Memory chip prices have quadrupled in three quarters — and on June 25, Apple became the most prominent consumer electronics company to stop absorbing the cost and start passing it to customers.
The Reading
What Happened: From Earnings Warning to Price Tag
Apple CEO Tim Cook sat down with the Wall Street Journal in mid-June and used language that was unusual for a carefully managed corporate interview. “This is a hundred-year flood,” he said. “I’ve never seen anything like it in any area in over 40 years.” He had already flagged the direction in April, telling investors on Apple’s quarterly earnings call that “significantly higher memory costs” were coming in the June quarter. By June 25, the warning had become a price list.
CNBC reported that Apple announced increases of hundreds of dollars on Macs and iPads that morning. Apple’s online store briefly went offline as the updated prices went live. Apple’s stock dropped more than 6% — its worst single-day fall since April 2025. Cook’s language in the Journal interview had already explained why. “Unfortunately, price increases are unavoidable,” he said. “We’re doing our best to mitigate the huge increases that are being passed to us, and we’ve been trying to shield our customers from the increases, but the situation has become unsustainable.”
Apple’s official statement was equally candid: “The consumer electronics industry is facing an unprecedented challenge. The rapid expansion of AI data centers has created an extraordinary surge in demand for memory and storage. We have never seen a component price increase this much, this quickly.”
Who Says So: The Musk Alignment That Matters
When Elon Musk read Cook’s quote and posted on X — “Biggest price jump in anything I’ve ever seen too” — it was a confirmation that cut across competitive lines. Cook and Musk are not often in agreement. On the severity of this shortage, they are completely aligned.
The alignment is more telling than it appears on the surface. SpaceX’s Colossus data center in Memphis is among the largest AI training clusters on the planet. Musk’s xAI and SpaceX operations are precisely the kind of large-scale AI infrastructure build-out that has been consuming high-bandwidth memory (HBM) and storage chips at a rate the industry’s supply chain was not designed to handle. Cook is paying the cost of a shortage that Musk’s own operations are structurally contributing to. That both men — one on the demand side of the crisis, one helping drive it — describe the situation in the same historic terms is the signal serious readers should not dismiss.
The Cook-Musk alignment is not merely a curiosity; it is a structural admission that the AI infrastructure build-out has now created a two-tier components market where hyperscale operators and consumer device makers are in direct, zero-sum competition for the same constrained chip supply. When the operator of one of the world’s largest AI training clusters publicly validates the severity of the shortage that his own infrastructure is helping cause, it forecloses the standard industry talking point that supply will “catch up soon” — because the entity with the most incentive to say supply is fine is instead saying it is historically bad.
Why It Matters: A Third Wave of Inflation, This Time in Silicon
Memory and storage chip prices have quadrupled over the past three quarters, according to data from Counterpoint Research cited by CNBC. The mechanism is structural rather than cyclical. Hyperscalers — Google, Microsoft, Meta, and Amazon — require enormous quantities of high-bandwidth memory to run AI inference and training at the scale their models now demand. Memory chip suppliers, recognizing that hyperscaler contracts are higher-margin and arrive in larger, more predictable batches, have re-oriented their production accordingly. Consumer electronics manufacturers — Apple, Dell, HP, Nintendo — receive what remains. Right now, that is not much.
Micron’s most recent quarterly results illustrate the distortion this creates. The company’s revenue quadrupled year-over-year, and its gross margin jumped from 39% to 84.9%, surpassing both Nvidia and Meta. For memory suppliers, the current environment is a windfall. For every company purchasing memory to put into a device a human being will hold in their hands, it is the inverse. As discussed in our earlier analysis of why AI token costs are exploding across the stack, the cost pressures now visible in consumer hardware trace directly to the same infrastructure arms race reshaping cloud pricing.
The Wall Street Journal published data showing computer software and accessories prices rising approximately 15% year-over-year — a rate last seen in the 1980s. The Journal framed the AI data center build-out as “a third wave of inflation,” following the energy and food inflation waves of 2022 and 2023. That framing carries analytical weight: it positions the memory crunch not as a temporary supply hiccup but as a macroeconomic force with a distinct cause, distinct beneficiaries, and a timeline tied to the pace of AI infrastructure investment rather than to traditional semiconductor production cycles.
Apple is the most visible company absorbing the shock, but it is far from alone. HP, Dell, and Nintendo have all raised prices on their products. Best Buy’s incoming CEO Jason Bonfig told reporters that the company expects its computing division to be the segment most affected by price increases in the coming quarters. The ripple is industry-wide. The dynamics driving this shortage connect directly to the broader questions about what AI infrastructure actually costs the planet — the memory crunch is the consumer-facing price signal of the same resource competition playing out in electricity grids and water supplies.
