In the third quarter of 2026, the global technology industry witnessed a perplexing paradox.

On one hand, the consumer electronics end-market experienced an unprecedented wave of price hikes. From the iPhone 18 Pro series' starting price increasing by 1,000 yuan to Huawei, Xiaomi, and Honor—the three major domestic manufacturers—raising prices simultaneously, the average retail price of global smartphones surged by over 20% year-to-date.

Latest data from Counterpoint Research shows that global smartphone shipments in the third quarter of 2026 plummeted by 20% year-over-year, reaching the lowest level for the same period in fourteen years. However, the average market price rose against the trend, presenting a typical pattern of "declining volume but rising prices."

The core driver of this round of price hikes points directly to upstream memory chips—DRAM and NAND flash memory prices continued to soar, with storage costs accounting for a surge from the traditional 10%-15% to 30%-40% of a phone's bill of materials (BOM), nearing 50% for some flagship models.

On the other hand, SK Hynix, one of the world's top three memory chip giants, saw its stock price continuously plummet amid the widespread price increases. From its historical high in late June to mid-October, SK Hynix's market value evaporated by over 40%, with a maximum intraday decline nearing 50%. On July 29, the stock plummeted by 19% in a single day, marking the largest daily decline since the company's listing and even directly triggering the Korean stock market's circuit breaker mechanism.

Logically, product price hikes should benefit upstream suppliers, with stock prices expected to rise accordingly. However, with memory chip prices continuously climbing and smartphones collectively raising prices, why did industry leader Hynix's stock price instead fall relentlessly? Behind this seemingly contradictory phenomenon lies profound industrial logic and capital game.

The Cost Tsunami Triggered by Memory Chips

In September 2026, Apple's autumn event unfolded as scheduled, but the pricing of the iPhone 18 Pro series exceeded market expectations, with domestic starting prices rising by 1,000 yuan from the previous generation and top-tier versions surging by as much as 3,500 yuan. This was not an isolated choice by Apple but a collective move across the industry.

Ahead of the event, Huawei, Xiaomi, and Honor—the three major domestic manufacturers—had already taken the lead in raising prices for their main models, with increases ranging from 200 to 1,000 yuan. OPPO and vivo followed suit, adjusting prices for their flagship series. By then, all major global smartphone brands had joined the price hike camp.

This round of price hikes exhibited distinct structural characteristics: higher-end models saw greater increases than mid-to-low-end ones, with Pro series models rising more than standard versions. For instance, in the Android camp, flagship Pro series models generally saw price hikes of 500-1,000 yuan, while standard versions mostly increased by 200-500 yuan.

More notably, besides direct price hikes, manufacturers also controlled costs through "covert price increases via specification reductions," with some mid-range models' memory configurations dropping from 8GB to 6GB and storage specifications downgrading from UFS 4.0 to UFS 3.1, as if reverting to 2018 hardware levels overnight.

A senior analyst at Counterpoint Research pointed out: Severe memory shortages are pushing the smartphone market toward a more selective stage, reshaping OEM strategies and even influencing consumer behavior. Consumers are holding onto their phones longer than expected, increasingly considering refurbished models, and prioritizing high-value devices; meanwhile, OEMs are adjusting product portfolios, pricing, and market strategies to safeguard profitability.

This round of price hikes was not a proactive choice by manufacturers to pursue profits but a passive transformation forced by soaring costs of upstream core components. Among them, memory chips were the primary driver of price surges.

Over the past decade, memory chips typically accounted for only 10%-15% of a phone's material costs, with the processor (SoC) being the most expensive component. However, this ratio was completely rewritten in the current price hike cycle. According to Counterpoint Research, smartphone memory prices surged by over 80% quarter-over-quarter in the second quarter of 2026, with DRAM memory surpassing SoC processors to become the single most expensive component in flagship phones.

Specifically, the cost of 12GB LPDDR5X memory rose from about 200 yuan in early 2025 to nearly 600 yuan in the first quarter of 2026; 1TB UFS 4.0 flash storage also climbed from over 200 yuan to nearly 600 yuan. A mainstream 12GB+1TB storage combination saw procurement costs soar from about 400 yuan to around 1,200 yuan, a surge exceeding 200%.

