A new report from TrendForce suggests that the iPhone 18 Pro BOM costs could surge by 38%, costing more for Apple to make. Apple won’t be alone under pressure from the memory crunch, as the situation is even worse for Android brands.

The bill of materials for the 256GB iPhone 18 Pro is projected to jump nearly 38% year over year, and memory is the single biggest reason why.

A year ago, memory made up roughly 10% of the total BOM for the 256GB Pro model. By the third quarter of 2026, that share has climbed to around 34%, and TrendForce expects it to cross 40% sometime in the first half of 2027.

For years, the processor and display were the two components that decided how expensive an iPhone was to build. Memory is now taking over that role, and it’s not a small move; it’s a complete reordering of where Apple’s money goes on the parts list.

Apple may prioritize shipments over margins

Apple has options here, and TrendForce thinks the company will lean on the same trick it used with recent MacBook launches: eat into gross margin rather than pass the full cost hike onto buyers.

There’s also a decent chance older iPhone models get quiet price bumps around the time the iPhone 18 Pro series launches, giving Apple another lever to soften the blow from rising memory costs.

Android smartphone makers are in a tougher spot. Memory prices have gone up five to sevenfold since early 2025, and budget and mid-range phones don’t have the pricing cushion flagships do.

TrendForce expects several brands to either raise prices sharply or quietly kill off models that have slipped into negative margins. Expect this pressure to keep dragging on global smartphone production right through 2027.

Samsung Medical Center has just given us another reason to believe the future of neurological screening, such as Parkinson’s disease, might not need a single dose of radiation.

A research team led by Professor Joo Eun-yeon (Neurology), along with Professors Kim Eung-yeop and Son Beom-seok (Radiology), in collaboration with Professor Kim Jae-rim from Sejong Chungnam National University Hospital, has found that Neuromelanin-sensitive MRI (NM-MRI) could help identify patients at high risk of developing Parkinson’s disease, without the need for radioactive isotopes.

The RBD connection

The study centers on REM sleep behavior disorder (RBD), a condition where people physically act out their dreams, sometimes shouting, kicking, or flailing their arms while asleep. RBD is often seen as an early warning sign of Parkinson’s, but here’s the catch: not every patient with RBD actually goes on to develop the disease.

That uncertainty has made early, reliable screening a genuine headache for doctors trying to figure out who actually needs closer monitoring.

Why MRI instead of PET

Right now, Dopamine Transporter (DaT) PET scans are the gold standard for evaluating Parkinson’s risk. But they come with real drawbacks: high costs, limited access at many hospitals, and the use of radioactive isotopes that make repeated testing less than ideal.

This is where the Samsung Medical Center team’s research comes in.

Neuromelanin is a pigment found in dopamine neurons, and it naturally declines as those neurons get damaged. By analyzing neuromelanin volume and signal intensity through MRI, researchers can indirectly gauge how much dopamine neuron damage has occurred, all without radiation.

What the study found

The team scanned 66 RBD patients along with 30 healthy individuals for comparison. The RBD group was further split based on dopamine PET results: 37 with abnormal findings and 29 with normal ones.

Neuromelanin volume and signal intensity dropped significantly only in patients who already showed abnormal PET findings, while those with normal PET results looked no different from the healthy control group.

Not a replacement, but a smart first step

The researchers were careful to clarify that NM-MRI isn’t meant to replace PET scans entirely. Instead, it could serve as a screening layer, helping identify which patients genuinely need detailed PET evaluation, cutting down unnecessary testing and freeing up resources for those who need it most.

Apple is reportedly in early talks with China’s ChangXin Memory Technologies (CXMT) to source DRAM chips. The company plans to use Chinese memory in select iPhone and MacBook models sold in China.

According to a Wall Street Journal report, Apple has been testing CXMT DRAM memory chips across product lines, including the iPhone and MacBook, with the goal of using them in devices sold within China.

Apple is also said to be seeking White House approval before finalizing any deal, given the sensitivity around Chinese semiconductor suppliers.

Notably, Apple wouldn’t be the first major name to turn to CXMT. HP and Acer have already started using the company’s memory chips in select laptops sold outside the US.

• Apple’s unsuccessful CXMT DRAM talks benefit Samsung and SK Hynix

The AI boom has sent memory prices soaring, and Apple hasn’t been immune. The company has already raised prices on some devices globally as component costs climb.

Some of CXMT’s memory products are now priced on par with, or even above, offerings from Samsung Electronics, SK Hynix, and Micron, largely due to tight supply.

• Samsung secures 39% DRAM market, Micron narrows gap with SK Hynix in Q2 2026

For now, CXMT’s production is running near capacity and mostly serving Chinese giants like ByteDance and Xiaomi, but the company plans to more than double capacity by 2028.

Samsung and NTT Docomo announced on August 10 that they’ve successfully verified a new AI-RAN-based technology that customizes networks for individual users, rather than treating every connected device the same way.

The new AI networks tech demonstrated by Samsung and NTT Docomo learns how each user moves and how they typically use their connection, then predicts when quality is about to dip before it actually happens.

The network adjusts things like frequency bands ahead of time. So if you’re about to walk into a dead zone while streaming a video, the system has already reshuffled your settings to keep the picture smooth.

Samsung and DOCOMO ran simulations back in January using data pulled from DOCOMO’s actual commercial network in Japan, along with local 5G test environments.

The results were notable: transmission speed drops fell from 13.1 percent down to 7.2 percent, nearly cutting the failure rate in half.

Rather than collecting and processing wireless environment information in bulk, which strains the network, the new process only pulls what’s relevant to a specific problem, easing the load considerably.

Looking ahead, both companies submitted their data processing methods to 3GPP, the body that sets global mobile standards, back in February, as part of broader 6G groundwork.

Samsung Research’s Jin-Gook Jeong called it proof that AI can meaningfully improve individual user experience, while DOCOMO’s Masafumi Masuda framed it as a step toward genuinely user-centric networks going forward.

Samsung Electronics’ HBM4 yield has climbed into the 80 percent range, a milestone reached roughly six months after mass production kicked off, and it’s shaking up a memory market SK Hynix has controlled for years.

What’s driving this jump is Samsung’s integrated approach.

Instead of memory, Foundry, and packaging teams working separately, they collaborated from the design phase itself, and that synergy appears to be paying dividends faster than expected.

SEDaily suggests Samsung also cracked one of its toughest technical hurdles, the Thermal Compression Non-Conductive Film process, managing to balance speed, power efficiency, and heat control all at once.

To be fair, SK Hynix hasn’t been sitting idle. Industry estimates suggest its own HBM4 yield has also touched the 80 percent mark, and the company continues to lean on its mass-production track record.

SK Hynix executives have stressed that real competitiveness comes from combining stable yields with consistent quality at scale, not just hitting a yield number once.

Samsung HBM4E

Samsung’s next-gen HBM4E has reportedly crossed 70 percent yield. The company is expanding capacity at its Hwaseong and Pyeongtaek facilities to meet demand from Nvidia, AMD, and other major AI customers.

Samsung Electronics achieved an 80 percent yield rate just six months into mass-producing its 6th-generation High Bandwidth Memory (HBM4), pushing the HBM market, previously dominated by SK Hynix, into a new phase.