{
  "id": 3397370,
  "title": "When will AI stop making memory more expensive? Probably much later than you hope",
  "url": "https://urgent.news/2026/08/26/when-will-ai-stop-making-memory-more-expensive-probably-much-later",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-08-26T01:30:09.000Z",
  "source": {
    "name": "e27",
    "slug": "e27",
    "url": "https://e27.co/when-will-ai-stop-making-memory-more-expensive-probably-much-later-than-you-hope-20260823/"
  },
  "original_language": "en",
  "account": "Microsoft's recent decision to raise Xbox console prices by $100 and $150, respectively, for 512GB and 1TB models marks more than a mere gaming headline. This move is tied closely to the growing demand for Artificial Intelligence (AI) infrastructure, specifically high-bandwidth memory (HBM), server memory, and flash storage required for AI development. According to Micron, data-center DRAM and NAND bit shipments in 2026 are expected to double their 2024 levels, while Samsung predicts demand for server memory, enterprise storage, and HBM will remain undersupplied through the second half of 2026. This shortage extends beyond data centers, impacting various products not directly linked to AI, such as phones, PCs, and even refrigerators, due to the shared manufacturing processes of wafer fabrication, cleanroom space, and equipment investments. While established producers like Samsung, SK hynix, and Micron prioritize AI, consumer electronics face the most significant challenges as they have less margin to absorb rising memory costs. For example, a mid-range smartphone's memory can account for 15-20% of its bill of materials, whereas flagship models have 10-15%. Low-end phone makers are particularly vulnerable, as they cannot absorb large increases. Meanwhile, PC manufacturers face a similar problem, with local AI demand raising their memory requirements just as costs rise. Microsoft's price increase serves as a stark illustration of what happens when a fixed hardware design meets a rapidly escalating input cost. The situation is further compounded in the automotive sector, where vehicles with advanced driver-assistance systems require significantly more memory and storage than average vehicles. Additionally, China's expanding semiconductor industry may provide some relief by reducing the demand for memory from major suppliers, but its impact is indirect and may not significantly alleviate the AI-memory shortage. While some relief is expected by late 2027, substantial output from new DRAM capacity is not anticipated until 2028, and broader normalization is expected around 2028.",
  "summary": "On August 1, Microsoft raised Xbox console prices by US$100 for 512GB models and US$150 for 1TB models. The company said storage and memory prices had risen more than 2.5 times and could roughly double again by fall 2027. That sounds like a gaming story. It is really an AI infrastructure story. The world is […] The post When will AI stop making memory more expensive? Probably much later than you…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}