TrendForce data shows that conventional DRAM contract prices in the first quarter of 2026 rose by 90%-95% quarter-over-quarter, while NAND Flash prices increased by 55%-60%. In the third quarter, traditional DRAM contract prices are expected to rise by another 13%-18% quarter-over-quarter, with NAND flash memory contract prices projected to increase by 10%-15%.

Entering the fourth quarter, not only did price increases not narrow but they accelerated. TrendForce forecasts a 10%-15% quarterly increase in general-purpose DRAM contract prices and a 15%-20% increase in NAND Flash contract prices in the fourth quarter, with overall DRAM contract prices (including HBM) expected to rise by 15%-20%.

The root cause of the memory chip price surge lies in a global capacity battle. The explosive growth of generative AI has spurred explosive demand for HBM (high-bandwidth memory), the core matching (supporting) storage chip for AI server GPUs, with bandwidth dozens of times higher than traditional DRAM and far higher unit prices and profit margins than consumer-grade storage chips.

Driven by profits, Samsung, SK Hynix, and Micron—the three major memory manufacturers—shifted 70%-90% of their advanced process capacity to HBM and server-grade DRAM, continuously squeezing consumer-grade DRAM and NAND flash memory capacity. Omdia data shows Samsung's NAND wafer output fell from 4.9 million units to 4.68 million units in 2026, while SK Hynix's dropped from 1.9 million units to 1.7 million units.

More critically, producing HBM of the same capacity consumes about three times more wafers than ordinary DRAM. This means that even if manufacturers shift 20% of their wafer capacity to HBM, the impact on consumer-grade storage supply far exceeds 20%. Bank of America data shows the DRAM supply-demand adequacy rate at only about 79% in 2026, well below the 100% equilibrium line. Citigroup expects the DRAM supply-demand ratio to deteriorate to -8.7% in 2027, a rarity in nearly 30 years.

The computing power demand from AI servers has created a powerful siphon effect, continuously drawing storage capacity away from the consumer electronics sector. Phone manufacturers can only compete for the remaining limited capacity, naturally having to accept rising prices. This is the underlying logic behind smartphone price hikes—not that phones are better, but that memory is more expensive.

SK Hynix's Stock Plunge: The AI Darling Falls from Grace

In the first half of 2026, SK Hynix was still the brightest star in the capital markets. Benefiting from the AI storage boom, its stock price soared from the end of 2025, reaching an all-time high on June 22, 2026, with a market value exceeding 2,100 trillion Korean won (about US$1.6 trillion), ranking among the top ten companies globally by market capitalization.

As the global leader in HBM market share, SK Hynix was seen as the "biggest winner in the AI era," with investors widely believing it would fully benefit from AI infrastructure construction.

However, the situation took a sharp turn downward from late June. On July 13, SK Hynix plunged by 15.37% on the Korea Exchange, marking its largest daily decline in 18 years and directly triggering the KOSPI index's circuit breaker mechanism. Just two trading days earlier, its ADR had debuted on NASDAQ with a 12.8% surge, reflecting unprecedented market enthusiasm.

On July 29, after SK Hynix announced its Q2 2026 financial results, its stock plummeted by 19% again, Refresh (setting) a new record for the largest daily decline. That day, the KOSPI index fell by over 6%, triggering the circuit breaker for the second consecutive trading day.

The decline continued into October. On October 13, influenced by news of technological breakthroughs by Chinese memory chip companies, SK Hynix fell by another 10.09%, with Samsung Electronics dropping by 7.39%. On October 14, SK Hynix fell by 3.9% in the U.S. market, with the memory sector collectively weakening.

By mid-October, SK Hynix's stock price had fallen by about 36% from its June historical high, with its market value evaporating by over 750 trillion Korean won (about US$570 billion). This decline far exceeded Samsung Electronics (about 23%) and was greater than the overall adjustment of the Philadelphia Semiconductor Index.

Ironically, the direct cause of the July 29 plunge was a "record-breaking" financial report. SK Hynix's Q2 2026 results showed: total revenue of 79.3 trillion Korean won, up 51% quarter-over-quarter and 257% year-over-year; operating profit of 60.5 trillion Korean won, up 61% quarter-over-quarter and 557% year-over-year; an operating profit margin of 76%, a new historical high; and net profit of 93.9 trillion Korean won, with a net profit margin of 118%.

