AI can be a very controversial topic. On the one hand it can help with the analysis of large sets of data and may find patterns a human would not be able to find, but on the other hand it is used to replace human workers, especially in fields like customer service and support, driving people into unemployment and, in some cases, drastically worsening the experience for users if they need help. But one thing that is often overlooked in this debate, is how AI and especially the recent AI-Boom increases the prices of consumer electronics, especially those on the higher end like computers or smartphones.

What does AI need?

We need to differentiate here, because there are many different forms in which we might interact with AI. For this I am going to focus on probably the two most known variations: Large Language Models (LLMs) and Generative AI (GenAI).

A LLM requires the most resources during the training process, since it typically does not change once trained and optimized and just needs to output text. Examples of LLMs are OpenAI's GPT-3.5 or GPT-4 models used by ChatGPT.

GenAI on the other hand also uses a lot of resources when actually generating content, since they are much more versatile and can for example also generate images or videos. DALL-E 2 is an example of generative AI.

What the companies behind those AI models need in any case, are computing resources, especially with the increase in complexity the models have made. Especially the demand for RAM has increased in the last few months, and that is where the problem for consumers starts.

The RAM shortage

One thing especially high in demand right now is computer memory (RAM). Since the manufacturing capacities are limited, setting up new production lines is very costly and the few manufacturers that can produce high speed memory chips don't see that the current demand spike will last long enough to economically justify the investment, it has become increasingly more costly to get memory.

This does not only affect the big AI companies, but also makes the prices on the consumer market skyrocket. The price for a DDR5-6000 memory kit with 64GB for example has increased by almost 400% over the last 4 months, and the ceiling has not yet been reached there:

Source: https://pcpartpicker.com/trends/price/memory

Another example of how this affects consumers is that one of the Big 3 memory manufacturers, Micron, has decided to shut down their consumer brand Crucial in favour of selling the chips directly, both to third party manufacturers of memory modules and to data center customers. In their words they are 'helping' the customers with this, but since we all know how businesses work, they will just sell to the highest bidder, thus further increasing the prices for everyone.

The Apple case

Another field where we will probably see price increases soon, will be Apple products. For years Apple has been the largest customer of chips from Taiwanese manufacturer TSMC and managed to save a lot of money compared to competitors due to a deal between those companies. But this deal has now expired and since Nvidia has taken over the crown of the largest and most important customer, Apple will no longer get the sizeable discounts they had so far. Instead they will have to fight a price war against Nvidia, who uses the chips mostly for their AI products, which in turn are mostly bought by huge corporations and not by individual customers like most of Apples products are, meaning they have a lot more money to throw at this problem, further increasing the prices Apple has to pay. Combined with the above mentioned price increases for memory, it would be difficult for Apple to not increase the prices significantly or loose a huge part of their margins.

* * *

Sources:

Large Language Models (LLMs) vs. Generative AI: What’s the Difference? | Coursera

https://pcpartpicker.com/trends/price/memory/

Micron says it's 'helping' consumers — by not selling RAM to consumers | PCWorld

Apple's sweetheart deal with its chipmaker comes to abrupt end | Cult of Mac

🗃