Semiconductor machines keep getting better, getting them to work together is the hard part - SDxCentral

Search

Show Menu

Close modal

Search Suggestions

Get the Report: How leading storage architects are solving the AI memory crunch

Home

Industry Views

The Silicon & Supercomputing Channel

Semiconductor machines keep getting better, getting them to work together is the hard part

Ch'ng Eng Yong, Country Manager, Delta Electronics Malaysia

Ch'ng Eng Yong is country manager at Delta Electronics Malaysia, with 19 years of experience bridging strong technical expertise and commercial leadership in the energy and automation sector. At Delta Electronics, he focuses on advancing clean energy and smart automation by integrating intelligent systems across power, infrastructure, and industrial applications, which ultimately enhances operational efficiency and supports sustainable growth.

What should have been a seamless addition to the production line becomes yet another silo

August 17, 2026

Have your say

Imagine bringing in a new process engineer to manage equipment, systems, and workflows in your semiconductor fab. They are highly capable but refuse to talk to anyone else on the factory floor.<br>Now replace that engineer with a newly installed piece of equipment. It operates in isolation, unable to integrate or share data with other systems. Instead of improving performance, it creates throughput slowdowns and data bottlenecks. What should have been a seamless addition to the production line becomes yet another silo.<br>Such situations are not uncommon, but with the semiconductor industry expected to reach $1 trillion to $1.1 trillion by 2030, the costs of getting integration wrong can be exceptionally high.<br>The integration problem<br>Organizations invest millions of dollars in equipment to enhance their manufacturing capabilities. However, the decision to purchase such equipment is rarely made by a single individual due to its high cost. Often, stakeholders involved in making the choice operate independently.<br>Failure to collaborate across functions before purchasing the equipment can result in many drawbacks, such as having a state-of-the-art machine that is unable to function within the existing fab ecosystem or one that the staff are unable to maintain or operate. Integration must be treated as a design priority instead of an afterthought.<br>Semiconductor wafers pass through multiple machines as part of the fabrication process. These machines often come from different original equipment manufacturers (OEM), run different software stacks, and may even run on different communication protocols.<br>The real challenge here lies in not just simply connecting individual machines but ensuring the entire system functions as a cohesive unit, so the full production line can operate well.<br>While a communication protocol that aims to unify communication between a variety of factory equipment does exist, it requires lengthy and laborious development. Established by the Semiconductor Equipment and Materials International (SEMI), the SECS/GEM standard is a highly complex communication standard that can take up to three years of development.<br>As a result, some equipment manufacturers have turned to software tools to simplify SECS/GEM implementation. Such software can help facilitate communication between machines and control systems, enabling more consistent data exchange, improving production efficiency, and shortening development time. For instance, a Taiwanese semiconductor packaging equipment manufacturer reportedly reduced the development time by 80% through such software.<br>Having a more standardized set of tools and technologies can also help reduce integration complexity, ultimately improving throughput and agility. Long-term maintenance is also simplified while fab lifespan is lengthened.<br>Simulate first, commit second<br>Before committing to a costly piece of equipment that may not have a meaningful impact on productivity and output, organizations should consider the use of digital twin technology. Simply put, a digital twin is a dynamic virtual replica of a physical system. This allows engineers to move from guesswork to predicting the most probable results.<br>Simulating the addition of new equipment and its impact on existing processes and systems lets organizations test a myriad of potential scenarios and identify potential issues and bottlenecks before pulling the trigger. And with fab equipment designed to last a long time, companies can use digital twinning to continuously test and reconfigure facilities and processes so that their fabs do not become obsolete.<br>Across the manufacturing sector, digital twins are becoming an increasingly important tool for supporting industrial AI initiatives.<br>AI’s growing role<br>The importance of AI in semiconductor manufacturing grows as wafer sizes shrink. Costing $30,000 per two-nanometer wafer, a small drop in yield can result in millions in lost revenue. The conversation about AI in...