Aerps.com/Unsplash

Global smartphone shipments fell to their weakest second-quarter level since 2013 on Monday as fresh data from Counterpoint Research and IDC confirmed that the AI-driven memory shortage has now fundamentally broken the economics of affordable devices — and that prices are not coming back down even after the shortage ends.

Counterpoint's preliminary Market Monitor, released July 13, put Q2 2026 shipments down 11% year-over-year, the steepest second-quarter decline in 13 years. IDC's Worldwide Quarterly Mobile Phone Tracker, released July 14, placed the absolute figure at 277.5 million units — a 6.7% year-over-year drop in IDC's methodology — with both firms agreeing that the sub-$200 segment is bearing the catastrophic share of the loss. What neither figure fully communicates is the structural turn underneath it: IDC has explicitly described this reallocation of silicon capacity as "potentially permanent." When new fab capacity comes online in 2028, memory manufacturers will face the same economic incentive to keep making high-bandwidth memory for AI data centers rather than return to lower-margin consumer DRAM. The floor is moving, and it is not moving back.

Why AI Data Centers and Your Phone Fight Over the Same Factory

The shortage traces directly to a single engineering reality: high-bandwidth memory (HBM) — the stacked, three-dimensional DRAM chip that sits alongside the processor in every NVIDIA AI accelerator — and the low-power DRAM (LPDDR5X) inside a smartphone are both produced by the same three companies, on the same DRAM fabrication lines, using the same limited supply of silicon wafers.

Samsung, SK Hynix, and Micron together control more than 95% of global DRAM production. Beginning in 2024, all three began systematically converting fabrication capacity toward HBM, because the economic case is overwhelming: HBM generates three to five times more revenue per wafer than standard mobile DRAM. Micron has publicly disclosed a 3:1 conversion ratio — producing one wafer's worth of HBM requires the fabrication capacity that would otherwise yield three wafers of conventional DDR5. Every AI accelerator shipped is, in that sense, three smartphones' worth of memory manufacturing capacity that no longer exists for consumers.

By mid-2026, analysts from CCS Insight estimated that AI data centers were consuming approximately 70% of all memory chips produced worldwide — up from 20–30% as recently as 2022. That estimate reflects not just HBM but also the large volumes of conventional DRAM server memory that hyperscalers like Microsoft, Google, Meta, and Amazon are buying under multi-year locked contracts, leaving consumer and smartphone OEMs to compete for whatever remains.

The result is a zero-sum allocation problem that IDC's analysts put directly: every wafer allocated to an HBM stack for an NVIDIA GPU is a wafer denied to the LPDDR5X module of a mid-range smartphone.

Why Budget Phones Get Hit Three Times

Not all smartphones suffer equally, and the mechanism behind the asymmetry is specific.

Memory and storage account for roughly 10–15% of an iPhone or Galaxy flagship's total bill of materials — painful when prices surge, but manageable. For a sub-$400 Android device, that proportion had already reached nearly 60% by the first quarter of 2026, according to Omdia's Quarterly Smartphone Technology Trends report. In the sub-$99 ultra-budget tier, the share has crossed 64%. When two-thirds of a device's manufacturing cost is tied to a single input category that has more than doubled in price in less than a year, there is no financial engineering available. There is nothing left to cut.

Budget phones face a compounding disadvantage that the draft coverage of this crisis has largely missed. The memory standard used in most affordable Android devices — LPDDR4X, a low-power DRAM designed specifically for cost-constrained mobile platforms — is not simply being deprioritized. It is being discontinued. TrendForce described the phase-out in its June 2026 bulletin as "an irreversible trend, signaling an impending and severe supply disruption." Samsung, SK Hynix, and Micron are not merely making less LPDDR4X — they are retiring the production lines for it.

