Server prices are set to increase as the AI memory shortage continues to bite, with Samsung looking the likely winner from proceedings.

The Register reported various channel sources expecting price increases of around 15% for servers, with all manufacturers expected to get in on the hikes, including Hewlett Packard Enterprise (HPE), HP, and Dell. Those three companies saw their ratings downgraded by Morgan Stanley in response to what it dubbed the “memory supercycle,” which has seen dynamic random access memory (DRAM) and NAND flash memory prices spike between 50% and 300% in the past six months.

A price rise of 5% is also expected to hit PCs, while a separate report from WCCF Tech claimed sources are predicting memory shortages across all products, including DDR5 and DDR4 memory, to last into Q4 2027.

An expected winner from the turmoil is Samsung, with The Korea Herald expecting Samsung Electronics to reclaim its status as the world’s leading DRAM provider. Sources claimed the division is to account for 15.1 trillion won ($10.2 billion) of Samsung’s projected 18 trillion won ($12.2 billion) operating profit across the current quarter, up by 422% year on year.

That momentum saw Samsung Electronics spin out a new unit solely focused on high-bandwidth (HBM) chips such as its 40 Gbps GDDR7 offering, as led by Hwang Sang-joon, head of Samsung’s DRAM product and technology group.

A spokesperson for Samsung Electronics' memory business directed SDxCentral to its recent earnings report in response to questions.

A Samsung Networks representative, meanwhile, said the company’s networking division “has always embraced openness and optionality, working closely with CPU and GPU partners like Intel, AMD, and Nvidia to unlock AI for its operator customers.”

“Samsung fully understands the critical role of chips in a RAN network, which is why it develops in-house chips that set the company apart from others. However, this strategic capability is separate from Samsung’s memory business,” they added.

The OEM response

At HPE’s recent Discover Barcelona event, EVP & GM of HPE Financial Services Maeve Culloty told an audience of press and analysts that the memory issue is “causing customers to deliberate where are they going to spend the budget and when are they going to spend it … so the shape and nature of data centers is changing [across] the infrastructure stock tying together, storage, server, networking.”

A separate panel saw Simon Ewington, HPE SVP channel and partner of ecosystem sales say that HPE “continues to work with partners and customers to make sure we address those as best we can.” Ulrich Seibold, VP for global partner hybrid cloud solutions & service providers, added stability is all but confirmed for Greenlake, which guarantees prices for most components of the cloud service.

These comments were made before HPE’s most recent earnings call, in which CEO Antonio Nero noted the firm already increased prices in November, adding his firm had “strong capabilities in our supply chain to secure the allocation of components we need,” as well as “discipline in passing through the cost through our pricing.”

The CEO also drew comparisons to the COVID period, suggesting the company had learnt some lessons on the allocation of supply from the pandemic.