Samsung’s MX division (which makes its own smartphones) may suffer an annual loss for the first time — meanwhile, the folks at Device Solutions (where the memory business is located) have posted guidance for a record-breaking Q1. You have already understood that both these incidents have the same reason.
To make it clear, rising prices of DRAM and NAND memory are hurting margins. according to analysts competitionThe bill of materials (BOM) for a flagship phone ($800+) is projected to increase by $100-$150. 23% of the total memory is spent in RAM and 18% in storage.
According to industry insiders, TM Roh, head of Samsung’s DX and MX divisions, has warned management that the mobile division could post an annual loss.
Ratio of DRAM price to smartphone cost by price range: blue = sub-$200, purple = $400-$600, orange = $800+ (chart by Lee Ji-hye)
The flagship Galaxy S26 series is performing very well – it broke pre-order records in Korea, saw 25% more pre-orders in the US than the S25 series and saw 20% more pre-orders in Europe. In all three regions, the sales trend is in favor of the Galaxy S26 Ultra, which is the priciest of the three flagships.
Too, IDC The report said Samsung was the top smartphone vendor in Q1 this year with 62.8 million units shipped – despite a 4% decline in the overall market, a 3.6% increase in shipments compared to Q1 2025.
LPDDR5X is twice as fast as LPDDR4X
In any other year, such a performance would have been cause for celebration. However, Korean publications money todayWho broke the news puts things in perspective. The base configuration of the Galaxy S26 Ultra has 12GB of LPDDR5X RAM. However, a typical AI supercomputer requires the same amount of RAM as it takes to make 4,600 phones.
Low power DDR was primarily used in smartphones, tablets, and laptops. However, power and cooling are the two biggest costs of running an AI cloud infrastructure, so servers started using LPDDR as well.
This leads us to a problem – or a huge advantage, depending on what you’re after. Samsung is already shutting down its LPDDR4 production lines to increase LPDDR5 production capacity.
Smartphone memory cost share estimates across different price segments (source)
To make matters worse, the market is suffering from several issues at once – AI hyperscalers are ordering chips en masse, TSMC has limited capacity and Taiwanese semiconductor giants are raising their prices. This may have prompted Qualcomm to use Samsung fab for the upcoming Snapdragon 8 Elite Gen 6. this was another issue competition Highlighted – Flagship chipset prices are also rising, so flagship phones are in double trouble.