Budget phone manufacturers cannot simply upgrade to LPDDR5X. The newer standard requires different chipset designs that most budget-tier platform vendors do not currently support. Transitioning to LPDDR5X also costs significantly more per gigabyte. The result is that entry-level device makers face a compounded squeeze: the memory they were built around is disappearing from production, and the replacement costs more and requires hardware they would need to redesign around. As Nothing co-founder Carl Pei put it when explaining the CMF Phone 3 Pro cancellation in June: "Brands now face a simple choice: raise prices by 30% or more, or downgrade specs. The 'more specs for less money' model that many value brands were built on is no longer sustainable in 2026."

Three Methodologies, One Direction: Down Sharply

The Q2 2026 data carries an important nuance worth stating clearly. Three major research firms measured the quarter and reached three different headline numbers: Counterpoint Research's 11% year-over-year decline is the sharpest reading; IDC's tracker puts the absolute shipment count at 277.5 million units, representing a 6.7% year-over-year decline by IDC's methodology; Omdia measured a 4% decline using different survey methodology and shipment definitions. These figures do not contradict each other — they reflect genuine methodological differences in what counts as a "shipment" — but combining them as mutual corroboration of a single decline rate, as some coverage has done, overstates the certainty of any one number.

What all three firms agree on: the direction is unambiguously down; Apple and Samsung are the only two major vendors in the top five that grew shipments; and the pain is concentrated at the bottom of the market, where price-sensitive consumers in emerging markets are absorbing the steepest increases.

"Q2 confirms exactly what we predicted: this is not a uniform downturn; the memory crisis is favoring premium players and punishing vendors exposed to the low end," said Nabila Popal, senior research director for worldwide consumer devices at IDC. Francisco Jeronimo, IDC's vice president for worldwide client devices, framed the structural divide: "This memory crisis has split the smartphone market in two. At the top, Apple and Samsung are pulling away because they secured supply early and sell where memory is a smaller share of the bill of materials. At the bottom, the vendors exposed to cheap, high-volume devices are absorbing the pain — and so are their customers."

Apple's Counterintuitive Win, Samsung's Strategic Position

Apple posted the quarter's most statistically surprising result. iPhone shipments rose 3% year-over-year, lifting Apple to a record 20% global market share for any second quarter in company history — despite the most adverse memory-cost environment any major OEM has faced in the modern era.

The mechanism was strategic discipline: Apple was the only major manufacturer to avoid raising smartphone prices during the quarter. While Xiaomi, OPPO, and vivo were hiking sticker prices to preserve thin margins, Apple held the iPhone 17 line steady. Consumers in a market defined by sticker shock responded by gravitating toward price stability.

Apple absorbed the cost through a combination of factors: long-term supply agreements secured with memory partners before the crisis peaked — Apple had reportedly locked in DRAM contracts 12–24 months in advance — a premium price architecture that leaves memory at a smaller share of its total bill of materials, and the financial cushion of a services business that provides revenue the company's hardware-only competitors cannot match. That hold, however, is not guaranteed to extend to the iPhone 18. Apple has already raised prices on Macs and iPads by as much as $300. CEO Tim Cook told the Wall Street Journal that memory costs have become unsustainable internally, signaling that the iPhone's price stability through Q2 2026 was tactical, not structural.

Samsung reclaimed the global number-one position with a 24% market share — its strongest quarterly showing against the top five in recent memory — posting the strongest year-over-year growth among the industry's five largest vendors. The Galaxy S26 series drove demand, particularly the S26 Ultra. Samsung also avoided major price increases in India and the Middle East, supported by aggressive seasonal promotions and better product availability than its rivals.

Samsung's vertical integration provides a structural hedge unavailable to pure-play handset makers. As both a memory manufacturer and a smartphone brand, Samsung's semiconductor division benefits directly from the same shortage that pressures its mobile business, giving the company a degree of supply-chain visibility and allocation control that brands like Xiaomi or OPPO simply cannot replicate.

