MacBooks are expensive because of AI: Memory chips, data centers, and the price ordinary users are paying
When Apple announced a price increase for MacBooks and iPads on June 25, Tim Cook explained in a short sentence: not tariffs, but a lack of memory chips. It sounds simple, but if you stop and think, the story behind it is quite interesting. Why is the memory chip missing? Who's buying it all? And why should the buyer of a personal laptop have to bear the consequences of something happening inside the data centers of OpenAI, Google or Microsoft?
The memory chip is disappearing to somewhere
There are two most common types of memory chips in electronic devices: dynamic random access memory (DRAM - Dynamic Random Access Memory) used to handle temporary tasks while the device is running, and NAND flash memory used for long-term storage such as SSD hard drives, memory in phones, MacBooks. Both are in a situation of record supply and demand.
Memory chips are becoming a scarce commodity at this time
The numbers speak
The consequences on the market are very clear through contract price data. In the first quarter of 2026, contract DRAM prices increased about 90 to 95% compared to the previous quarter, according to TrendForce data. NAND flash prices are expected to continue to increase by 70 to 75% in the second quarter. In the first half of 2026, memory chip prices have increased from 50% to 200% depending on type, pushing server costs up to more than 125%.
Samsung is one of the leading HBM manufacturers today Consumer electronics manufacturers have been trying to absorb these price increases for months, hoping the situation would ease. But as the trend persists, they have no choice but to pass on costs to buyers. Apple is a strong enough company to delay this decision longer than most, but at some point the wall has to break.
Who is winning in this game?
What's interesting is that the memory chip crisis isn't bad for everyone. Samsung, SK Hynix and Micron, the world's three largest memory chip manufacturers, are directly benefiting. Higher prices, strong profits, and long-term contracts with AI data center operators ensure stable revenue for years to come.
When will this end?
The answer is not very positive. To increase the supply of memory chips, it is necessary to build more factories. TwitterHere a chip factory takes three to four years and costs tens of billions of dollars. Samsung, SK Hynix and Micron are expanding capacity but the process cannot be accelerated as desired. According to most analysts, the shortage of supply and high prices could last until at least the end of 2027.
And users will be the ones to bear the consequences of memory chip shortage What's interesting is that we've seen a similar scenario once, albeit on a much smaller scale. Around 2020 to 2022, a wave of cryptocurrency miners bought up GPUs worldwide, pushing graphics card prices to two to three times the original price. Many people who want to buy GPUs to play games or do graphics cannot buy them because miners have already bought them all. Finally, when that wave passed, the market recovered. With AI, the question is whether this wave will "pass", or if it is a permanent structural change.