By any metric, this was a super financial report worthy of entering the history books. What does a 76% operating profit margin mean? In the tech industry, even consumer electronics giants like Apple typically have operating profit margins in the 25%-30% range; NVIDIA, the biggest winner in the AI era, had an operating profit margin of about 72% in Q2 FY2026, lower than Hynix's.

The issue was that market expectations were even higher. Previously, Wall Street analysts generally expected SK Hynix's Q2 operating profit to reach about 64 trillion Korean won, with revenue around 84 trillion Korean won. The actual results fell short of expectations, with operating profit about 3.5 trillion Korean won lower and revenue about 4.7 trillion Korean won lower.

With the stock price at a historical high and market sentiment extremely optimistic, the phrase "below expectations" was enough to trigger selling. Moreover, investors were not focused on achieved results but on future growth momentum.

Some analysts pointed out that SK Hynix's failure to meet analyst expectations was precisely because its business exposure in high-end memory chips for AI data centers was higher than that of competitors. This meant that when traditional memory chip prices rose strongly, Hynix benefited relatively less, as more of its capacity was allocated to HBM, which uses long-term agreement pricing and has less short-term price elasticity than consumer-grade storage in the spot market.

The Logic Behind the Paradox of Hynix's Stock Plunge

Returning to the original question: Why is Hynix's stock price falling when phones and memory chips are rising in price? This seemingly contradictory phenomenon actually contains five layers of profound industrial and capital logic, each pointing to the same conclusion. Smartphone price hikes are not a positive for Hynix but rather an corroborate (affirmation) of market pessimism.

Many mistakenly assume that rising memory chip prices automatically mean higher profits for Hynix, but in reality, this "memory" is not the same as that "memory."

SK Hynix's revenue and profit growth are primarily driven by HBM and server-grade DRAM, not the LPDDR memory and UFS flash storage used in phones. In Q2 2026, Hynix's server-related storage business accounted for over 60% of revenue, with HBM business contributing nearly 30% of total DRAM revenue, and this proportion continues to rise.

In contrast, consumer-grade storage (including phones and PCs) contributes less and less to Hynix's profits. The reason is simple: HBM's gross margin is far higher than that of consumer-grade storage. A single wafer used for HBM is worth several times that used for phone memory. With limited capacity, Hynix inevitably prioritizes high-margin AI server orders, actively reducing its capacity share for consumer-grade storage.

Thus, smartphone memory price hikes are essentially the result of Hynix's proactive supply contraction. The higher consumer-grade storage prices rise, the more it indicates Hynix's reduced capacity investment in this area. The additional costs paid by phone manufacturers do not flow significantly into Hynix's pockets, as Hynix is selling less to phone makers.

More precisely, consumer-grade storage price hikes are "supply-contraction-driven" rather than "demand-pull-driven." The result of declining volume but rising prices is that Hynix's revenue growth in the consumer-grade storage market is far smaller than the price increase amplitude (magnitude).

The 20% plunge in smartphone shipments is not just a consumer electronics issue but an important indicator of global tech demand. Market concerns are: If even the terminal market like consumer electronics is so weak, how strong can the long-term demand for AI infrastructure construction really be? After all, AI technology ultimately needs to land in specific application scenarios and generate actual revenue to support sustained capital expenditures.

Phone price hikes lead to further lengthening of consumer replacement cycles, shrinking the global smartphone market size, which will transmit to the entire semiconductor supply chain. While current AI server demand is strong, if terminal application scenarios fail to open up and AI monetization falls short of expectations, the capital expenditure boom among cloud providers will eventually recede.

What investors see in the smartphone "declining volume but rising prices" is not storage manufacturers' pricing power but global consumption weakness. This weakness will inevitably transmit upstream—it's just a matter of time. The current memory chip price hikes are more driven by supply-side contraction than demand-side explosion. Supply-driven price hike cycles are often shorter and more fragile than demand-driven ones.

The memory chip industry is highly cyclical, a consensus in the capital markets. Historically, every memory price hike cycle has ended in overcapacity and price collapses. How long this AI-driven "super cycle" can last is the market's biggest point of contention.