Who Is Paying: Emerging Markets and the Sub-$100 Collapse

The human cost of the memory crisis is most acute at the bottom of the market. In India — one of the world's largest smartphone markets by volume, and one where sub-$100 devices are the primary means of internet access for hundreds of millions of people — the sub-$100 segment collapsed 59% year-over-year in the first quarter of 2026, according to CCS Insight data. Across Southeast Asia, Africa, and Latin America, retail price hikes in the 40–50% range are already documented, per IDC.

IDC's Kiranjeet Kaur, associate research director for worldwide consumer devices, noted that even as the top-line rankings for Xiaomi, OPPO, and vivo remain unchanged from last quarter, the pace of decline among Chinese vendors has accelerated. Most large Chinese players fell by double digits year-over-year in Q2.

Counterpoint's Shilpi Jain, who authored the Q2 preliminary, put the transition in the starkest terms: "What started as a components issue is now a full-blown demand issue. The global memory crisis has now overtaken every other factor as the single biggest drag on the smartphone industry."

The practical response from manufacturers is telling. Brands are cutting low-margin models entirely, reverting to older generation processors to save 30% on chip costs, switching from LTPO to LTPS OLED displays to save $3–5 per unit, and in the most extreme cases — as with CMF Phone 3 Pro — canceling devices outright rather than launching at a price point that would be inconsistent with their brand identity. The spec "shrinkflation" pattern that TrendForce warned about in early 2026 is now observable: a phone that would have shipped with 8GB of RAM and 256GB of storage in 2024 may now debut with 4GB of RAM and 128GB of storage at the same price point.

Xiaomi, OPPO, and vivo together held a combined market share that fell from 35% to below 31% in the quarter. Xiaomi's CFO had warned as early as late 2025 that memory cost pressures would force price increases, and the company has since simplified its lineup and reduced low-end volume exposure — a cautious bet that its Redmi Note 15, Redmi K90, and Xiaomi 17 series can partially compensate in the premium segment for volume lost at the entry level.

How Long the Shortage Lasts — and Why Prices May Not Fully Recover

Counterpoint expects global smartphone shipments to decline roughly 14% for the full year 2026 compared to 2025 — which Counterpoint's latest forecast (June 5, 2026) set at 13.9%, the worst annual contraction in smartphone history. IDC separately projects a 13.9% decline to approximately 1.09 billion units for the year. Both firms expect the memory shortage to continue into at least 2027.

New semiconductor fabrication capacity is coming — Micron's Idaho and New York fabs, SK Hynix's Indiana plant, Samsung's Texas facility — but the engineering reality is that a new fab takes three to five years from groundbreaking to meaningful commercial output. None of these announcements will provide relief before late 2027 or early 2028 at the earliest.

The more consequential question is what happens in 2028 when capacity does expand. Multiple analysts now project that even post-shortage normalization will not restore consumer DRAM prices to their 2025 levels. IDC calls the current reallocation "potentially permanent" and a "structural reset." SK Group chairman Chey Tae-won stated in March 2026 that the global wafer shortage is likely to persist until 2030. A Kearney PERLab analysis released in 2026, corroborated by CEO survey data, puts the shortage through at least 2030.

The economic logic behind this pessimism is simple: as long as AI data center spending continues to grow — every major forecast projects it will — chipmakers face a durable incentive to keep fabrication capacity in HBM production rather than return to consumer DRAM. HBM generates three to five times the revenue per wafer. The shortage is not being caused by manufacturers failing to produce enough memory. It is being caused by manufacturers choosing to produce a different, more profitable kind. That distinction matters for the recovery timeline: a demand-driven shortage self-corrects when prices rise; a choice-driven reallocation corrects only when the economic incentive to reallocate changes. That has not happened yet.

IDC's forecast is explicit: the sub-$100 segment, which accounted for more than 170 million shipped devices in 2025, "becomes economically unviable as memory and NAND costs settle at a permanently higher level, even after the memory shortage stabilizes in 2028."

