top of page
hero-bg-08.jpg
Profet AI Insight

Know about the latest news and industry views through us

Profet AI Collaborates with Leading Semiconductor Clients through the Virtual Data Scientist Platform to Accelerate the Adoption of Citizen AI and Explore More AI Applications

The most crucial industries of the 21st century and the foreseeable future.

The semiconductor industry is undoubtedly one of the most crucial industries of the 21st century and the foreseeable future. Its widespread applications span computers, mobile devices, industrial applications, vehicular communication, autonomous vehicles, electric vehicles, AIoT, and 5G. In 2021, the semiconductor industry continued its robust growth, extending the supply-demand tightness from 2020. The World Semiconductor Trade Statistics (WSTS) projected an 8.4% annual increase in the global semiconductor market value in 2021.


In summary, industries such as electric vehicles, artificial intelligence, cloud services, and mobile devices will continue to see upward development in chip and panel industry chains with Taiwan holding a leading global position in this sector.


Taiwan's semiconductor achievements are not accidental but are the result of past strategic planning and actions.


The story dates back to the 1970s and 1980s when leaders like then-Premier Sun Yun-suan and Minister Lee Kwoh-ting envisioned the future, establishing science parks, recruiting international talent, and founding companies like United Microelectronics Corporation (UMC). Dr. Morris Chang, then-president of the Industrial Technology Research Institute, established Taiwan Semiconductor Manufacturing Company (TSMC), enhancing the local integrated circuit manufacturing capabilities.


Dr. Morris Chang's consideration of Taiwan's competitive edge led to the creation of a new business model: the world's first dedicated semiconductor foundry. This new division of labor within the semiconductor industry and the success of this differentiation in the foundry market underlined Taiwan's solidification in semiconductor manufacturing. This evolution brought prominence to companies like TSMC, ASE Technology Holding Co., Ltd., and GlobalWafers.


Around 2013-2014, with the global advocacy for Industry 4.0, smart manufacturing, and later digital transformation, Profet AI's team began promoting these concepts and AI solutions across various industries in Taiwan, including electronic manufacturing services (EMS), machine tool equipment, semiconductors, and the optoelectronics industry chain.


Semiconductor manufacturing processes are highly precise and intricate, continuously challenging the limits of manufacturing technology.


Over decades, the semiconductor industry has evolved into a complex and professional ecosystem, starting from crystal growth, slicing wafers, IC fabrication through thin-film deposition, photoresist, exposure, development, etching, and stripping processes. The layered structure of IC circuits requires multiple iterations of mask input, pattern creation, and the formation of lines and components.


With technological advancements, the semiconductor process technology continues to miniaturize, constantly breaking new records in transistor density.

With the continuous evolution and miniaturization of semiconductor process technology year by year, the density of transistors keeps setting new records. (Image source: TSMC official website)


With technological advancements, the semiconductor process technology continues to miniaturize, constantly breaking new records in transistor density. After wafer fabrication, the product moves to packaging and testing facilities, where it undergoes WAT testing, followed by slicing, bonding, packaging, and final testing.


The leading semiconductor and packaging/testing companies have long incorporated big data analytics into their DNA.


These processes are supported by extensive, high-end equipment, stringent environmental controls, material purity, quality consistency, and automated, monitored manufacturing. Historically, manufacturing and environmental parameters have been analyzed to ensure product stability. In cases of yield issues or efficiency improvements, various data analysis techniques assist decision-making. Compared to other industries, the semiconductor sector has a more mature capability in data governance and analysis.


For decades, the industry has cultivated top-tier talent in quality enhancement, efficiency improvement, preventative maintenance, and environmental conservation, integrating AI technology into leading companies' data usage culture.


 

Having hired dozens to hundreds of data scientists, companies yet still find the output of AI-related projects to be too slow.

Despite having hundreds of data scientists, these resources remain scarce for high-revenue clients. While AI machine learning technology is relatively mature, its application in industries is just beginning.


Complex semiconductor processes and manufacturing quality departments continuously explore how AI can be applied. In successful cases, deep involvement of domain experts is key, following a three-step process:

  1. Problem Transformation: Domain experts translate problems into data analysis.

  2. Data Preparation and Model Evaluation: Data scientists develop programs for data processing and model creation, evaluating model viability.

  3. Application and Expansion: Explore and expand the application of AI in various scenarios.


This cycle repeatedly occurs in enterprises, but only about 10% of AI application scenarios are successfully implemented, overburdening data scientists.


The Virtual Data Scientist Platform can be utilized after just 2-hours training. AI topics exploration process boosted by 5 times.

In leading enterprises with established data lakes, process integration engineers analyze data post-production for improvement and optimization. Profet AI AutoML Platform becomes one of the analytical methods, enabling various improvement suggestions based on data.


Typical manufacturing clients focus on:


  1. Identifying critical factors in quality anomalies.

  2. Virtual measurement/simulation of quality goals.

  3. Recommending startup parameters.

  4. Assisting in material formulation development and material property simulation.


Applications in the semiconductor industry, like thin film and exposure processes, have seen practical use and efficacy. In some leading enterprises, Profet AI's products facilitate real-time assistance, advancing towards R2R applications.


In the packaging and testing industry, clients use the Virtual Data Scientist Platform to explore AI scenarios in complex processes, accelerating data monetization and benefits.


 
Unlike traditional AI project outsourcing, Profet AI's solution encourages clients to boldly explore AI applications, safeguarding critical business secrets.

The transition from one-off outsourced projects to an internal virtual data scientist platform is a trend. The platform's model configuration management enables future understanding of internal AI projects, detailing data used, features selected, objectives, chosen algorithms, and model scores. Enterprises use Profet AI's platform for model configuration management, preserving and disseminating AI modeling experience.


 
Domain experts plus Profet AI's Virtual Data Scientist Platform are key to significantly enhancing the semiconductor industry's competitiveness.

The key to sustainable development in the semiconductor industry is maintaining technological leadership and continuously optimizing operational performance.

Successful enterprises stand out often due to their unique corporate culture, a reflection of how a group thinks and acts. The success of AI involves a cultural shift in problem exploration and data application. Historically, data analysis was the purview of specific individuals, but Profet AI envisions a culture where everyone in an enterprise can create value with AI, turning every employee into an AI contributor.


Through Profet AI AutoML Platform, manufacturing industries can quickly enhance their personnel's analytical skills and quality, expanding AI's internal application scope and improving operational performance, ultimately expanding their competitive advantage.


 
 

Comentários


bottom of page