Bulls argue that AI-driven storage demand is structural and long-term, fundamentally different from past cycles. The HBM supply-demand gap may persist for years, ushering in a long-term boom for the memory chip industry. Bears contend that no industry can sustain ultra-high profit margins of 76% indefinitely. High profits will inevitably attract new entrants and capacity expansion, eventually leading to a supply-demand reversal. Technological breakthroughs by Chinese manufacturers are accelerating this process.

The phenomenon of rising smartphone prices has, in fact, deepened the doubts of the shorts. When storage costs become so high that they start to suppress end-user demand, it indicates that prices have approached or reached the critical point that the end market can bear. Further price increases will only further compress demand and ultimately backfire on the upstream.

From a historical perspective, when terminal manufacturers begin to collectively pass on costs to consumers and configuration downgrades occur in mid-to-low-end products, it often means that the price increase cycle has entered its latter half. The excessive profits of the upstream sector are unsustainable, and the turning point of the cycle will inevitably arrive sooner or later.

In the first half of 2026, the market priced Hynix as an "AI growth stock." Investors believed that AI would unlock long-term growth potential in the storage industry, weakening its cyclicality and allowing Hynix to enjoy a high valuation similar to that of NVIDIA.

However, as the stock price plummeted, the market began to re-examine this pricing logic. More and more investors realized that memory chips remain fundamentally a cyclical industry. Even HBM cannot escape cyclical patterns. Cloud providers' capital expenditures are cyclical, and AI construction also experiences peaks and troughs. Once the major cloud providers complete their data center construction, HBM demand will similarly decline.

An operating profit margin of 76% is clearly unsustainable. Once the supply-demand balance reverses, profit margins could rapidly plummet from 76% to 0% or even negative, which is a historical norm in the storage industry. If valued as a cyclical stock, Hynix's stock price should command a low price-to-earnings ratio at the peak of the cycle, not a high one. This is a classic case of "all good news being priced in at the cycle top." The best performance often coincides with the highest stock price.

The collective price hikes of smartphones, as a signal from the latter part of the cycle, precisely remind investors that now may be the peak of this cycle.

In the long term, what worries investors more is the rapid rise of Chinese storage manufacturers. In the past, the global memory chip market was monopolized by Samsung, SK Hynix, and Micron, forming a stable "Big Three" pattern (translated as "Big Three" pattern for consistency, though it literally means "pattern of three giants"). The three companies tacitly controlled production capacity and adjusted prices to jointly maintain industry profits.

But now, this pattern is being broken. In the DRAM sector, ChangXin Memory Technologies has made rapid technological advancements, increasing its global market share from 4% to 10%, achieving the world's first mass production of LPDDR6, and preparing to launch its 4F architecture. In the NAND sector, Yangtze Memory Technologies has risen to third place globally in market share, with a quarterly profit of RMB 33.379 billion in the first quarter of 2026.

The rise of Chinese manufacturers means that the storage industry will shift from a "triopoly" to a "multi-polar competition." Historical experience shows that intensified competition will inevitably lead to a decline in overall industry profit margins. Even if AI demand continues to grow, Chinese manufacturers will capture a significant portion of market share and profits.

Against the backdrop of rising smartphone prices, domestic smartphone manufacturers have a stronger incentive to adopt domestic memory chips to reduce costs and ensure supply chain security. For Hynix, this means not only facing price competition but also losing market share.

The seemingly contradictory phenomena of collective smartphone price hikes and Hynix's stock price plunge essentially reflect the market's re-pricing of the storage industry's position in the cycle. When consumer terminals respond to rising costs with "reduced volume and improved quality" and when signals of weakening end-user demand become increasingly clear, investors have reason to question how long the upstream super boom can last. No matter how important memory chips are, they cannot exist independently of the broader tech industry and macroeconomy.

Cycles always repeat; there is no perpetual rise, nor perpetual fall. The key is to remain rational amid the noise and discern the underlying logic behind the phenomena. Smartphone price hikes are not due to value enhancement from technological progress but rather cost increases from capacity shifts; Hynix's stock price decline is not because the company has deteriorated but because market expectations have returned to rationality.