iPhone Prices Set to Rise: Here’s Why

When the next generation of iPhones hits the market this week, expect to see a price increase that many consumers won’t welcome. This potential hike signals that rising memory costs are becoming a significant issue, with no immediate relief in sight.

Dubbed “chipflation” or “RAMageddon,” this memory shortage is reversing a long-standing trend of decreasing memory prices that had previously allowed consumer electronics to become more powerful without a corresponding rise in costs. The terms “memory prices” and “memory shortage” were mentioned in 473 company earnings calls last quarter, indicating that the entire industry is grappling with this challenge, yet a solution remains years away.

The prevailing theory points to AI as a major contributor to this crisis. The technology is not only driving up energy costs and causing layoffs but is also pushing up prices for everything from smartphones to gaming consoles. However, the roots of this shortage were evident long before the rise of AI, which has only intensified the demand for RAM.

“We need to build more wafer capacity,” said Manish Bhatia, president and COO of Micron, one of the top three memory manufacturers. He noted that the industry now faces challenges that differ significantly from those of the past, where technology advancements could keep pace with demand.

For years, memory producers could ramp up production by fitting more chips onto each wafer, but those gains have diminished. In 2021, amid a surge in demand for electronics during the pandemic, Micron recognized that technological improvements alone wouldn’t suffice to meet long-term demand. Manufacturers would need to process more wafers and construct large new facilities to increase capacity.

Then, the memory market took a downturn. The spike in purchases of computers, tablets, and phones during the pandemic led to a temporary surge in demand, but as consumer spending declined, manufacturers found themselves with excess inventory, resulting in financial losses and a slowdown in expansion plans.

As the market began to recover, the rise of generative AI created an unprecedented demand for memory that far exceeded manufacturers’ expectations. The memory sector is also highly concentrated, with three companies controlling about 90% of the market. This leaves the industry reliant on a few key players to allocate limited resources between AI infrastructure and consumer devices.

The memory in your devices generally falls into two categories: DRAM (dynamic random-access memory), which temporarily holds data for active applications, and NAND flash memory, which provides long-term storage for files and photos. In AI data centers, specialized processors depend heavily on a type of DRAM known as high-bandwidth memory (HBM).

“It’s not as simple as saying data centers are consuming RAM,” explained David Naranjo, associate director at a market research firm. “The RAM is not the same.” HBM, which stacks memory chips and utilizes advanced connections, allows for rapid data transfer, making it more complex and costly to produce but also more profitable.

AI chipmakers like Nvidia and AMD, along with tech giants such as Meta and Microsoft, have a voracious appetite for HBM, often willing to pay a premium. Producing HBM requires significantly more manufacturing capacity than conventional DRAM, as it demands more silicon due to its stacked design. Micron estimates that creating a given amount of HBM requires about three times the number of wafers compared to conventional DRAM.

The incentives are increasingly skewed toward AI. Memory manufacturers can secure long-term contracts with some of the wealthiest companies, making it more appealing to prioritize HBM production. “We have been engaging in discussions with customers, prioritizing those who can guarantee committed future captive demand,” said Jaejune Kim, Samsung’s executive vice president of memory.

AI is also driving up demand for conventional DRAM, as phone and PC manufacturers seek to run smaller AI models directly on their devices, necessitating more advanced memory configurations. Future models may require significantly more memory to support these capabilities.

Both Samsung and Micron have expressed their commitment to conventional DRAM, with Micron stating it still represents the majority of their wafer capacity. However, the combination of limited supply, soaring demand, and AI customers willing to make long-term commitments has made it increasingly profitable to allocate resources to HBM.

Recent financial reports show that SK Hynix achieved a record 76% operating margin, up from 41% the previous year, while Micron’s adjusted gross margin reached an impressive 85%. Samsung’s semiconductor profits skyrocketed by approximately 250 times compared to the previous year.

Counterpoint estimates that DRAM prices for smartphones surged by about 56% in the first quarter of 2026 and around 83% in the second quarter. For instance, the cost of 16GB of DRAM for smartphones jumped from approximately $42 in the second quarter of 2025 to about $181 a year later, marking an increase of over 300%.

While these figures may not directly reflect Apple’s pricing strategies—given its strong negotiating power with suppliers—they illustrate the dramatic shifts in component costs for high-end devices.

The straightforward solution would be to increase memory production, but this process is lengthy and complex. Micron is currently demonstrating the scale of this challenge. In July, the company began construction on a massive manufacturing complex in upstate New York, which will feature 2.4 million square feet of cleanroom space, making it the largest semiconductor manufacturing site in U.S. history.

The cleanrooms will occupy only a small portion of the entire facility, which is roughly the size of 350 football fields. “By the time we’re done, we’ll have built probably 15 to 20 million actual square feet of building space to support the 2.4 million square feet of cleanroom,” Bhatia noted.

The construction process unfolds in three main phases. Ground preparation can take six to nine months, followed by another 18 months for the concrete structure and facade. The most complex phase involves installing the extensive mechanical, electrical, plumbing, and piping systems necessary for a semiconductor fab, ensuring that not a single microscopic particle contaminates the chips.

