The enormous increase in memory costs will also affect Nvidia customers. The group has reportedly begun notifying customers of price increases of more than 15% for finished products, all in the high-performance AI product segment. According to Bloomberg, the news has already been sent to the group’s main customers, including Microsoft, Alphabet, and Oracle.

In reality, Nvidia was not the first to act. In recent weeks, Apple had imposed significant price increases on notebooks and desktop PCs, citing the same reason: the costs of next-generation memory have soared, while the components remain extremely difficult to source.

The memory bottleneck?

The memory bottleneck is among the most important challenges facing the AI sector. Major manufacturers are facing significant backlogs, to the point that Apple itself had asked the US government for authorization to begin purchasing high-performance RAM from Chinese manufacturers as well.

Now the wave of high and steadily rising prices is also hitting Nvidia and its customers. According to Bloomberg, substantial price increases have reportedly been communicated for the high-end range of Nvidia products, which includes Vera Rubin and Grace Blackwell. The increases could exceed 15%, depending on the chip generation and the specific configurations requested by end customers.

Even more capex?

Another issue remains: the major companies in the technology sector have already invested hundreds of billions of dollars in AI, specifically in infrastructure. Spending more money, however, cannot make fast memory appear out of nowhere. Such memory is in high demand and is now inaccessible, in certain configurations, to the consumer segment.

The concern expressed in many analyses is that this bottleneck will not be easy to overcome, frustrating hyperscalers’ spending plans and their ability to continue supporting a cycle fueled precisely by their spending.

As a knock-on effect, Microsoft, Google, and Oracle have reportedly already notified customers with long-term contracts of possible price increases. This confirms the widespread nature of the problems, which begin with chip manufacturers—whose product supply is insufficient to meet demand—and then affect the entire sector. It is an exceptionally difficult situation, which the industry will attempt to overcome through substantial investments in new factories—investments that have already been approved by the sector’s major players.

For now, the AI segment’s greatest concern is being unable to spend the billions that have already been raised or are the subject of financing contracts that have already been signed.