We are heading towards the smart world that offers smart homes, smart cities, and industrial and business automation through IoT and Artificial Intelligence. In essence, AI makes it possible for machines to think. Today’s AI applications can not only process data but also learn from experience. Moreover, the applications can apply that experience to improve how they function. AI demands will have lasting impacts in complex manufacturing environments, such as those found in the semiconductor industry.
In large part, this is because the amount of data processed and stored by AI applications is massive. Saurabh Gupta, Research & Strategy Manager at Netscribes sheds light on the increasing benefits of AI adoption in the semiconductor equipment industry and the implications for the ecosystem as Industry 4.0 becomes increasingly crucial.
1. AI has now become a common phrase in the semiconductor industry. With businesses utilizing data to accelerate their growth, how is the use of AI preparing the industry for a smart future?
AI isn’t just a buzzword anymore. The last few years have seen the adoption of AI in the consumer, automotive, and industrial manufacturing sectors. Now, the semiconductor industry is standing at a stage where AI has started to assist in generating business and economic value at every stage, be it in operations, design, fabrication, R&D, manufacturing, sales, and even supply chain.
Moore’s law had laid the foundation of a futuristic shift towards increasing scale complexities. The law stipulated how the no. of transistors will be doubled every two years. This would help increase the number of data sources in new technology chips. The data explosion of the past few decades has finally allowed AI to disrupt reality. Today, every company and the whole industry wants to turn smart and deliver business value by leveraging their data. Because of the high CAPEX requirements, the semiconductor industry thrives on bringing the innovations and R&D costs to the high-volume production floors, within the smallest time possible and with the broadest range of errors. They persistently attempt to reduce their product life cycles and pursue R&D and innovation aggressively, to introduce products faster in the markets.
Now, the stakes are getting higher. With every decreasing technology node, expenses for chip design R&D and production at fabs have increased exponentially. These high stakes are applicable for chip manufacturers investing approx. USD 12 Bn to equip a new 100,000 wafer starts/month fab. In fact, the stakes impact the entire technology ecosystem downstream.
2. With costs and stakes on the rise in the technology ecosystem, can big data help semiconductor equipment companies establish successful AI strategies?
The primary question which arises around the increasing stakes and costs is this: How difficult is the current industry scenario?
Let us draw a simple analogy of board games to understand the above complexity running downstream the entire technology ecosystem. The complexity of board games increases exponentially based on changes to one of three parameters. These include: the size of the board, different types of pieces, and variations in the moves which can be played. For instance, the game ‘Go’ has around 3.2*10^11 possible combinations, while chess allows for 10^120 combinations.
Circling back to the semiconductor equipment industry: the complexity is very high, even unimaginable for a human mind to some extent. The data is huge and difficult to comprehend. It’s difficult to generate useful insights to improve process and equipment without using complex algorithms and supercomputing mechanisms.
This roadmap will integrate the entire value chain. It will also focus on novel advancements in shrinking chips, packaging, physical IP designs, 3D ASICs, and materials such as Indium, GaN, SiO2, and others.
3. How will the evolution of AI 2.0 or the new era of drawing insights from data shift the current dynamics of the semiconductor equipment industry?
AI 2.0 is basically a reference for reimagining AI, from ‘Artificial Intelligence’ to ‘Actionable Insights’. The industry will flourish from applying AI to the big data collected through the numerous sensors placed on various equipment processes. This will enable a data-driven input feedback mechanism to generate better outcomes for the engineering teams. Furthermore, it will transform the industry dynamics by accelerating product development lifecycle, reducing cost and risk implications, and maximizing return on investments.
4. In taking a data-first approach and maximizing the potential of AI, what challenges will the industry have to overcome?
Companies aim to become increasingly capable of analyzing and generating insights from the available massive data volumes. The first challenge, thus, will be on the computing front. New computing based on capacity-specific hardware systems will bring a new fraternity of chips into being. This will push established companies to deliver power consumption improvements in computing platforms and solutions.
Secondly, reducing the time-to-market for the new semiconductor equipment is key for equipment companies. AI will enable a simulation framework on digital twins and digital thread workflows. This will ensure the possibility for combination trials, errors, retrials, and iterative improvements. This step will be taken before actual funds, resources, and time is invested to develop new equipment for new technology nodes and to ship them from labs to fabs.
This benefit will be possible only when AI gets deployed across R&D, scale-up, and high-volume manufacturing (HVM).
5. Moving forward, which industries will drive the demand for increased adoption of AI and big data in the semiconductor equipment space?
Over the last decade, the demand for AI adoption has been driven largely by the consumer electronics industry. However, the COVID-19 pandemic has forced consumer scenarios such as WFH, online education, digital and telehealth, and online retail segments. These have significantly driven investments into cloud data centers and telecommunications and networking infrastructure. The equipment industry has seen a boom in big data as a result.
However, the willingness of companies as well as nations to transform themselves digitally will be the long-term growth drivers for the increasing relevance of AI in the semiconductor equipment industry. Further, an influx of commercial investments into IoT, big data, 5G, robotics, quantum, cloud computing, and AI/ML will drive the semiconductor equipment industry’s growth for the next 15-20 years.
Innovation Research and Business Strategy Expert
Saurabh has been in the strategy consulting and advisory space for over nine years, working closely with global clients on resolving their business, corporate, and operational challenges. His experience in client projects spans the sales, marketing, IP legal, and operations functions.
When it comes to the technology vertical, he has specific interests in semiconductors, electronics, nanotechnology, and how digital transformation impacts these segments. In his free time, he loves traveling, hiking, reading extensively, and spending quality time with family.