The Semiconductor Industry has been at the forefront of digitalization by supplying products which have enabled other industries to pursue and pivot towards a more digitized operation. Chips manufactured by the industry have enabled increased data storage, real- time data processing at the edge and the very manifestation of IIoT and AI in value chains worldwide. However, in terms of progress, the automotive and medical device industries are outpacing semiconductor in terms of digitalization, with the semiconductor industry lagging behind. The interesting aspect is that the increased pace of adoption for automotive and other industries are made possible largely by the products supplied by the semiconductor industry.

As the KPMG Semiconductor Outlook highlights, “semiconductor companies had to re-trench during the early days of the pandemic and focus on immediate supply chain and other business continuity issues. This may have been done at the expense of longer-term digital optimization investments that have a longer payoff period.”

As an indicator of the lapse, McKinsey in a recent article highlights that only 30% of the semiconductor industry device makers surveyed claim to benefit from AI/ML and related Industry 4.0 technologies. The other 70% are either in the pilot purgatory or have their project stalled. These numbers clearly reflect that the industry lags behind in this important technology benefitting Industry 4.0, but what are the main reasons the overall progress of Industry 4.0 and digital transformation are slow in the industry?

Growth matters

From a scale perspective, the semiconductor industry is forecasted to achieve total sales of $ 527.2 Billion in 2021, with an 8.8% growth for 2022, per SIA. In a world quickly moving towards a ‘smart everything’ scenario, the requirement of chips, from standard to highly customized, high precision and high performance, is only going to increase. The need to increase capacities and/or add new capabilities translates to massive capital expenditures, especially in the semiconductor manufacturing space.

Every new chip designed and every new process step added is significant cost, but one which needs to be incurred if incumbent leaders and market players are to retain or consolidate their place is an ever-changing and highly demanding marketplace. McKinsey estimates that the R&D cost for chip development from a 65 (nm) node to a leading edge 5 (nm) node has risen from US$ 28 Million to US$ 540 Million; fab construction costs for the same nodes increased from US$ 400 Million to US$ 5.4 Billion. The demands on capital for new fabs mean most chip companies, with few exceptions, are not investing in net new construction, but rather upgrading and expanding capabilities when they can.

From a digitalization perspective, the semiconductor industry has been the premier supplier of IoT chips, while other industries were scrambling to become Industry 4.0 compliant and utilize IoT in their digitalization efforts.

This leads to a paradox. It is evident that there is a need to add new capacity and/or capabilities for fabs; capacity constraints amid a growing demand for product and an uncertain supply chain are tapping production output. Tackling high demand with the COVID pandemic causing disruptions and outages shifts the focus away from long-term vision and digitalization strategy to short-term operational enhancements and capacity management.

The challenge is further magnified for the industry as process complexity increases with the fast pace of digitalization outside the industry. Besides basic product quality, product reliability becomes critically important. A chip failure in a self-driving car for example can be lethal; a faulty chip in a medical device can lead to a misdiagnosis; this raises the bar for chip manufacturers supplying these industries.

Barriers to digitalization in semiconductor

The emergence of Electric Vehicles, 5G communications chips, IoT sensors and wireless technology further exacerbates the demand. Semiconductor companies continue to fall behind in their ability to produce. Forecasts show that unless the semiconductor industry drastically changes their methods and technology, adopting AI and digitalization in addition to capacity increases, there will be serious ongoing problems by 2025-2027.

Here’s a view of global fab utilization: the industry is near (or is at) saturation.

Source: www.semiconductors.org

The reliability/performance challenge is a symptom of a larger problem: how to add automation and other Industry 4.0 technologies to an already-stressed industry? We find that the challenges which keep semiconductor manufacturers from pursuing Industry 4.0 and digital transformation, then, come down to the following reasons:

  1. Demand pressures, shifting focus from digital transformation to meeting existing requirements.
  2. Perceived higher costs of digital transformation and fear of operational disruption leading to losses in the short term.
  3. Data existing in siloes and reliance on point solutions.
  4. Lack of integration across organizational IT applications, preventing the convergence of IT/OT infrastructure which forms the basis of digital transformation.
  5. Inherent complexity of the semiconductor manufacturing process, with up to 1,400 process steps in the production of a wafer alone. Each process step involves a variety of sophisticated tools, machines and people. 
  6. Inertia towards automating existing manual jobs.

