The AI-Driven Memory Crisis: Why RAM Prices Are Skyrocketing and What It Means for Tech

December 4, 2025

Memory price inflation is an unavoidable reality, and its impact is only beginning to be felt across the industry. Just last month, when I was assembling a new PC with components acquired earlier this year, a 64 Gigabyte T-Create DDR5 memory kit cost $209. Today, that identical kit is priced at a staggering $650.

Recent developments underscore this trend: Raspberry Pi has announced increased prices for its single-board computers, and Micron is discontinuing its Crucial brand for consumer RAM and storage, reducing competition. Even Samsung is reportedly struggling to source RAM for its own smartphones. Smaller vendors like Libre Computer and Mono are experiencing memory price increases of two, three, or even more times, despite not even utilizing the very latest RAM technologies.

PC builders appear to be among the first groups to feel the full brunt of this inflation. Data from PC Parts Picker reveals drastic price hikes, with DDR4 RAM skyrocketing from approximately $30 to $120, and 64GB DDR5 kits surging from around $150 to $500. However, these impacts are only just beginning to ripple into other market sectors.

Libre Computer recently noted on Twitter that a single 4-gigabyte LPDDR4 memory module now costs $35 – exceeding the combined price of all other components on one of their single-board computers. Operating at a loss is unsustainable, meaning current product batches will inevitably lead to either increased prices or product line discontinuations.

Smaller companies are disproportionately affected. Even Raspberry Pi, likely with healthier margins, has already raised SBC prices and introduced a 1GB Pi 5 model – perhaps an unintended catalyst for developers to reconsider memory-intensive JavaScript frameworks and optimize for lower memory footprints.

Ultimately, virtually any device containing memory – from cameras and gaming consoles to tablets – will face these price pressures. It's almost unbelievable, but Apple's notoriously high memory upgrade costs now appear to be surprisingly competitive within the current market landscape.

The root cause of this market upheaval is the massive buildout of AI datacenters. While the possibility of price fixing, reminiscent of past decades, remains a topic for speculation, the fundamental issue lies in the limited number of companies controlling global memory production.

These dominant manufacturers have recognized the potential to earn billions more by exclusively producing RAM for AI datacenter products, thereby sidelining the broader consumer and enterprise markets. Consequently, they are phasing out consumer memory lines to dedicate all production capacity to AI-specific demands.

Even established GPU board manufacturers are feeling the squeeze. Nvidia is reportedly no longer supplying memory alongside its chips as it once did, effectively telling partners, "good luck, you're on your own for VRAM now!" This situation is particularly ironic, given Nvidia's immense profitability from the very AI boom driving these changes.

Despite the grim outlook, some optimists foresee a "silver lining," suggesting that a potential AI bubble burst would flood the market with cheap hardware. While this might have been true in past tech cycles, the current situation is different.

The majority of RAM produced for AI is either integrated into specialized GPUs incompatible with standard computers or designed into proprietary memory modules (e.g., HBM) that are unsuitable for consumer PCs. Furthermore, the GPUs and servers deployed for AI require substantial power and specialized cooling, forming part of massive, complex systems far removed from consumer-grade equipment like a classic Dell R720, which only requires basic air circulation and a wall outlet.

This situation signals a potentially strange era where the PC building hobby becomes untenable, single-board computers become prohibitively expensive, and individuals who haven't already stockpiled components find themselves at a significant disadvantage.

Even major players like Lenovo have reportedly admitted to stockpiling RAM, echoing the toilet paper shortages of 2020, but on a corporate scale. Limited supply means financially capable companies are buying up available stock, hoping to mitigate shortages that may be prolonged precisely because of such stockpiling.

It's not far-fetched to imagine some companies resorting to scavenging memory chips from other systems for inventory, a practice reminiscent of "dosdude1," especially if RAM prices continue their ascent. The alternative is simply ceasing production. This scenario bears unsettling resemblances to the global chip shortages of 2021-2022, which severely impacted smaller businesses.

It's disheartening to witness a similar crisis unfold just a few years later, with the burgeoning AI bubble as the primary culprit. While there's no positive note to conclude on, perhaps this is an opportune moment to revisit unfinished projects rather than investing in new hardware. The duration of this trend remains uncertain, but many, including myself, have already postponed planned projects for 2026 due to these market conditions.