Samsung has reportedly raised its memory prices in response to the AI boom, as Morgan Stanley cut ratings on the likes of Hewlett Packard Enterprise (HPE) due to margin pressures stemming from the current “memory supercycle.”

Reuters reported that the Korean giant raised prices of its dynamic random access memory (DRAM) wares by 60%, citing two persons with knowledge of the matter.

In the wake of the rises, financial analysts from Morgan Stanley downgrade HPE, HP, and Dell due to what they dubbed as a “memory supercycle,” seeing DRAM and NAND flash memory prices spike between 50% and 300% in the past six months.

HPE was downgraded to Equalweight from its previous Overweight rating due to memory cost inflation pressures expected in 2026, compounded by issues such as the ongoing Juniper integration.

“We see risk of integration of legacy HPE and Juniper operations, with less gross cost savings flowing through to the bottom line … combined with a weaker than expected outlook for HPE's new networking business and incremental margin pressure on the back of rising memory costs,” wrote analyst Erik Woodring.

Rising costs have been exemplified by Samsung’s reported price hikes, with Tobey Gonnerman, president of semiconductor distributor Fusion Worldwide, telling Reuters that Samsung contract prices for 32GB DDR5 memory chip modules escalated to $239 this month, up from just $149 in September.

Gonnerman added that Samsung has also raised the prices of its 16GB and 128GB DDR5 chips by roughly 50%, to $135 and $1,194, respectively, while prices of 64GB and 96GB DDR5 chips have increased by more than 30%.

Rising interest in inference

Manufactured by the likes of Samsung and South Korean peer SK Hynix, as well as America’s Micron, DDR memory chips are used in servers and more to enhance computing performance by temporarily storing data and managing high-speed data transfers and retrievals.

This is especially helpful for AI inference workloads, a growing use case going into 2026, which has seen Oracle and the hyperscalers turn to Nvidia Dynamo in recent weeks, and AI storage vendor Vast Data team up with Google Cloud for user inference needs.

A big consumer of memory chips is AI giant OpenAI, which recently signed a contract with both Samsung and SK Hynix for a cumulative 900,000 DRAM wafer starts per month – around 40% of global DRAM output.

Despite questions over its finances, OpenAI has made big commitments to fuel its growing Stargate data center venture. It also has the advantage of not being a hardware vendor, with Morgan Stanley’s recent report warning that hardware original equipment manufacturers (OEMs) such as HPE historically face gross-margin compression six months to a year after memory costs begin rising.

Analyst Woodring warned that OEMs are usually able to absorb only about 70% of the rise in memory costs due to inflation, even with mitigation strategies such as increasing device prices, cutting other bill-of-materials expenses, or reducing operating costs.

Overall, Morgan Stanley expects a median 60-basis-point decline in global OEM margins next year as memory demand outpaces production.

The apparent unease around the memory has reportedly already seen some panic buyers rushing to secure hardware before prices skyrocket further. According to DigiTimes, OEMs including Asus and MSI have been aggressively stockpiling in preparation for a potential rise in demand.

The impact is already being felt at the consumer level, with some PC builders already reporting price rises of around 95% in just a matter of months.