--- title: "Is Memory Critical for AI CPUs? Korean Media: DRAM Shortage to Persist for Another Year" type: "News" locale: "en" url: "https://longbridge.com/en/news/284989898.md" description: "AI inference architectures are reshaping the computing landscape, with a surge in CPU memory demand extending the DRAM supply gap into 2027. As the focus of the AI industry shifts toward inference, CPUs have become major consumers of memory, tightening global DRAM supplies. Intel's latest AI CPUs will feature up to 400GB of DRAM, and this growing demand suggests the memory supercycle could extend to 2027. Rising spot prices for DDR5 reflect these market shifts. The transition to the AI inference phase is the fundamental driver behind the surge in CPU memory demand" datetime: "2026-05-03T09:40:53.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/284989898.md) - [en](https://longbridge.com/en/news/284989898.md) - [zh-HK](https://longbridge.com/zh-HK/news/284989898.md) --- # Is Memory Critical for AI CPUs? Korean Media: DRAM Shortage to Persist for Another Year AI inference architectures are reshaping the computing landscape, with a surge in CPU memory demand extending the DRAM supply gap into 2027. On May 2, according to a report by the Korean media outlet Sedaily, as the focus of the artificial intelligence industry shifts from training to inference, **CPUs are becoming new major consumers of memory, further exacerbating the tight global DRAM supply situation**. Citing industry sources, Korea's Securities Daily reported that **current DRAM market supply is approximately 10 percentage points lower than demand**, with the supply-demand imbalance unlikely to ease in the short term. The report stated that Intel's newly released AI CPUs are expected to feature up to 400GB of standard DRAM, four times that of standard CPU products. Meanwhile, next-generation AI chips from manufacturers such as Nvidia, AMD, and Google continue to drive up demand for high-bandwidth memory (HBM). > **Amidst this convergence of multiple demands, industry forecasts suggest that the memory supply capabilities of Samsung Electronics and SK Hynix will struggle to keep pace with demand growth. The memory supercycle, previously expected to end in 2026, is now likely to extend into 2027.** At the same time, **DDR5 spot prices have been the first to reflect market conditions.** According to data from Mirae Asset Securities, in April this year, the spot price for DDR5 (16GB standard) used in AI CPUs rose by 2.8% month-on-month, while the older generation DDR4 fell by 16% over the same period. The diverging trends indicate a widening premium for DDR5 prices. ## CPUs Become the "Scheduling Core" for AI Inference, Driving a Sharp Rise in Memory Demand According to reports, the AI industry's transition to the inference phase is the fundamental driver behind this surge in CPU memory demand. In the past, AI data centers built their computing infrastructure around GPUs, using HBM to support large-scale parallel training tasks, with server configurations typically consisting of one CPU paired with eight GPUs. However, as AI inference demand expands rapidly, especially with the rise of "Agentic AI," the role of the CPU is undergoing a fundamental transformation—**upgrading from a peripheral supporting role to an "orchestrator" that coordinates and manages various AI agent workflows.** The report noted that **the core of this role shift lies in "contextual memory" capabilities.** CPUs need to continuously track and call upon the output results of various agents, requiring them to be equipped with far greater memory capacity than before. An Intel executive stated at a recent earnings conference call, "In AI inference infrastructure, the compute architecture ratio has shifted from 1 CPU to 4 GPUs, and is narrowing further toward a 1-to-1 ratio." > **This means the proportion of CPUs in AI servers has increased significantly, leading to a multiplied increase in the consumption of standard DRAM.** ## GPU and CPU Compete for Memory, Revealing a Snowball Effect in Demand The report pointed out that the race for memory capacity is spreading from GPUs to CPUs, showing a "snowballing" expansion of demand. On the GPU side, Nvidia's next-generation AI chip, "Vera Rubin," configures 288GB of memory through eight HBM chips, while AMD's next-generation GPU, the MI400, reaches an ultra-large capacity of 432GB. Google's recently released eighth-generation Tensor Processing Unit, TPU 8i, is also expected to feature 288GB of HBM. On the CPU side, industry insiders revealed that CPU manufacturers are actively pushing for AI CPUs to be equipped with 300 to 400GB of DRAM. **Intel's "Xeon" series and AMD's "Epyc" series have already begun adopting high-capacity DDR5.** In comparison, standard CPU products have DRAM configurations of only 96 to 256GB, meaning the memory demand for new-generation AI CPUs has increased by up to four times. The simultaneous expansion of memory demand by both GPUs and CPUs has further intensified supply pressures in the already tight DRAM market. ## Exacerbated DDR5 Supply-Demand Imbalance, Supercycle May Extend to 2027 Prices for standard DRAM have risen by more than 100%, delivering unprecedented performance for the memory industry, while the supply-demand imbalance remains difficult to reverse in the short term. According to reports, an industry insider stated: > "Currently, DRAM market supply is about 10 percentage points lower than demand. With the surge in demand for standard DRAM beyond HBM, it is highly likely that the supercycle will extend from the previously estimated 2026 to 2027." Price trends in the DDR5 spot market confirm this judgment. According to Mirae Asset Securities data, the spot price for DDR5 (16GB standard) rose by 2.8% in April this year, while DDR4 plummeted by 16% during the same period. This price divergence directly reflects how strong demand for DDR5 from AI CPUs is reshaping the supply and demand structure of the memory market. The report pointed out that for Samsung Electronics and SK Hynix, against the backdrop of already highly constrained HBM production capacity, the additional explosion in demand for standard DRAM will further test their overall supply allocation capabilities. **The market generally expects that the boom cycle in the memory industry will last longer than previously predicted.** Risk Disclosure and Disclaimer Investing involves risks; please proceed with caution. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment decisions made based on this content are the sole responsibility of the investor. ### Related Stocks - [INTC.US](https://longbridge.com/en/quote/INTC.US.md) - [04335.HK](https://longbridge.com/en/quote/04335.HK.md) - [SOXX.US](https://longbridge.com/en/quote/SOXX.US.md) - [SOXL.US](https://longbridge.com/en/quote/SOXL.US.md) - [159516.CN](https://longbridge.com/en/quote/159516.CN.md) - [588780.CN](https://longbridge.com/en/quote/588780.CN.md) - [512760.CN](https://longbridge.com/en/quote/512760.CN.md) - [INTW.US](https://longbridge.com/en/quote/INTW.US.md) - [XSD.US](https://longbridge.com/en/quote/XSD.US.md) - [512480.CN](https://longbridge.com/en/quote/512480.CN.md) - [588170.CN](https://longbridge.com/en/quote/588170.CN.md) - [SMH.US](https://longbridge.com/en/quote/SMH.US.md) - [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md) - [AMD.US](https://longbridge.com/en/quote/AMD.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [SSNGY.US](https://longbridge.com/en/quote/SSNGY.US.md) - [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md) - [SMSN.UK](https://longbridge.com/en/quote/SMSN.UK.md) ## Related News & Research - [Intel Stock (NASDAQ:INTC) Surges as SambaNova Deal Clears Regulators](https://longbridge.com/en/news/284931575.md) - [Altera Brings Determinism to Physical AI Systems with Latest Release of FPGA AI Suite](https://longbridge.com/en/news/284800493.md) - [BUZZ-J.P. Morgan sees AI hardware investment widening beyond GPUs](https://longbridge.com/en/news/284193418.md) - [BUZZ-Street View: AI demand boosts Intel, but capacity limits loom](https://longbridge.com/en/news/283982891.md) - [Intel warns China of severe server CPU shortage as AI demand surges](https://longbridge.com/en/news/284144041.md)