Apple hasn't named the next iPhone's price, but Tim Cook has now made the direction clear: the AI memory squeeze is moving from data centers into your phone bill.
The AI buildout is no longer something happening in someone else's server room. It's starting to show up in the price of the devices you buy. In a Wall Street Journal interview published June 17, Apple CEO Tim Cook said the company can no longer fully absorb the surge in memory chip costs, after trying to shield customers from supplier increases.
That is the part you should take seriously. Cook didn't say the iPhone 18 Pro will start at $1,299. Apple hasn't announced that phone yet, and final pricing is still expected at its September 2026 event. But when the company that built its modern supply chain on squeezing component costs says the situation has become unsustainable, you don't need much imagination to see where this goes.
Cook called the memory squeeze a hundred-year flood, according to the Journal. That sounds dramatic until you look at the numbers. Tom's Hardware, citing TrendForce, reported that DRAM contract prices are expected to rise 58 percent to 63 percent in the second quarter of 2026, while NAND flash prices are expected to rise 70 percent to 75 percent. Those are the two categories that sit inside phones, laptops, tablets and the AI servers now swallowing chip supply.
The pressure is coming from the hyperscalers. Goldman Sachs analysts, cited by Axios, put 2026 hyperscaler capex at about $770 billion, roughly equal to their operating cash flow. Business Insider recently reported Goldman expects Amazon, Alphabet, Meta and Microsoft to spend up to $725 billion this year, more than double their 2025 total. You can call that conviction if you want. Frankly, it also looks like a bidding war for the same factories that make the memory in your next iPhone.
Apple's problem is simple and ugly. High-bandwidth memory for AI servers uses the same broad manufacturing base as conventional DRAM, and suppliers make more money selling into Nvidia-heavy data center demand than into consumer electronics. SK Hynix, Samsung and Micron are not charities. If AI customers are willing to sign bigger and longer contracts, the memory makers will follow the money.
The iPhone math is already moving. The Wall Street Journal, using TechInsights estimates, reported that the iPhone 18 Pro's production cost could rise to about $726, up from roughly $582 for the iPhone 17 Pro. The 12GB of DRAM alone could cost $145, compared with $39 in the prior model, while 256GB of flash storage could rise to $51 from $13. That is not a rounding error. If Apple wants to protect its usual margins, a $1,299 starting price becomes a live possibility, with a higher figure possible if the camera system also gets more expensive.
Cook also ruled out the fantasy solution. Apple is not going to build its own memory fabs just because DRAM and NAND got expensive. A new memory manufacturing operation would take years, cost tens of billions of dollars and pull Apple into a commodity business it has spent decades avoiding. Apple designs its own processors because that gives it control over performance and product identity. Memory is different. It is brutal, cyclical and capital hungry.
The supplier side shows why this shortage won't clear quickly. Tom's Hardware reported in October 2025 that OpenAI's Stargate project had preliminary agreements with Samsung and SK Hynix for up to 900,000 DRAM wafers a month, a figure described as potentially equal to about 40 percent of global DRAM output. That is one project. Add cloud data centers from Microsoft, Amazon, Google and Meta, and the consumer electronics aisle starts looking like the customer at the back of the queue.
Apple is not alone here. Premium laptops, Android flagships, tablets, consoles and gaming PCs all lean on the same memory supply chain. If Apple raises prices before or around the iPhone 18 launch, Samsung's Galaxy line, Microsoft's Surface devices and other high-end hardware makers will have an easier time doing the same. Nobody likes being first. Everybody likes cover.
There is a harder question underneath the component squeeze. The cloud giants are spending hundreds of billions of dollars on AI infrastructure before the commercial returns are fully proven. Some of that spending may pay off. Some of it may turn into stranded capacity, ugly depreciation and shareholder arguments later. But the memory shortage is not theoretical. It is already priced into supplier contracts, analyst models and now Apple's public posture.
So when you see the next iPhone cost more, don't treat it as a normal Apple premium. Part of that price will be the cost of the AI race being passed down the chain. Amazon, Google, Meta and Microsoft are buying the memory first. You may end up paying for it second.
Also read: South Korea's food delivery duopoly faces its biggest antitrust reckoning yet • Tesco is suing Broadcom for over £100 million and it may be the lawsuit that breaks VMware's stranglehold on enterprise IT • EarnOS raises $18.5 million to pay real people for engaging with brands as bot traffic swamps the web
