The 2026 Mac Squeeze: How Apple's Lineup Has Changed

Apple has quietly raised prices, eliminated configurations and launched its first iPhone-chip Mac — all in response to a once-in-a-generation memory shortage. Here's what's happened, why, and what it means if you're thinking of buying a Mac in 2026.

Section 01What caused all this?

Memory and storage are the two most volatile components in any computing device. They are also, for a brief and miraculous period running from roughly mid-2023 to mid-2025, the two components whose prices were trending down. Apple — like every other PC and smartphone manufacturer — was enjoying margin tailwinds it had not seen in years.

Then artificial intelligence happened. Or more precisely, the infrastructure build-out for artificial intelligence happened. Hyperscale data centres needed HBM — High Bandwidth Memory — a specialised form of DRAM in which multiple memory dies are stacked vertically and connected directly to a processor through thousands of microscopic wires. The stacking is what delivers the bandwidth that AI accelerators need, and it is also what makes HBM expensive: it consumes roughly four times the wafer capacity of conventional memory to produce the same total volume, and it sells at several times the price.

Samsung, SK hynix and Micron — who together control essentially the entire memory market — made a rational choice. They shifted production to HBM, where the margins were extraordinary, and let conventional DRAM and NAND flash storage production stagnate. The result, by late 2025, was a supply crunch in the regular memory used in every PC, Mac, phone and console on the planet.

For Apple, this created an immediate problem. Apple Silicon is built around unified memory — the GPU, CPU and Neural Engine all sit on a single piece of silicon called an SoC (system on chip) and share the same physical memory pool. This design is what gives Apple Silicon its efficiency advantage over traditional PC architectures where memory sits inches away from the processor in a separate DIMM slot. The catch is that Apple's unified memory has to be packaged tightly against the SoC using premium LPDDR5X chips, in volumes that scale linearly with how much RAM each Mac ships with. There is no way for Apple to engineer around higher memory prices through clever firmware. Every additional gigabyte of memory in every Mac is a real, fungible cost line item.

And that cost has multiplied. Here's how Apple has responded across each product line.

Section 02The MacBook Air: Apple ate the cost

The MacBook Air is Apple's biggest-volume laptop, and it's a useful canary for the rest of the lineup. When component costs move, what Apple does to the Air tells you what it's willing to do publicly. The two-year story of the base-spec MacBook Air looks, on the surface, like a series of consumer-friendly price cuts. Underneath, it's a story of vanishing gross margin.

Two years ago, you could buy a 16GB / 256GB MacBook Air M3 for £1,299 (the 16GB tier was a paid upgrade from the 8GB base). By March 2025, the M4 launched with 16GB as the standard base for £999. By March 2026, Apple retired the 256GB tier entirely — the base MacBook Air now ships with 16GB and 512GB at £1,099.

The headline price has barely moved. The estimated cost to make that machine, however, has more than doubled. Memory and storage have gone from roughly £37 of BOM (bill of materials — the total cost of physical components Apple buys to make one unit) to around £200. Everything else has crept up modestly with inflation. The end result: Apple's hardware gross margin on this SKU has compressed by an estimated 37 percentage points in two years.

Section 03The MacBook Pro: The storage tier trick

The MacBook Pro tells a more clever story. When Apple launched the M5 Pro and M5 Max MacBook Pros in March 2026, the headline retail prices looked almost identical to the M4 generation. The 14-inch M5 Pro is £2,199 (was £1,999). The 14-inch M5 Max is £3,599 (was £3,199).

That looks like a £200 and £400 price increase, which is unusual but not extraordinary. What Apple actually did, however, was eliminate the entry storage tier on these machines. The M4 Pro 14-inch base shipped with 512GB; the M5 Pro 14-inch base ships with 1TB. The M4 Max shipped with 1TB; the M5 Max ships with 2TB. The "price increases" are, almost to the pound, what Apple was previously charging as an optional storage upgrade.

This is the cleanest example of forced ASP recovery anywhere in the Mac lineup. ASP — average selling price — is the figure that matters to Apple's revenue: the actual amount of money Apple collects per unit sold across the full mix of configurations, as opposed to the list price advertised on the storefront. By eliminating the cheaper storage tier, Apple has not raised the headline price of either machine. It has simply removed the option that previously dragged the ASP down. Every M5 Pro and M5 Max buyer is now paying for an upgrade they previously had the option to decline, and Apple's ASP on these SKUs has risen by exactly the cost of the eliminated tier.

From an internal margin standpoint, it's brilliantly designed. Apple gets to report the same headline pricing while collecting an extra £200-£400 per unit sold, all of which helps cover the NAND cost surge. The customer experience is "the price didn't change," even though the price they pay actually has.

Section 04The Mac mini and Mac Studio: Where Apple stopped pretending

The desktops are where Apple's strategy starts to visibly break. Apple has not refreshed either product to the M5 generation. The Mac mini still ships with M4 silicon from October 2024, and the Mac Studio still ships with M4 Max and M3 Ultra silicon from March 2025. There has been no public statement explaining the delay.

