Google: Research shows that transistor costs will no longer decrease after 2012, but have decreased by an average of 30% per generation before that

According to recent news, the latest research by Milind Shah, head of Google's integrated circuit packaging department, shows that after the 28nm process node in 2012, the average cost of transistors has stopped decreasing, which naturally leads to high costs and prices of chips.

He pointed out that before 28nm, the average cost of transistors would decrease by 30% with each new generation of process, but after 28nm, each generation remained basically unchanged, or even increased slightly, until the latest 3nm, which only decreased slightly compared to 5nm, but the decrease was very small.

In fact, as early as 2014, Zvi Or-Bach, CEO of 3D semiconductor integration company MonolithIC 3D, came to the same conclusion. It can be said that "Moore's Law" has stopped at the 28nm node and is no longer truly effective.

One important reason why the cost of transistors or chips remains high is that advanced processes are becoming more and more complex, and the required manufacturing equipment and factories are expensive. For example, an advanced lithography machine is estimated to cost $200 million, and an advanced wafer fab costs $20-30 billion.

Because of this, the entire industry is no longer overly keen on pursuing the most advanced technology, but instead vigorously develops various packaging technologies to better balance performance, power consumption, cost, etc.

Of course, advanced technology is still very important, but it is no longer for the purpose of making chips cheaper, and fewer and fewer companies and products will be able to afford it.

Zikuai Technology