Anders Bylund, The Motley Fool
Thu, February 26, 2026 astatine 10:57 AM CST 5 min read
Last summer, the workstation I usage for penning these articles felt sluggish. You cognize however it goes, right? I'm utilizing the aforesaid web browsers and connection processor arsenic always, but each leafage comes with a twelve postulation trackers and a clump of ads that weren't determination 5 years ago, erstwhile I past upgraded my machine memory. It was clip for a mates of caller representation sticks successful my Frankenbox with parts acquired and upgraded successful waves betwixt 2010 and 2022.
Will AI make the world's archetypal trillionaire? Our squad conscionable released a study connected the 1 little-known company, called an "Indispensable Monopoly" providing the captious exertion Nvidia and Intel some need. Continue »
So I ordered 64 gigabytes of DDR4 representation successful October 2025, tripling the representation capacity. It outgo $95 with escaped shipping from Amazon (NASDAQ: AMZN), and I could person gotten a akin woody from Newegg (NASDAQ: NEGG) oregon Best Buy (NYSE: BBY). It was the close medicine, and this aged Linux container has been creaseless arsenic silk successful the past 4 months.
Out of axenic curiosity, I checked what the aforesaid representation upgrade would outgo maine today. The Amazon listing jumped to $450. People are buying utilized ones connected eBay (NASDAQ: EBAY) for $250 oregon more. I could marque a important nett by reselling this 2023 product, designed for an outdated procreation of processors and motherboards.
As a longtime machine enthusiast with a college-aged gamer successful the family, this makes maine cringe. But it definite feels bully to ain Micron Technology (NASDAQ: MU) banal successful this epoch of upside-down terms trends for machine memory.
You see, the soaring prices of consumer-grade representation sticks are a broadside effect of a broader marketplace trend. I'm definite you cognize the drill already, but it goes similar this:
-
Artificial quality (AI) titans similar OpenAI, Microsoft (NASDAQ: MSFT), and Alphabet (NASDAQ: GOOG) (NASDAQ: GOOGL) are gathering monolithic information centers to bid and run their AI platforms.
-
These systems necessitate a ton of memory. One Nvidia (NASDAQ: NVDA) GB300 NVL72 rack ships with 17 terabytes of DDR5 and 20 terabytes of adjacent faster precocious bandwidth representation (HBM). My 64 GB of slower representation chips whitethorn beryllium plentifulness for a penning workstation oregon adjacent a gaming PC, but it's little than 0.2% of that azygous AI computing system's representation needs.
-
One GB300 rack has an tremendous magnitude of computing power, but the existent giants don't halt there. For instance, Microsoft added a clump of 64 of these systems to the Windows Azure cloud-computing service. My brace of representation sticks isn't adjacent a rounding mistake successful this immense representation pool.
-
Nvidia and others are selling these AI accelerators arsenic accelerated arsenic they tin marque them. As a result, there's astir bottomless request for high-end representation chips for the AI industry. The Big 3 representation makers -- Micron, Samsung (OTC: SSNLF), and SK Hynix -- are besides selling chips earlier they're adjacent made. Micron has already sold retired its expected 2026 accumulation volume, with immoderate contracts locking down supplies successful 2027 and 2028.

3 weeks ago
17



.png)

English (CA) ·
English (US) ·
Spanish (MX) ·