Why Apple Lead Times Are Stretching — and How UK Businesses Are Coping

What has actually happened to Apple lead times

The pattern through 2026 has been consistent: standard configurations stay broadly available, while anything with more memory in it goes to the back of a long queue.

In April, TrendForce reported Mac mini and Mac Studio delivery estimates stretching dramatically by configuration. A Mac mini with the M4 Pro chip and 64GB of memory was quoting 16–18 weeks in the United States, while the base 16GB model was showing delays of around a month. High-memory Mac mini and Mac Studio configurations in China were quoted at four to five months. Over the same period the cost to upgrade a top-tier M3 Ultra Mac Studio from 96GB to 256GB rose to $2,000, up from $1,600.

By August the squeeze had reached Apple’s highest-volume laptop. TechCrunch reported the global memory shortage hitting the MacBook Air directly, with most configurations quoting two to three weeks and a 13-inch Air specified with 32GB of RAM not shipping for over a month. Coverage at the time noted Apple steering shoppers towards the MacBook Pro, and a back-to-school promotion that slipped from its usual June start.

Writing in TidBITS on 10 August, Adam Engst added a caveat worth carrying into any procurement plan: some consultants were seeing real delivery slip beyond Apple’s own published estimate after an order had been accepted. In other words, the quoted date is a forecast, not a commitment.

The cause: AI data centres bought the memory first

This is not an Apple manufacturing problem. It is an allocation problem across the entire memory industry.

Samsung, SK hynix and Micron have been shifting production capacity towards high-bandwidth memory (HBM) and enterprise server DRAM, because that is where AI infrastructure demand — and margin — sits. Cloud providers have locked in multi-quarter purchase agreements to secure supply. What is left over is what the PC, smartphone and embedded markets get to buy.

TrendForce’s contract-price data shows how violent the repricing has been. Its first-quarter 2026 outlook was revised sharply upward mid-quarter, with conventional DRAM contract prices rising 90–95% quarter on quarter against an initial estimate of 55–60%, and NAND flash up 55–60% against an initial 33–38%. The second quarter brought a further 58–63% on DRAM and 70–75% on NAND.

The one piece of good news in that dataset is the trend line. TrendForce put third-quarter increases at 13–18% for DRAM and 10–15% for NAND — still rising, but at a fraction of the earlier rate as consumer demand weakened and the comparison base got high. Prices are not falling. The rate of increase is slowing.

Apple has put this in writing to the SEC

Most supply-chain commentary is analyst inference. This is not. In its Form 10-Q for the quarter ended 27 June 2026, Apple states plainly:

“The Company is experiencing a period of supply constraints and increasing costs for components driven by factors such as industry supply-demand imbalances for components, including advanced semiconductors, storage (NAND) and memory (DRAM).”

And on the outlook: “The Company expects these trends to intensify, which may materially negatively impact the Company’s revenue, costs, gross margin, results of operations and financial condition.”

It is worth being precise about what this did and did not do to Apple’s numbers, because the story has been overstated elsewhere. Apple’s products gross margin actually rose year on year, to 40.1% for the third quarter of 2026 from 34.5% a year earlier. The filing attributes that increase to product mix and tariff refunds — “partially offset by higher costs, including memory”. Memory is a drag on the margin, not yet a collapse of it. The forward-looking sentence is the one that matters for buyers: Apple itself expects the pressure to get worse before it gets better.

Apple has also been looking for supply outside its usual roster. Reuters reported in August that the company has been testing memory from Chinese manufacturer CXMT across product lines including iPhones and MacBooks, specifically to mitigate the component shortage.

What it means for iPhone — and why timing matters this week

The iPhone side of this is less about delivery dates so far and more about cost, which tends to become a delivery problem later.

