Amazon did not send a press release. The price hikes took effect around August 22, 2026, without any prior official announcement. Customers simply noticed, as they tend to do, when the bill got larger. Amazon raised prices across its hardware product line overnight, including the base Echo Dot from $49.99 to $79.99, a 60% increase for what had been the company’s entry-level device. The 16GB Kindle went from $109.99 to $149.99, and the Fire TV Stick 4K Max moved from $59.99 to $84.99. Amazon confirmed the logic in a statement to the press: the consumer electronics industry is “facing significant increases in memory and storage component costs” and, after absorbing them as long as it could, adjusted pricing across its product lines.
The more revealing detail sits in the timing. Roughly four weeks before the price hikes, Amazon CEO Andy Jassy told investors the company’s 2026 capital spending would reach $220 billion, up from a previous estimate of $200 billion, with higher memory costs driving the revision. The same scarce input that is inflating the price of a smart speaker is also inflating the cost of every AI data center Amazon is racing to build. This is not a supply chain footnote. It is the defining capital allocation story of this technology cycle.
The Anatomy of the Shortage
Just three companies, Samsung, SK hynix, and Micron, control roughly 95% of global memory chip supply, and all three have been shifting wafers away from consumer memory and toward high-bandwidth memory as fast as they can. The economics are straightforward. TrendForce has estimated that roughly one gigabyte of HBM consumes about four gigabytes’ worth of standard DRAM wafer capacity, and that AI-related demand will absorb close to 20% of the entire industry’s DRAM wafer capacity in 2026. Every AI accelerator shipped is a direct tax on the consumer aisle.
The magnitude is not subtle. TrendForce upgraded its first-quarter 2026 conventional DRAM contract price forecast to plus 90–95% quarter-over-quarter, with PC DRAM expected to at least double. For the third quarter, TrendForce now predicts conventional DRAM contract prices will rise another 13–18% quarter-over-quarter. The rate of increase is moderating only because so much of the damage has already landed on buyers’ bills of materials.
The demand destruction is real and measurable. IDC’s latest tracker forecasts worldwide smartphone shipments falling 13.9% in 2026 to 1.09 billion units, which it describes as a record annual decline. Gartner projects worldwide PC shipments declining 10.4%, with a 130% surge in combined DRAM and SSD prices lifting PC prices by 17% and smartphone prices by 13% against 2025 levels.
Where the Pricing Power Sits
For investors, consumer pain is producer opportunity, provided the producer commands the right position in the stack. Alongside SK hynix and Samsung, Micron is benefiting from a supply crunch that has pushed memory prices sharply higher, with the three companies controlling about 90% of the DRAM market and effectively all near-term HBM supply, giving them significant pricing power as shortages are expected to persist until around 2028.
The financial evidence is striking. Samsung’s first-quarter operating profit increased about 755% year-on-year, and SK hynix’s revenue exceeded KRW 50 trillion for the first time with an operating profit margin of 72%. Micron has reported a sharp year-on-year surge in profit as memory prices reset higher, but the specific 770.8% figure is not a reliable baseline for comparison across the company’s fiscal periods and is better understood simply as a cycle-level jump.
The question for a disciplined long-term investor is not whether prices are high right now. They obviously are. The question is which of the three producers sustains pricing power through the next capacity cycle, once new fabs begin yielding volume. Significant new manufacturing capacity is generally not expected to arrive until late 2027 or 2028, creating a prolonged supply gap. SK hynix holds the strongest HBM franchise today: Samsung reclaimed the top spot in conventional DRAM revenue in the final quarter of 2025, though SK hynix continued to dominate HBM with a 57% market share.
The Honest Counterargument
None of this is without risk. Long-term supply agreements between hyperscalers and memory makers are already capping upside. Pricing is climbing more slowly than anticipated, Micron has covered roughly 20% of its DRAM volume and about one-third of its NAND volume with long-term agreements, and those contracts can limit how much upside it captures if spot markets run hotter than contract floors and caps. A demand collapse in consumer devices, if severe enough, could eventually free conventional DRAM capacity faster than the market expects.
Charlie Munger’s first question would be whether the moat is structural or merely situational. Three producers, no new entrants, a decade-long wafer investment cycle, and an AI buildout that shows no credible sign of reversal: that combination looks more structural than most. Amazon raising the price of a speaker is not the story. It is the symptom. The story is which memory producer locks in the economics of scarcity for the next five years, and whether the current valuation offers a reasonable entry for a patient owner.
