Many smartphone users love the depth and quality of information they receive from their AI-powered assistants. Others love to use features that rely on AI such as live translation, photography-related AI, website and message summaries, and more.
Demand for memory chips force price of limited supply higher
The incredible demand for AI has forced the price of memory chips to skyrocket. AI data centers use 70% of the world’s memory chips. Foundries that manufacture cutting-edge chips like TSMC are so busy that the firm’s best customers, such as Apple and Nvidia, get the first shot at using TSMC’s 3nm capacity. That leaves other tech firms like Ericsson, who use memory chips for network equipment, on a waiting list paying higher prices for the silicon components they need.
Ericsson uses application-specific integrated circuits (ASICs) for radios and baseband gear. It is interesting that networking gear uses chips that are a node behind smartphones and AI hardware, which would be 5nm for the most part. However, mobile networks demand chips that are just behind smartphones and AI workloads.
The first smartphone powered by a 2nm application processor was released this year
It’s been a while since we discussed why lower process nodes are so important. When process node numbers decline, it means that the feature size for the chip gets smaller allowing the transistor density to increase, which leads to improved performance and energy efficiency.
Ericsson and Nokia are trying to get customers to renegotiate contracts
Narvinger says, “Right now, many of the AI workloads are competing for the same wafers that we also are interested in.” The hope is that as AI firms move to the 2nm process node in a few months, TSMC is able to lower lead times reducing production bottlenecks allowing prices to drop. But just in case none of this happens, Ericsson has reportedly approached its customers seeking to renegotiate the deals it has made at lower prices.
AI is driving up the demand for semiconductors in general. We are seeing extremely high demand for some of these components, and that drives up prices for us.
Per Narvinger, head of Ericsson’s mobile networks business group
If Ericsson can’t convince its customers to pay more for its networking equipment gear, the company’s profit margins will shrink. AI is already being blamed for the drop in head count at Ericsson, which employed as many as 105,500 people back in 2022, and this has declined to 88,000 three months ago.
Per Narvinger, Ericsson Executive. | Image by Ericsson
Just like Ericsson seeking to renegotiate contracts with its customers to adjust for the higher chip prices it has to pay, manufacturers might be forced to raise the prices of their handsets. For those of you who are not AI fans, you have another reason not to like artificial intelligence.
Nokia CEO Hotard says many customers understand why they must pay higher prices
Ericsson’s chief rival in networking equipment is Nokia, and the latter’s CEO, Justin Hotard, also has seen lead times grow and its costs rise. Like Ericsson, Nokia has been talking to its customers trying to get them to agree to renegotiate their contracts and raise prices. Nokia’s Hotard says that when it comes to asking them to pay higher prices because AI has forced memory chip costs to rise, “There are many customers that understand that and accept it.”
BT (formerly British Telecom) CEO Allison Kirkby said that AI data centers require a huge haul of chips, the same silicon that many smartphone manufacturers need to obtain. If you recall in Econ 101, too much demand for a limited supply of a product leads to higher prices, which explains the inflation in memory chips and the resulting higher prices for consumer devices like smartphones.
#Prices #smartphones #mobile #infrastructure #rise #due #heavy #demand #memory #chips