered Simulation tools from EDA vendors

???????????SNPS & CDNS both introduced the first versions of their AI-powered simulation tools for chip design earlier this year. Clearly, they are highly likely to win the simulation segment within the semiconductor industry. CDNS is also in trials for a CFD system simulation product which will compete directly with Ansys and others in the space. CDNS recently announced the acquisition of OpenEye, a molecular biosimulation company. Future Facilities, which is going after the data center market and digital twins simulation for data centers is another acquisition by CDNS.

Conclusion: Simulation is a large TAM but with strong incumbents. It is going to be difficult for EDA vendors to gain market share in the highly competitive Simulation market. They are starting with a few niches within the simulation market with smaller acquisitions but unless they acquire a large player in the space, it seems difficult to gain any material market share in the simulation space in the short-mid term. On the other hand, the Simulation market is highly fragmented with many small vendors specializing in a particular industry vertical which combined with the high cash flow generation of EDA companies provides an option to acquire niche simulation companies. AI-powered Simulation in chip design is a much smaller market and is more like another tool offered in the EDA toolset vs. general simulation which includes CFD & FEA.

Subscribe to our newsletter on LinkedIn to receive updates or join our open to all Telegram group where we post additional quarterly stock reviews at: https://t.me/bayfortcapital

Links to all articles in this EDA series:

1)???Part I: EDA Triopoly - https://www.dhirubhai.net/posts/ketuls_nvidia-synopsys-cadence-activity-6975015392578555905-qYc-?utm_source=share&utm_medium=member_desktop

2)???Part II: EDA TAM - https://www.dhirubhai.net/posts/ketuls_synopsys-cadence-mentorgraphics-activity-6977506349010038784-ir2R?utm_source=share&utm_medium=member_desktop

Part III: EDA Risks - https://www.dhirubhai.net/posts/ketuls_semiconductors-eda-nvda-activity-6981552643856846848-YtUr?utm_source=share&utm_medium=member_desktop