Interview with de CEO of Synopsys, Soort Geeus: Ki-Simple (2023)

The establishment of Chatgpt at the end of 2022 and at the beginning of 2023 promoted a new wave of Ki trends.The semiconductor sector has received part of the most attention in this trend.This is because AI can not only offer massive business opportunities, but can also reach the entire semiconductor industry, in particular the chip design, with the help of AI, with the further breakthrough.

Aart de Geu, chairman and CEO of manufacturers of electronic design automation (EDA) Synopsys, believes that AI is actually an important trend that can change the entire semiconductor industry, including the world.However, we still have to make this vision many challenges to go beyond technology, talent, energy and geopolitics.

The following is a summary of the interview by Aart de Geus with Digitimes:

Question: What changes does the AI trend bring to EDA tools?

A:In about 12 years there were currently many questions, because the next growth distribution is for the semiconductor industry?I predicted the moment it would be "intelligent".The conclusion was that it would have a major impact on the semiconductor industry if everything happens.

It is worth noting that Machine Learning (ML) was already visible as a direction for technological development.In addition to the current appearance of generative AI, Synopsica invested in the ML sector and improved the development tool.

Synopsys continued to invest in the AI and automated construction sectors about five years ago.At the beginning of 2021 we have the first proof of a customer who was able to use our AI tools to record a large circuit and to produce successfully.

Within the Synopsy AI tools there was an important function called DSO (Design Room Optimization).Chip products that continued had better computing power, lower power consumption and in some cases of smaller size.

Until the beginning of 2023, the AI tools of the Synopsys have reached more than 100 glue tape in automated designs.In the middle of the year this number has grown quickly to 230, which represents an amazing growth rate.

In general, Synopsys took over the AI technology much earlier than the current generative Ki -Boom.

The strength of the generative AI that we see now is that you can request huge amounts of different information and create interesting and valuable content.However, it can also create a fully defective content many times.We have no room for errors in the EDA tool area.If a single transistor does not work, nothing works.

This is not negative for generative Ki.I think what we see here is a fantastic application.We have already taken over a number of generative AI technologies within Synopsys.The degree of use in EDA tools is not yet very large.

Question: It seems that you believe that the next generation EDA tools is developing in the direction of "automation of automation".Does this direction solve the problem of the lack of half -intestinal talent within 5 to 10 years?

A:Exactly 37 years ago Synopsys invented the Zo -driven synthesizer.As we did the users for the first time, about 2% stated that this "removes their work", but 98% welcomed the convenience of new tools because they could become more efficient and do so much more.

Now, 35 years later, this is exactly the same situation.If automated tools become more efficient, human intelligence must look: "What can I do with these new tools?"How can we be more creative when reducing the degree of placking in the design of new architectures and systems?

A few years ago I said that we are moving from the Moore Law and enter what I like to call the set "SYMOORE".

Photo: The previous BOGSCH Act emphasizes the exponential scale growth.The Sysmoore -Act emphasizes systemic complex growth.

For example, the functions and structures of auto systems are becoming increasingly complex.A chip no longer solves all problems.It requires many different technical and professional knowledge, which increases the technical demand for semiconductors, hence the lack of talent.

AI can certainly release more potential for people and enable us to think about what else can be done.Most people, if they come across a useful tool, will be very happy because their work is simply easier.

Now chip design tools differ from normal tools.They must have a good training and knowledge to use these tools.Those who have been invested influence the size of the talent pool.

Some countries in the 1990s died of engineering.During this time everyone wants to be a financial person or a lawyer.Currently, jobs such as Engineering have returned to the mainstream of society, and we can absolutely acknowledge that there is more interest in universities.Despite the field, they are in, the required core power to perform complex tasks and to solve problems.

When I speak with graduates in the engineering, I always tell you that you can be a deep engineer in your field if you know that the right fields that are side by side are very valuable.Take, for example, cars and software are now also important.In addition to the chip, the engineers must also understand warmth, mechanic and even reliability.Is something that everyone has.

Question: Due to a push from AI, a very popular option has become a very popular option in these years.How do you see this new ASIC trend?Are there more and more companies in this model applicable?

A:Regardless of ASIC or the standard market, I believe that the visible trend in the future is that we see more and more inter -industrial cooperation between companies that bring together different skills to build a good solution.

The same development will also take place in the AI sector.We can already see that some companies that specialize in AI algorithms collaborate with chip design houses to develop faster and cheap chips.Such as specialties in every technology.

Another part I like to look for is that the law of the Trend Moor was based and makes things smaller, does more of it and makes it cheaper, still developed, but because this trend is delayed, people's attention has gradually changed to work more efficiently on the same chip.

To achieve this, the signal transmission speed between chips must be fast enough.An even more complex production technology is required, which considerably increases the design complexity.During some companies, some companies can take on all functions in an IC themselves, some product technology from different design companies, for which a close team work is required.

This applies to the importance of the system complexity that I have mentioned earlier.

Question: What are the most important functions in future business opportunities for AI companies that need semiconductor companies to achieve success?

A:This is a very complex question, but it actually depends on two important factors.The first is to be good at what you do.

Previous experiences have taught us that the high-tech sector has a function, namely that one or two important changes will determine what will happen in the next generation.The people asked me why synopsia has been successful for so many years, and one of the reasons is that we have never been behind the semiconductor industry because they have increased exponentially.We are always aware of the latest technology.

The second factor that in my opinion has become more important than the first is that you really have to be good for your partner.In other words, you don't just think as a supplier or customer.Interaction with your partners from the perspective of cost -effectiveness.You must think about what you can do to help your partner company to create more differentiation and benefits.

There is nothing wrong with reducing costs, but increasing differentiation is even better. It is not only very good at reducing costs, but also more differentiation and unique benefits.