Samsung’s mobile division faces a potential first-ever annual loss as its own semiconductor unit prioritizes AI customers over internal orders. The Mobile Experience (MX) division head TM Roh recently warned senior management about the dire financial situation, according to reports from Korean media.
This internal conflict highlights a broader shift in the tech industry where smartphone makers are no longer the priority customers for memory suppliers. AI companies like Nvidia can pay premium prices and order massive quantities, fundamentally changing supply chain dynamics that have favored mobile devices for over a decade.
The trouble started in late 2025 when Samsung’s mobile division approached its semiconductor sibling for a long-term memory supply contract. The MX division wanted to lock in DRAM prices and supply for at least 12 months ahead of the Galaxy S26 series launch. But the DS semiconductor division refused, insisting on quarterly price negotiations that follow market rates.
DS executives made their position clear: external customers offer better prices, and business is business – even between Samsung divisions. This strict approach reflects Samsung’s long-standing policy of independent accounting for each business unit, designed to prevent inefficient cross-subsidization.
The semiconductor division’s stance makes financial sense. In Q1 2026, DS posted operating profits of $3.8 billion, more than seven times the previous year. Meanwhile, AI demand has created an unprecedented memory shortage, with companies willing to pay whatever suppliers ask.
Memory prices have skyrocketed across the board:
- LPDDR5x memory for flagship phones jumped from $33 in early 2025 to $70 by November
- All LPDDR memory prices nearly doubled in Q1 2026
- Industry insiders expect another 80% increase by Q2 end
- DRAM now accounts for 20% of flagship phone material costs, double the Q4 2025 level
The price surge stems from AI’s massive memory appetite. A single Nvidia Vera Rubin system requires 75TB of memory – equivalent to roughly 6,000 flagship smartphones. With companies like Nvidia, OpenAI, and Anthropic announcing multi-year data center projects, AI customers offer both higher prices and guaranteed long-term demand.
This supply crunch has already impacted the broader smartphone market. Global phone shipments fell 4.1% year-on-year in Q1, with rising component costs forcing manufacturers to increase retail prices by $150-$200 for flagship models. Even Qualcomm’s CEO recently flew to Seoul to secure memory supplies from Samsung and SK Hynix.
Samsung’s internal battle illustrates the company’s decentralized structure, where each division operates as an independent profit center. This system typically promotes efficiency by preventing profitable units from subsidizing struggling ones. But when market conditions create direct conflicts of interest between divisions, the policy can backfire spectacularly.
The semiconductor division now treats AI customers as premium accounts while viewing Samsung’s own phones and tablets as lower-priority orders. Without guaranteed memory contracts, the mobile division faces ‘price extortion’ from all suppliers, not just its corporate sibling.
This situation represents a fundamental shift in technology priorities. For the past decade, smartphone demand drove memory manufacturers’ expansion plans and determined chip specifications. But AI has flipped this relationship, making phones secondary customers in their own supply chain.
Samsung finds itself in a unique position as both one of the world’s largest memory producers and the second-biggest smartphone maker. The company’s commitment to market-based internal pricing means its mobile division gets no special treatment, even as it struggles with unprecedented cost pressures.