Apple's memory-shortage price hikes, tested against the arithmetic: mostly real cost, with the one markup it cannot explain landing on the cheapest iPad

On Thursday 25 June Apple raised prices across its Macs, iPads, HomePodHomePodApple's smart speaker. and Vision ProVision ProApple's mixed-reality headset, a high-priced wearable computer you put on your head., blaming a shortage of memory and storage chips. It left the iPhone, Apple Watch and AirPods untouched. Apple said the rapid build-out of AI data centers had created an extraordinary surge in demand for memory and storage, and that it had never seen a componentcomponentOne of the individual parts, such as a memory chip or a screen, that make up a finished device. price rise this much, this quickly. Apple shares fell about 6% on the day.

What the stock market does and what the underlying economics say are two separate questions, and we will not read a motive into the share price. A price increase that simply passes on real costs protects a company's profit marginprofit marginThe share of a product's price that is left over as profit after the costs of making and selling it., so a falling stock is at least as consistent with investors fearing that higher prices will scare off buyers, or doubting that Apple can pass the full cost on, as with a belief that Apple is padding its profits. We make no claim about why the stock moved. We test only the economic question, which Apple's own published price list lets us answer directly.

Across the lineup, price and memory are tangled together: Macs cost more and also carry more memory, so a hike that rose with either one would look the same. Exactly one pair of devices breaks that tie. The MacBook Air and the Vision Pro ship with the identical memory setup: 16GB of working memory and 256GB of storage. The Air started at $1,099; the Vision Pro at $3,499, more than three times the price.

Both received the same $200 increase.

Same memory, same dollar hike, three times the price. A flat percentage markup cannot produce that. Passing on a fixed cost per gigabyte of memory is the simplest explanation that can.

A percentage markup would have lifted the Vision Pro by about $630 to match the Air's 18%. Instead it rose 5.7%. That is a strong single example, not a proof. With only eight repriced devices we cannot run a proper statistical test, because everywhere else memory and price rise together. So we lean on this one pair and on the cost arithmetic below, and we flag that as a limitation rather than dress a single matched pair up as a fitted statistical relationship.

To turn the word 'shortage' into dollars, we need to know how much more memory costs now than when these products were last priced, in early-to-mid 2025. We use a range rather than a single number: roughly $8 to $14 for each extra gigabyte of working memory (DRAM), plus about $0.04 for each gigabyte of storage (NAND flash). The low end of the working-memory range comes from Counterpoint's price ladder for 64GB server memory sticks ($255 in Q3 2025, $450 in Q4, about $900 in Q1 2026). The high end comes from contract prices for the mobile-style memory Apple actually uses (LPDDR5LPDDR5A low-power type of working memory used in phones, tablets and Apple's laptops; the 'LP' stands for low power.). The storage figure comes from the price of a typical 1TB consumer SSD roughly doubling, from about $45 to $90.

Be clear about how weak this test is. A range of $8 to $14 per gigabyte spans 75% from bottom to top. So a hike 'fitting inside the range' is an easy bar to clear, and it should be read as 'not contradicted by memory costs,' not as 'proven to be nothing but memory costs.'

We leave the 14-inch MacBook Pro out of the tally. Its base model became 32GB of memory and 1TB of storage in March 2026, when Apple dropped the cheaper 512GB version, so part of its $300 increase is simply buying more memory, not paying a shortage premium on the same memory. We show it only for context; blaming a spec upgrade on the shortage would muddy the comparison rather than reveal anything.

On the devices we can compare like for like, the dollar hikes do grow with gigabytes: devices with 8GB of memory rose about $100, devices with 16GB rose about $200. That is the signature of a real supply story. But the honest counter-argument is visible in the same chart, and we did not hide it: most hikes sit near the TOP of their estimated memory cost, not the middle. The Air's $200 against a $138-234 range, the Vision Pro's $200 against the same, the mini's $100 against $69-117. Consistently pricing memory at the expensive end of a wide range is itself a mild markup spread across the whole lineup. The data is consistent with 'genuine cost, rounded up a bit,' but it does not rule that reading out, and a skeptic is entitled to read the clustering at the top of the range as a soft skim laid on top of a real shock.