TechInsights’ estimate, reported in August, is that the 12GB DRAM package inside an iPhone 18 Pro costs Apple around $145 per unit, against roughly $39 for the equivalent memory in an iPhone 17 Pro — a 272% increase for the same quantity of the same class of chip. Memory has reportedly reached a record share of the handset’s build cost. Forbes reported the same month that iPhone 18 production was running into a mounting DRAM shortage, and MacRumors flagged the risk that iPhone 18 Pro and the foldable model could sell out quickly for the same reason.

Apple holds its product event on Wednesday 9 September 2026. Nothing is announced yet, and analyst price expectations circulating before a keynote are exactly that — expectations. But the direction of travel on component cost is documented, and constrained launch supply is a reasonable planning assumption rather than a pessimistic one. Our companion piece on what UK business buyers need to know before the event covers the fleet-refresh question in detail.

A caution on reading Apple’s own store

Checking Apple UK on 8 September 2026, the headline pages look calm. MacBook Air shows standard free shipping with no delay banner. MacBook Pro starts at £1,999 for the 14-inch and £2,999 for the 16-inch. Mac mini shows “available from 22.09” — but that is a pre-order date for the newly announced M6 and M5 Pro models, not a shortage queue, and it would be easy to misread as one.

That gap between the headline page and the configured order is the whole point. Entry configurations are the ones Apple protects, because they are the ones people see. The lead time appears when you specify the machine your business actually needs — more memory, more storage — and it appears at the bag, not on the marketing page. Any procurement process that checks availability by glancing at a product page is measuring the wrong thing.

It is not only Apple — the same queue holds your servers

Focusing on Apple understates the problem. Every vendor buying DRAM and NAND is in the same allocation queue: server, storage, networking and Windows endpoints alike. Custom builds with specific memory configurations are reported to be the worst affected across the board, which is precisely the profile of most business server orders.

We have tracked this from the infrastructure side all year. Our analysis of whether to buy servers now or wait out the 2026 memory spike and the ongoing enterprise server price index cover the same underlying cause with server-specific numbers. If your Mac orders are slipping, assume your next server quote is exposed to the same forces — and price the risk in before the budget is signed off, not after.

How UK businesses are actually coping

Four responses come up repeatedly, and they suit different balance sheets rather than being ranked best to worst.

1. Order earlier and standardise. The single most effective change is boring: shrink the number of configurations you buy and place orders further ahead. Unusual custom builds wait longest, so a fleet standard that maps to a commonly-stocked configuration will land weeks before a bespoke one. Standardisation also makes bulk allocation possible, which is how larger buyers are securing stock at all.

2. Specify what you need, not what you might. Memory is the constrained, repriced component. Habitually over-specifying RAM “to be safe” now costs materially more and pushes the order into the slowest queue. This is the year to size properly rather than round up.

3. Bring refurbished into the mix. Where a specification is fixed and the requirement is immediate, existing stock avoids the queue entirely because the memory in it was bought at a previous price. Our new versus refurbished calculator runs that comparison honestly, including the cases where it does not stack up.

4. Smooth the cost rather than absorb it. A front-loaded, higher-than-budgeted capital cost is the immediate problem for many organisations. Spreading it changes the cash-flow shape without delaying the hardware — the arithmetic is in our IT finance calculator.

The tempting fifth option — extend everything and revisit next year — deserves more scepticism than it usually gets. Deferring a refresh converts a known capital cost into a less visible operational one: more support incidents, more out-of-warranty repair, and a growing security surface as devices age past vendor support. That can still be the right call. It should be a decision with a number attached, not a default.

What to do this quarter

Three practical steps while the market is like this.

First, re-baseline the budget. If your hardware plan was costed before 2026, it is wrong — memory-heavy configurations have moved most. Second, move your decision dates earlier than your delivery dates by a realistic margin, and treat a vendor’s quoted estimate as a forecast that has already slipped for other buyers. Third, decide deliberately between buying now, extending, and buying existing stock, rather than defaulting into whichever the deadline chooses for you.

Our IT hardware refresh planner models a whole estate across those options, and our procurement team works to current UK lead times rather than list availability. If you are specifying laptops in particular, the business laptop finder filters by the configurations that are realistically obtainable.