Bhatia likened the precision required to “trying to build all of New York state and not have any ants anywhere in the entire state.” However, even once the building is complete, it won’t immediately produce memory. Micron anticipates activating essential systems by the end of 2028, with pilot production beginning in 2029 and meaningful output not expected until 2030. This timeline means that the journey from permitting to production will take roughly six years.

Micron’s new facility in Idaho is further along and is expected to start wafer output in mid-2027. The company is investing heavily to expedite this process, doubling its capital expenditure to over $25 billion this year. However, even with this investment, Micron cannot pinpoint when supply will catch up with demand, which continues to grow rapidly.

“We’ve said that right now it’ll be beyond 2027, and we don’t see when it closes just because demand continues to grow so fast,” Bhatia stated. Market analysts expect that capacity won’t meet demand until late 2027 or early 2028 at the earliest.

PC and smartphone manufacturers can’t indefinitely absorb skyrocketing component costs. They have a few options: raise prices, reduce production, cut memory capacity, or focus on higher-end models with better profit margins. Recently, Microsoft raised Xbox prices by $100 to $150, with some models now costing up to $300 more than their launch prices. Meta also increased the price of its Quest 3 headset by $100.

Apple, with its significant leverage over suppliers and a customer base willing to pay premium prices, is a key case study. Despite its advantages, Apple has already raised prices across its Mac and iPad lines. Now, all eyes are on the iPhone.

Instead of launching its entire lineup this fall, Apple is expected to debut the iPhone 18 Pro and Pro Max alongside its first foldable iPhone, delaying the standard iPhone 18 and refreshed Air until spring 2027. This strategy concentrates Apple’s fall offerings at the higher end of the market, where it can better protect its profit margins.

Counterpoint’s supply-chain analysis indicates that Apple is preparing to produce about 10% more iPhone 18 Pro and Pro Max units for the September to December period compared to the previous year’s iPhone 17 Pro and Pro Max.

During its last earnings call, Apple acknowledged that supply constraints, including memory chip shortages, were impacting growth and that rising memory costs had directly affected its gross margin. Former CEO Tim Cook referred to memory pricing as a “100-year flood,” highlighting the “exponential increases” that have forced Apple to adjust its prices.

The iPhone 18 Pro could start at $1,299, which is $200 more than the iPhone 17 Pro, according to estimates based on component costs. Other manufacturers like Samsung and Google have also raised starting prices for new models by $100 this year, although many come with increased storage.

Apple’s strong market position may allow it to implement higher prices more effectively than its competitors. Its loyal customer base and integrated ecosystem make its devices feel more like luxury items than standard electronics. The company has gained market share by maintaining stable prices while lower-end competitors have raised theirs, making the premium iPhone a more appealing choice in markets like China, where alternatives are priced similarly.

For other manufacturers, the situation is more challenging. Analysts are already observing smartphone companies reducing shipment expectations while shifting focus to more expensive models. If they can’t sell as many devices, they’ll aim to maximize revenue from each unit sold.

This shift means more premium features like OLED displays, AI capabilities, and larger memory configurations, justifying higher prices. Consequently, revenue for PCs and smartphones could remain stable or even increase, despite declining unit sales.

In summary, AI is not just inflating electronics prices; the memory shortage is reshaping the types of devices companies choose to produce.

Eventually, more supply should become available. Samsung, SK Hynix, and Micron are all heavily investing in new manufacturing capabilities. However, since new facilities take years to complete, immediate relief is unlikely.

SK Hynix is investing 600 trillion won in its Yongin Semiconductor Cluster, a massive complex in South Korea, with plans to double its capacity within five years, although it still anticipates demand will outstrip supply through 2030. Samsung and SK Hynix are also planning to invest a combined 800 trillion won (around $588 billion) in four new memory-chip factories in southwestern South Korea. Micron’s projects in New York and Idaho are part of its commitment to invest over $250 billion in U.S. manufacturing and research through 2035.

A potential new player is emerging in China, with CXMT building additional fabs, though its technology is not yet on par with that of the leading manufacturers. Apple is currently testing CXMT chips but would need U.S. government approval to engage in business with them.

Alternatively, it’s possible that supply won’t need to catch up because demand could collapse first. If the AI investment bubble bursts, memory manufacturers might face a very different economic landscape.

For now, this remains speculative. The long-term contracts that AI companies have signed with memory suppliers provide a clearer picture of the immediate future.

“2027 is pretty locked and loaded,” Naranjo stated. Analysts expect price increases to slow from the rapid pace of the past year to single or low double digits. However, slower increases do not equate to falling prices.

“Not to say prices will come down to 2025 levels, but the price increases will stabilize at a new normal,” said a senior director at a market research firm. This new normal is “at least triple what they used to be.”

New manufacturing facilities take years to come online, while AI data centers and increasingly sophisticated consumer devices continue to demand more memory. Manufacturers have little incentive to prioritize cheaper consumer electronics over clients willing to pay a premium and commit long-term.

The outlook is clear: “It’s not a happy ending anytime soon.”