The Mac mini's quiet £200 price increase

In October 2024 the base Mac mini sold for £599 with 16GB of memory and 256GB of storage. By May 2026, the £599 configuration has been retired. The new entry point is £799 for 16GB and 512GB. That's a 33% price increase on the cheapest Mac in the lineup, and Apple didn't announce it. The 256GB option simply stopped being available.

The Mac mini Pro has also lost build-to-order options. The 32GB and 64GB memory upgrades have been pulled. You can now configure a Mac mini Pro with 24GB or 48GB — and that is it. If you want 64GB on an Apple desktop priced under £2,000, you can't have it any more.

The Mac Studio rationing

Then there's the Mac Studio, which is the most extraordinary case study in the entire lineup. Apple has done something it almost never does: publicly withdrawn product configurations rather than reprice them.

Configuration withdrawals are unusual behaviour from Apple. The standard playbook when components are constrained is to extend delivery estimates while keeping the option visible. Pulling configurations entirely — without notice or explanation, while the tech specs page continues to advertise them — is the behaviour of a company that doesn't expect supply to recover quickly enough to keep the option in front of buyers.

What it tells us about Apple's internal economics is that the 256GB and 512GB Mac Studio configurations had become unprofitable. By Q2 2026, a 512GB Mac Studio carries an estimated £3,250 of memory cost alone — against a previous retail price of around £6,200 ex-VAT. Add the SoC, chassis, storage, cooling, assembly, and Apple's margin on the SKU had collapsed below their internal threshold. Rather than raise the price to £8,000+ and create a PR problem, they pulled it.

Section 05The MacBook Neo: Apple's strategic retreat

In the middle of all this, in March 2026, Apple announced an entirely new MacBook. It's called the MacBook Neo. It costs £599 (£499 for students). And it's built around a chip Apple has never used in a Mac before.

The MacBook Neo runs on the Apple A18 Pro — the same chip in the iPhone 16 Pro, with one fewer GPU core. In every meaningful sense, it's an iPhone chip in a laptop chassis. We covered the spec compromises in detail in our MacBook Neo buying guide, but the headline points are: 8GB of non-upgradeable memory, 60GB/s memory bandwidth (half the M4's), no Thunderbolt, no MagSafe, no backlit keyboard, basic 1080p webcam, no notch, no P3 wide colour display.

What is binning, and why does it matter here?

The detail of "one fewer GPU core" on the A18 Pro is more important than it sounds. It points to a manufacturing trick called binning, which is central to understanding the Neo's economics.

When TSMC fabricates a wafer of A18 Pro chips for Apple, a meaningful percentage come off the production line with minor defects — typically a single non-functioning GPU core, a failed cache block, or an inability to hit the full clock speed. These chips aren't broken; they just don't meet the full specification. Rather than scrap them, chip designers build in the ability to permanently disable the affected component and sell the chip as a lower-tier product. That's binning.

The iPhone 16 Pro gets the A18 Pro chips with all six GPU cores working. The MacBook Neo gets the chips with five working cores out of six. Apple has been doing exactly this with M-series for years — every base M3 with one disabled CPU core, every "11-core" M3 Pro that is really a 12-core with one disabled, is a binned chip that would otherwise have been scrap.

What makes binning so valuable to Apple in the Neo specifically is that the A18 Pro's enormous iPhone production volume creates an enormous supply of binned chips as a byproduct. The R&D, the mask sets, the wafer cost — all of it has already been paid for by iPhone economics. Apple's marginal cost for a binned A18 Pro destined for a Mac is close to zero. Compare that to an M-series Mac SoC, which has to be designed and tape-out funded specifically for Mac volumes of perhaps 10 million units a year.

The Neo's chip strategy isn't just "use a cheap chip." It's "use chips that would otherwise have been thrown away, from a production line whose costs are fully amortised against another product."

Why this works for Apple right now

Take the binning trick, combine it with the Neo's other cost decisions — locked 8GB memory in a single configuration, lower memory bandwidth that doesn't need premium packaging, no Thunderbolt, no MagSafe, basic display — and you can see what Apple has built. The Neo is a strategic retreat from the unified-memory premium that has been Apple's differentiator for five years. It doesn't try to compete with M-series Macs. It competes with Chromebooks and entry-level Windows laptops.

The catch is that even the Neo isn't immune to the memory crisis. Our model estimates that in the single quarter since launch, the Neo's hardware gross margin has compressed from approximately 49% to 40% — a 9 percentage-point fall in three months, driven entirely by ongoing memory and NAND price increases. Apple's escape hatch is already leaking.

Section 06What this means if you're buying a Mac in 2026

Step back from the individual products and a clear pattern emerges. Apple is running five different strategies at once to manage the same underlying problem.

The bigger picture

The component cost surge is not over. TrendForce's published forecasts have DRAM prices continuing to rise through at least Q3 2027. SK hynix's own internal analysis projects PC DRAM supply trailing demand until late 2028. Apple's CEO has publicly acknowledged that rising memory costs will begin compressing Apple's profit margins later in 2026.

Translated to consumer terms: this is likely to get worse before it gets better. If you've been considering a Mac purchase, the choice of whether to buy new or refurbished now has a different shape than it did even a year ago. Apple's headline prices have been artificially held down by absorbing margin. The configurations you actually want — reasonable memory, reasonable storage, sensible pricing — are increasingly the configurations Apple is steering you away from.