The reliance on human-based processes in semiconductor manufacturing is longstanding. In older 6- and 8-inch fabs, individual process tools were typically automated and computer controlled but there was no automation and coordination of functions such as material handling, product/process characterization, and data collection and analysis between different unit operations. Larger capacity fabs today have more digitalization, but it’s the experience and technological maturity of the manufacturer that distinguishes their ability to adopt Industry 4.0. 

Addressing the digitalization challenge

How then can these challenges be addressed to enable a fast-paced digital transformation in the semiconductor industry?

As one aspect of digitialzation, McKinsey predicts that through the implementation of AI/ML alone the industry can gain $35-40 billion in value annually. Over a longer timeframe—3 to 4 years—it could double, to almost 20% of the industry’s current revenue. These are staggering numbers, but the key is to understand that digital transformation is a process that requires commitment from top management and a clear, well defined approach, which then leads to the step change envisioned. 

With the highly automated and complex nature of operations in a semiconductor fab, the challenge certainly isn’t the availability of enough data. Rather, it is the use of generated data within a framework that makes the data accessible, relevant, timely for decision making and contextual. A modern MES data platform is a good starting point for manufacturers attempting to begin their digitalization journey:

  1. A modern MES integrates the entire operation, be it at the SECS/GEM Level 1 or at the ERP/business systems Level 4. Integration allows the creation of a ‘single version of truth’, which forms the basis of digital transformation. The MES platform allows true IoT and supply chain-wide integration. Data from the fab and from extended operations is accessed in real-time and informed actions taken automatically. The MES also enables advanced scheduling, recipe/route management, exception management and resource tracking, all of which contribute to the agility of an organization, which in turn allow them to meet both high and variable demand better.
  2. The collection of data from the manufacturing’s edge, operational siloes, supply chain partners and the elimination of dependence on point solutions enables greater insights from data collected. Analytics from ‘big data’ collected allows process owners to view KPIs and insights previously unavailable from the process as a whole. When data is collected from multiple sources, cleaned and contextualized, it starts delivering value by affecting faster decisions, better operational results and increased process efficiency, all of which are desperately needed by the industry.  
  3. The modern MES enables process owners to create a digital twin of the fab, thereby enabling the simulation of new process steps, complex designs and managing costs pertaining to real-world trials. It also helps train personnel through deployment of VR and better process management and maintenance through AR. Adding advanced technologies such as AR/VR and Digital Twin allows 3D plant monitoring, and is a first step in implementing a full-fledged MES.
  4. Quality and maintenance become predictive in nature, where data captured from the process prompts corrective action in response to possible defects or breakdowns. It allows for requisite steps to be taken in order to prevent any loss or disruption. Automation of quality management through mounted sensors and cameras, allow for defects to be detected during design through manufacturing. It allows for quality to become more reliable, managed in a deeper sense and with more precision than ever possible from human visual surveillance.

McKinsey estimates that just through the deployment of AI/ML, semiconductor manufacturers stand to achieve a reduction of 28-32% in research and design costs and between 13-17% in COGS.

Time to act is now

Market pressures, the pandemic, material shortages all contribute to the semiconductor industry’s downtrodden position. While the industry has understandably been lagging behind in digital transformation efforts, we believe it is prudent to act now.

Simple start are:

  • to focus on leveraging pre-existing data for increased visibility and intelligence to guide decision making
  • to automate manual labor-heavy activities, such as data collection, material movement, inventory management and quality data management
  • adopt technology starting with an MES (or upgrade your MES) to reap the benefits of advanced technologies such as digital twin; AR/VR and AI/ML.

Altogether, these technologies lower labor and production costs, while improving quality, yield and equipment management. MES-enabled integration brings increased process automation, visibility and intelligence. MES forms the very basis for meeting future demands, grappling with current shortages, and leveraging your existing assets and investments.

 ###