What Buyers Can Actually Do Right Now

For consumers facing buying decisions, the structural nature of the shortage reshapes the conventional wisdom on timing. The standard advice — wait for the next model, prices will stabilize — has inverted in this market.

In the sub-$400 segment, buying now rather than waiting is the better financial strategy. The devices available today were built with earlier memory contracts; devices launching through the end of 2026 and into 2027 will incorporate components procured at the elevated prices now fully flowing through the supply chain. A phone available today at $299 may cost $375–$400 in its successor model, or launch with downgraded memory specifications at the same price. The refurbished market — specifically certified refurbished devices from 2024 and early 2025 — represents unusual value, as those devices carry pre-shortage memory cost structures.

For premium device buyers, the calculus is different but no less urgent. Apple has held iPhone prices steady through Q2 2026, but Tim Cook has already signaled that the iPhone 18 will face upward pricing pressure it cannot fully absorb internally. TechInsights estimated that 12 gigabytes of DRAM for the iPhone 17 Pro cost Apple $39 last year; the equivalent package for the iPhone 18 Pro is estimated to reach $145. Buying an iPhone 17 now to lock in current pricing before the iPhone 18 launch is a defensible hedge.

What is not defensible, for buyers who need a budget device and are in emerging markets waiting for price relief: the wait. The structural shift in memory economics does not self-correct on a one-to-two-year horizon. What feels like a temporary squeeze is, on the evidence available today, the new floor.

Frequently Asked Questions

Will smartphone prices come down in 2027 or 2028 when new fabs come online?

Not fully, and possibly not at all in the segment most affected. IDC describes the current shift as a "potentially permanent, strategic reallocation" of global silicon wafer capacity — not a temporary bottleneck. When new memory capacity comes online in late 2027 or 2028, chipmakers will face the same economic incentive that drove the reallocation in the first place: HBM generates three to five times the revenue per wafer of consumer DRAM. IDC explicitly projects that even after the shortage stabilizes in 2028, memory and NAND costs will settle at a permanently higher level than 2025 prices. The sub-$100 smartphone tier, which shipped more than 170 million devices in 2025, may simply not be economically viable at any point in the current forecast horizon.

Is now a good time to buy a budget Android phone, or should I wait?

Buy sooner rather than later if you need a sub-$400 device. Devices available today were built with components procured before the full price shock reached contract markets. Devices launching through the end of 2026 will carry higher memory costs in their bills of materials, and those costs will appear either as higher sticker prices or as reduced RAM and storage configurations at the same price. The refurbished market is particularly attractive right now: certified refurbished 2024–2025 devices carry pre-shortage component cost structures and typically deliver better specifications per dollar than new 2026 equivalents.

Why can't budget phone makers just use less memory to keep prices down?

They can, and many are — but the savings are limited and visible to consumers as spec downgrades. A phone reverting from 8GB to 4GB of RAM saves some cost, but not enough to fully offset price increases of 250% or more on LPDDR4 memory since early 2025. The deeper problem is that LPDDR4X — the specific memory standard used in most budget Android devices — is being discontinued by all three major DRAM manufacturers, not just deprioritized. Budget makers cannot easily substitute LPDDR5X because it requires different chipset hardware they would need to redesign around, and LPDDR5X costs considerably more per gigabyte. The result is a compounded disadvantage with no short-term engineering escape.

What happens to the billions of people in emerging markets who depend on sub-$100 smartphones for internet access?

This is the most consequential and least-covered dimension of the shortage. IDC projects the sub-$100 tier is becoming economically unviable. The practical near-term paths for price-constrained buyers are: older-generation devices at reduced prices (brands are keeping 2024-era models in circulation longer to fill the gap), refurbished devices, or devices with downgraded specifications. The longer-term risk is a reversal of the decade-long trend of expanding internet access through affordable mobile devices. In markets like India, where the sub-$100 smartphone market contracted 59% year-over-year in Q1 2026, the implications for digital inclusion extend well beyond consumer electronics.