The memory-based explanation survives the test we can give it. That is not a clean acquittal, because the test we can give it is a generous one.

Two different measures are easy to confuse, so let us keep them apart. 'Share of price' (the bar chart) measures who pays the most in proportional terms. 'Above the cost range' (the range chart) measures who paid more than memory can account for. They are not the same, and the difference matters: the iPad Air's 25% is a steeper proportional jump than the iPad Pro's 20%, yet the iPad Air still fits inside its range while the iPad Pro sits just above it. A high percentage is not, by itself, evidence of a markup.

Only one device is both the steepest proportional hike and above its cost range: the $349 entry iPad. It rose $100 against an estimated $53-$89 of extra memory cost, a 28.7% increase, on the device with the least memory to blame. The reason is plain in the numbers: Apple rounded every increase to a tidy $100 or $150, and rounding up bites hardest where the underlying cost is smallest.

We narrow the claim on purpose. The iPad Pro also rises above its range, but only by about $22 on a $999 device, well within rounding and not a case of 'cheap and low on memory.' So the careful, defensible statement is not that the markup falls on cheap devices in general; it is that exactly one device, the entry iPad, shows an increase the shortage cannot fully explain, and on that device the effect is regressiveregressiveDescribing a cost or charge that falls hardest, in proportional terms, on those paying the least., meaning it falls hardest on the buyer paying the least. The dollars involved are small, on the order of $30-$50 across the whole lineup once the within-range hikes are netted out; the fairness point is real even though the amount is not large.

The cynical read (a margin grab across the board) and the corporate read (pure shortage, everywhere) are both too strong. The unexplained extra is small, and it is concentrated on the cheapest iPad.

What did not change is also evidence. A tarifftariffA tax a government places on imported goods, which can raise the prices buyers pay., currency or general profit-margin story would predict price rises spread broadly across the catalogue. A memory story predicts the hit lands on the devices with the most memory and spares those with the least. Apple left the iPhone, Watch and AirPods unchanged, which fits the memory story better than the alternatives, though it is consistent with that story rather than proof of it.

The cost shock shows up outside Apple too. Microsoft the same week raised Xbox and Surface prices, citing higher storage and memory costs, and PC makers say memory is now a far larger share of what it costs to build a PC than a year ago. When an entire industry moves together on the same component, that points to a shared input shock rather than one company's grab. The 2017-18 episode, when PC makers largely absorbed a memory spike instead of passing it on, is the natural comparison, but it differs in two ways: memory is reportedly a much bigger share of the build cost now (per a single HP estimate, not a measured figure), and demand conditions are different, so 'they absorbed it last time' is not proof they could now.

Each reading of the evidence makes a near-term, checkable prediction.

• If memory contract prices fall back in the second half of 2026 and Apple quietly reverses the hikes, that supports the pure cost-pass-through reading. Trigger: a sustained drop in DDR5/LPDDR5 contract prices followed by a price cut.
• If the autumn iPhone 18 launches with its own memory-driven price rise, that strengthens the broad cost-shock story, because the iPhone carries a lot of memory and would be the next device to absorb a real shortage. Trigger: an iPhone 18 priced above the iPhone 17 line and attributed to components.
• If the entry iPad stays elevated while everything else tracks falling memory costs, that confirms the small regressive markup is built in, not a temporary shortage premium. Trigger: the entry iPad holds at $449 after memory prices ease.
• If competitors do NOT follow suit on memory-heavy devices, the 'shared industry shock' reading weakens and a company-specific element becomes more likely. Trigger: a major PC maker holds prices flat on comparable configurations.