The results of this year's Nobel Physics and Chemistry Awards are unexpected.The physics award unexpectedly awarded the field of machine learning. The research of the two scientists in the figure above have played an important basic role in machine learning algorithms, and machine learning is the core of today's generation AI.

Three winners of the Chemistry Award are also closely related to AI. Two of them are from Google DeepMind. The AlphaFold AI model invented by them solves a series of difficulties in the predicting amino acid and protein structure, which greatly enhances the accuracy of structural prediction.

This year's Nobel Prize is far away from basic theory, and it has been awarded to experts in popular applications, which has attracted a lot of heated discussions and controversy, but it is enough to prove that AI influence is strong.Especially for this year's IC industry, AI looks particularly important, and even independently supports market growth.

AI becomes the largest industry incremental source

Course the United StatessemiconductorAccording to the monthly statistics of the Industry Association (SIA), the latest August data shows that global semiconductor sales have reached a high of 53.1 billion US dollars, an increase of 20.6%year -on -year, and an increase of 3.5%month -on -month.This is also the year -on -year growth of the 10th consecutive month, and the application that drives the incremental cycle of this round is undoubtedly AI.

worldwidesemiconductorSource of monthly revenue: SIA

Among major IC companies,NivineAs well asBroadcomAs well asMarvellWaiting for the beneficiaries, the performance has greatly surpassed other general -purpose IC companies.The current consumer electronics still look forward to recovery, AI related opportunities take over, becoming the main theme of peak season growth.In terms of industrial chain in Taiwan, China, the server -pulling effect has been stronger than consumer electronics. In the past, the growth driven by products such as iPhone and Mac computers has been applied to AI applications.

In the fourth quarter,NivineNew BlackwellchipIt is expected that a large number of shipments will be shipped. There will be about 450,000 pieces of Morgan Stanley prediction.NivineBring a income of about 10 billion US dollars.Based on the chip server, the unit price will also be greatly increased, which will promote the further growth of ODM enterprises' revenue.

In general, the tone of the active force of AI playing the demand for this year's peak season has been well played, all kinds ofComponentThe market will also be differentiated accordingly, especiallystorageThe market that has attracted much attention from device and MLCC is also under dual influence of consumer electronics and AI applications.

storageDevice: AI support demand, consumer category is weak

In the AI ​​server, HBM high -band wide memory is a hard match for GPU, mainlystorageThe competition of the original factory is also concentrated.The production rate of HBM is significantly lower than that of traditional DRAM, and has higher occupation of production capacity.Based on production capacity, the overall price support of DRAM this year is stronger than NAND flash memory, which reflects the overall influence of AI on the memory market.

In terms of the consumer market, because it is still looking forward to recovery, mainstreamstorageThe price of the device is difficult to boost.In mid -September, the investment bank Morgan Danley predicted that the memory would usher in the "cold winter", and pointed out that the demand for mobile devices and PCs for DRAM is slowing.According to the data published by IDC, the end of the global PC shipments in the third quarter, and the rise of the iPhone 16 series in the third quarter, this view has a realistic basis.

At the end of September, the module factory Kingston started the price reduction and clear inventory of the low -level products, which also increased the amount of this prediction.In addition, analysts pointed out that Taiwan, China, ChinastorageThe inventory of the module factory is generally as high as 11 months, and the inventory of low -level products will also be reduced in price. Others, such as the movement of Fitting, Creation, Qunlian and other factories, also need to pay attention.

According to a series of market expectations, the latest report of TrendForce Juanbang Consultation predicts that DRAM prices will slow down in the fourth quarter.The original factory will still seek room for price increases due to production capacity, but with the current demand, the downstream is difficult to accept, which will make the price flat or even fall.Those who can continue to increase in a large extent are only HBM necessary for AI applications.

MLCC: It will usher in a wider demand next year

AI servers, AI PC and other products have greatly improved the computing power, and the demand for MLCC will also increase in terms of quantity and quality.Passive factoryVillage, Guoshi, Huaxinke, etc. are optimistic about the future AI opportunities. This year's boom will be concentrated in cloud devices such as servers, and next year will extend to PC andcell phoneEnd side -side consumer goods.Correspondingly, the capacity utilization rate of the original factory will also increase.

But in the fourth quarter, exceptNivineIn addition to the server demand brought by Blackwell's large -scale shipment, PC andcell phoneThey all reflect the insufficient tide of the machine and have a certain impact on MLCC demand.

TRENDFORCE Jibang Consulting recently pointed out that the number of laptop forecast orders released by ODM has decreased by 5%-8%on average. It is estimated that the total number of MLCC suppliers in the fourth quarter was approximately 1.205 trillion yuan, a decrease of 3.6%from the previous month.

It can be seen that the main MLCC incremental force of this year is server, PC and other consumer electronics, which is expected to officially welcome AI empowerment next year, and the demand will be greatly improved at that time.

Conclusion

This year, the Nobel Prize in Physics and Chemistry was awarded to scientists in the AI ​​field, which was surprising, but it also reflected that the AI ​​had been valued by the academic community and would have a deeper and long -term confidence in AI.andComponentAs the carrier of the AI ​​algorithm, it will also usher in a long and stable demand.Whether it is the cloud or end -side device, it will usher in the opportunity to renew and empower.