Gartner’s forecast puts the demand destruction in concrete terms. CNBC reported that Gartner expects soaring memory costs to reduce global PC shipments by 10.4% and smartphone shipments by 8.4% in 2026. Gartner analyst Ranjit Atwal addressed Apple specifically: “Even Apple can’t be safe — as much as they have all the expertise and long-term planning, and everything else. This is beyond their capacity to limit the impact.” Counterpoint Research estimates that higher memory costs could add roughly $200 per iPhone. If that cost reaches consumer prices before the holiday cycle, the device that hundreds of millions of people treat as a default purchase becomes a meaningfully harder financial decision.
What to Watch: The iPhone Question Cook Didn’t Answer
Cook did not say when iPhone price increases would arrive. He did not need to. The Mac and iPad announcements on June 25 established the pattern. The consumer electronics industry’s assumption that Apple would absorb component cost increases rather than pass them on — an assumption built over years of Apple’s gross margin management — has now been explicitly retired by the company’s own CEO and its public-facing communications. The question for observers is not whether iPhone prices will rise, but how much, and whether the timing lands before or after the holiday quarter.
There is a secondary question that analysts are beginning to frame: whether the memory shortage accelerates a longer-term shift in how AI capabilities are delivered to consumers. If device-side memory becomes prohibitively expensive, the economic argument for running AI inference in the cloud rather than on-device strengthens. That would represent a structural tailwind for cloud providers — the same hyperscalers whose data center demand is causing the shortage in the first place. This dynamic is already visible in the way cloud AI token pricing is evolving as providers compete for workloads that might otherwise run locally.
The competitive pressure on device makers also raises questions about market concentration. If Micron’s margins are now approaching 85%, and that margin comes directly from the price pain being absorbed by Apple, Dell, HP, and consumers, regulators in the US and EU may find the memory market a more politically legible target than the AI model market itself. The shortage is already attracting the kind of bipartisan attention that tends to precede congressional inquiry.
How the Memory Shortage Compares to Previous Chip Crises
The current 2025–2026 HBM and DRAM shortage is different from earlier chip crises because it is being driven by structural AI data center demand rather than a temporary supply shock or manufacturing transition. The 2021–2022 semiconductor shortage was largely caused by COVID-era supply chain disruption and a sudden automotive demand surge, while the 2016–2018 NAND flash shortage was linked to the industry’s transition to 3D NAND manufacturing and related yield challenges. Both earlier shortages eventually eased as capacity expanded, manufacturing improved, and demand cooled. The current memory shortage is harder to resolve because hyperscalers building AI infrastructure are competing directly with consumer electronics companies for advanced memory supply, but they can offer higher prices and larger long-term volume commitments. This gives suppliers a strong incentive to prioritize AI customers over PC, smartphone, console, and consumer hardware makers. As a result, even new supply may not quickly normalize prices across the broader market, because AI data center demand continues to absorb incremental capacity at scale.
How Serious Players Should Respond
For device makers, the strategic window for pretending this is a temporary disruption has closed. Apple’s public acknowledgment that the situation is “unsustainable” and its decision to pass costs to consumers sets a precedent that competitors will follow — and that procurement teams, product roadmap planners, and CFOs at every hardware company now need to price into their models. Long-term memory supply agreements, investment in alternative suppliers, and accelerated research into memory-efficient chip architectures are no longer hedges; they are baseline requirements for companies that want to remain competitive in a market where the memory supply chain has been structurally re-prioritized by a class of buyers with deeper pockets.
For regulators and policymakers, the Gartner forecast — a projected 10.4% contraction in global PC shipments and 8.4% in smartphones — deserves serious attention as a consumer welfare issue, not merely an industry pricing story. The memory market’s structure, in which a handful of suppliers are now capturing windfall margins from a shortage they did not deliberately engineer but are rationally exploiting, is exactly the kind of market dynamic that antitrust frameworks were designed to examine. The question is whether regulators move before the holiday cycle makes the political pressure undeniable, or after.
For AI infrastructure investors and hyperscalers, the Cook-Musk exchange should prompt a harder look at the externalities of the data center build-out. The AI capital expenditure cycle has created genuine economic harm for the consumer electronics sector and — via device price increases — for hundreds of millions of consumers globally. That harm is now documented, attributed, and politically visible. Companies that frame themselves as responsible AI developers while simultaneously driving a shortage that is making smartphones and laptops unaffordable for a meaningful share of the global population will find that framing increasingly difficult to sustain.
