Broke a ten-quarter growth streak. Global smartphone shipments fell 2.9 percent year over year in the first quarter of 2026, according to IDC's Worldwide Quarterly Mobile Phone Tracker, the first decline the market had logged since mid-2023. IDC now expects the full year to come in nearly 14 percent below 2025, the steepest annual drop the firm has recorded. It is a chip allocation fight that smartphone makers are losing.
The Manufacturers Are Not All Losing the Same Amount
Five companies control most of the global smartphone market, and the 2026 shortage is sorting them by purchasing leverage rather than by brand loyalty.
- Samsung Electronics, first place. Reclaimed the top global position in the first quarter, helped by strong demand for its Galaxy S26 Ultra and the scale to lock in memory supply ahead of smaller rivals.
- Apple, second place. Held its position on the strength of premium pricing power and purchasing leverage.
- Xiaomi, third place. Posted the steepest shipment decline of any company in the top five, a decline the company partly engineered itself by deliberately scaling back older, lower-margin models rather than absorb the cost of a sweeping price hike across its catalog.
- Oppo, fourth place. Now integrated with Realme.
- Vivo, fifth place. Leaning on continued strength in China and a leadership position in India.
Below the top tier, the picture inverts.
- Honor posted the strongest growth of any company in the top ten, more than 24 percent year over year, as it pushed into overseas markets to offset pressure at home.
- Lenovo and Huawei also grew.
The pattern across the field: companies with either premium pricing power or aggressive overseas expansion are finding room to grow inside a shrinking market, while mid-tier Android brands selling mostly in the sub-$200 segment are the ones cutting targets.
That sub-$200 segment is where the damage concentrates. IDC expects regions with the heaviest concentration of low-end devices, the Middle East and Africa, Central and Eastern Europe, and Asia Pacific excluding Japan and China, to post double-digit declines approaching 20 percent or worse. North America, where 60 percent of shipments already sit above $800 and Apple and Samsung dominate, is forecast to decline by only about 6 percent. The shortage is hitting the part of the market that cannot pass rising costs on to price-sensitive buyers.
Three Companies Decide Who Gets a Phone Chip
Samsung, SK Hynix, and Micron Technology control more than 90 percent of global dynamic random access memory production, the chips that handle short-term data in nearly every phone, laptop, and server built today. All three have spent the past eighteen months shifting wafer capacity toward high-bandwidth memory, the stacked chip format that feeds AI accelerators in data centers, because that segment commands far higher margins than commodity phone memory ever did.
The reallocation is not subtle. SK Hynix told investors its memory capacity is essentially sold out for 2026. Micron has exited the consumer memory business outright to concentrate on enterprise and AI customers. Samsung's own memory division has warned that significant shortages across its product line will persist through at least 2027, with some customers already locking in allocations years in advance. This is the upstream mechanism behind the Xiaomi, Oppo, and Vivo shipment cuts Nikkei Asia reported in late June: low-power memory modules that phones depend on are competing directly against the high-bandwidth memory stacks inside a single AI server tray, and the server tray wins the allocation fight every time.
Low-power double data rate memory, known in the industry as LPDDR, sits at the center of the squeeze. It is the working memory inside every phone, built to run at lower voltage than the memory in a laptop or server so it can keep a battery the size of a stick of gum alive through a full day of use. It is soldered directly to the phone's logic board rather than installed on a removable module, which is part of why a phone's memory cannot be upgraded after purchase. For most of its history, LPDDR was a captive market that belonged almost entirely to mobile chipmakers. AI servers now need the same property, large amounts of fast memory that does not draw excessive power while feeding data continuously into a processor, and that overlap is what has pulled LPDDR into direct competition with the memory stacks inside AI accelerators.
Qualcomm and MediaTek, the two dominant mobile processor suppliers, have each pivoted meaningful engineering and go-to-market attention toward data center silicon over the past year, a sign that the chip industry's center of gravity has moved away from the device in your hand. Qualcomm's competitive position is unusual because it fights two separate battles at once. In mobile processors, its rivals are MediaTek and Samsung's own Exynos chips, with Apple's in-house silicon sitting outside the competitive set entirely since it is never sold to other phone makers. In data centers, where Qualcomm recently launched its Dragonfly brand, the competition shifts to Nvidia, AMD, and Intel, an entirely different field where Qualcomm is the new entrant rather than the incumbent.
