The semiconductor industry is changing rapidly thanks to artificial intelligence.

At the conference, opinions affirmed that AI and semiconductors are currently the pillars for the future of the digital economy. In particular, the two elements “AI” and “semiconductor” go hand in hand. Most obviously, AI helps automate the semiconductor manufacturing process, predict and detect product defects, improve production quality and efficiency.

Mr. Christopher Nguyen - CEO of Aitomatic gave an example, by 2030, some manufacturing plants, especially advanced manufacturing facilities, will require more stringent standards. For example, in the plasma processing process, parameters such as fuel diameter, pressure, temperature and dozens of other factors must be ensured to ensure near-absolute accuracy. AI will contribute to ensuring this accuracy.

“AI cannot develop without semiconductors , and vice versa, the semiconductor industry is changing rapidly thanks to advances in AI. It is a symbiotic relationship where both push each other forward,” he said.

The semiconductor industry is changing rapidly thanks to artificial intelligence photo 1

Mr. Christopher Nguyen - CEO of Aitomatic spoke at the Workshop.

Regarding the general technology picture, Mr. Christopher Nguyen cited Moore's law, affirming that the development speed of AI and semiconductors is very fast. Every 18 months, microprocessor technology has significant improvements.

As for the market, the world is witnessing remarkable growth with demand for AI processing chips expected to continue to increase sharply in the coming years. Countries such as the US, China, Japan, and South Korea are stepping up investment in this field. The race between leading countries in technology is extremely fierce.

In the field of chip manufacturing, Ms. Anna Goldie - Senior Research Scientist, Google - commented that while the computational needs of AI are growing exponentially, hardware capabilities are not keeping up, creating a growing gap. To solve this problem, new AI technologies such as AlphaChip - an AI chip design method - have been introduced. She said that thanks to the application of AI, the chip design process is incredibly accelerated, while helping to reduce costs and optimize performance.

“To fully exploit the potential of AI, we need to shorten chip design cycles, improve algorithms and make the most of data. In the future, AI will not only help improve hardware but also contribute to shaping breakthroughs in many other fields, from healthcare, finance to industrial manufacturing,” said Ms. Anna Goldie.

Specifically, Ms. Anna Goldie introduced the AlphaChip method that uses AI to optimize the layout of components on the chip, helping to reduce latency, save power and optimize production area. AI can improve the chip design process by shortening the time and improving product performance. AlphaChip has been applied to recent generations of Google TPUs, bringing significant efficiency compared to traditional design methods.

Meanwhile, Mr. Tran Thanh Long - Professor at Warwick University - shared more about the efforts around the world that are helping to increase the power of AI and semiconductor technology. For example, he mentioned how to use memory stores and Bayesian theory to improve the performance and scalability of artificial intelligence (AI). Memory stores help AI remember information for a long time and use past data to optimize decisions.

“Bayesian theory helps AI adjust its prediction probabilities based on new data, helping the system learn faster and more efficiently. This combination reduces the requirement for computational resources while still ensuring high accuracy,” said Mr. Long.

In addition, this approach helps AI operate more smoothly in areas such as healthcare, industrial manufacturing, and automation. In particular, AI can process data better without relying too much on large data centers, saving costs and resources. As a result, systems are smarter, more efficient, and self-adjusting without needing huge amounts of data .

Ms. Ngan Vu from Google DeepMind introduces a research direction that proposes using Circuit Neural Networks to create efficient logic circuit designs. By applying simulated annealing algorithms and other optimization techniques, her team of experts is aiming to shorten the circuit design cycle from idea to actual product.

One of the major challenges is balancing the accuracy and performance of circuits, ensuring that designs not only work accurately but also save resources. However, if the gap between AI software and hardware can be narrowed, it will open up many new opportunities in the semiconductor industry. “Applying AI to circuit design promises to change the way the semiconductor industry operates, helping to speed up the development process and deliver more optimal designs,” said Ms. Ngan Vu.

Comment (0)