Underneath both businesses sits a royalty stream that has nothing to do with whose chip ends up inside a given phone. Qualcomm holds a vast portfolio of patents covering the cellular standards, 3G, 4G, and 5G, that every connected device depends on. Nearly every smartphone maker pays Qualcomm a licensing fee on every unit sold, regardless of which company's processor is actually running inside it. That licensing business carries far higher margins than chip manufacturing, since it has no production cost attached, and it gives Qualcomm a stake in the outcome of the smartphone market no matter which brand wins any given quarter.
The Phones Are Designed in One Place and Built in Several Others
Smartphone assembly remains heavily concentrated in Asia. Mainland China accounts for the largest share of global smartphone export value, with Hong Kong, Vietnam, and India following behind it. Asia overall supplies the large majority of exported smartphones by value, while Europe and North America together account for a modest fraction, making both regions structurally dependent importers rather than producers.
A handful of contract manufacturers do the physical building regardless of whose logo is on the back of the device. Foxconn, also known as Hon Hai, remains the largest electronics assembler in the world and the dominant builder of Apple's iPhone line. Luxshare Precision, a mainland Chinese assembler that has absorbed manufacturing operations divested by Taiwanese rivals, has grown into a major iPhone assembler in its own right. Tata Electronics has emerged as India's anchor assembler, acquiring facilities from Wistron and Pegatron and signaling interest in assembling Android devices beyond its existing Apple work. India's share of smartphones shipped to the United States has climbed over the past two years, part of a broader diversification away from near-total reliance on Chinese assembly.
That diversification carries a complication worth sitting with.
Foxconn's own financial results explain why. The company has reported its strongest revenue growth coming from AI server production for hyperscale data center customers, even as its consumer electronics assembly business has softened. The largest contract manufacturer in the smartphone supply chain is reallocating its own capacity and capital toward the same AI infrastructure boom that is starving phone makers of memory. The assembly layer and the component layer are both tilting toward the same customer.
Where the Dependency Sits
Strip away the brand names and the dependency map is short. Three companies make almost all the world's DRAM. Two mobile processor designers, Qualcomm and MediaTek, supply the chips inside most non-Apple, non-Samsung phones, and both are now prioritizing data center silicon roadmaps. A handful of contract assemblers in China, Vietnam, and India build nearly every phone regardless of brand, and the largest of them earns more from AI server racks than from handsets. Apple and Samsung are the only two companies in the global top five with enough purchasing scale and in-house chip design to negotiate around the shortage rather than absorb it directly.
Everyone else in the smartphone business is, in effect, a downstream customer of the AI infrastructure buildout, whether they describe themselves that way or not.
The Replacement Cycle Was Already Stretching Before the Shortage Hit
The shortage is colliding with a separate, longer-running trend: people are keeping phones longer than they used to. Component suppliers told Nikkei Asia that demand for aftersales parts and refurbished handsets has risen this year, particularly in China, as buyers extend the life of devices they already own rather than absorb a price increase on a new one. That behavior predates the current memory crunch. Hardware has matured to the point where a three-year-old flagship phone still runs current software, on-device camera and battery improvements have slowed, and the marginal reason to upgrade has weakened independent of price.
The shortage now reinforces that behavior rather than causing it. Higher prices on new devices make an already-extending replacement cycle stretch further still. The two trends compound each other instead of canceling out.
What This Means for Wearables and the Next Device Category
The same component math applies to wearables, smart glasses, and any other connected device that depends on the same memory and processor families as phones. Foldable phones remain a growth pocket because they justify premium pricing that absorbs higher component costs, which is why Apple, Samsung, and the leading Chinese brands keep investing there even as overall volumes contract. Wearables face a tighter version of the same allocation problem phones face now, with even less purchasing leverage to negotiate supply, since most wearable makers buy at a fraction of smartphone volume.
On-device AI assistants are the wildcard. They increase the processing and memory demands of every device category at the exact moment memory is scarcest, which means the next wave of AI-capable wearables and phones will likely launch at higher price points and lower initial volumes than their predecessors. The components to build them at scale are committed elsewhere first. Demand is not the constraint.
The Enterprise Question Nobody Has Priced In Yet
Enterprise mobile fleets, IoT terminals, point-of-sale devices, and edge hardware draw on the identical memory supply chain now prioritizing AI data centers over consumer electronics. Most enterprise procurement planning still treats device refresh cycles as a budget and policy question. The 2026 shortage makes it a supply chain question first.
