---
title: "Budget Phones, \"Killed\" by AI"
type: "News"
locale: "en"
url: "https://longbridge.com/en/news/287775222.md"
description: "The stock prices of SK Hynix and Micron Tech have surged due to AI-driven demand, reshaping the valuation logic of the memory industry. Mid-to-low-end smartphone manufacturers face dual pressures from rising memory chip prices and supply being locked in by data centers. The golden age of budget phones is coming to an end. IDC predicts that global smartphone shipments will decline by 13% in 2026, with shipments of low-end devices potentially plummeting by more than 20%"
datetime: "2026-05-27T12:53:54.000Z"
locales:
  - [zh-CN](https://longbridge.com/zh-CN/news/287775222.md)
  - [en](https://longbridge.com/en/news/287775222.md)
  - [zh-HK](https://longbridge.com/zh-HK/news/287775222.md)
---

# Budget Phones, "Killed" by AI

On May 27, SK Hynix's stock soared by 11% at one point in the South Korean market, while Micron Tech surged 19% in the U.S. market, pushing the market capitalizations of both memory giants past the $1 trillion mark. Notably, SK Hynix's 12-month gain exceeded 1000%.

The valuation logic of the memory industry is being rewritten by AI: **Memory chips, which previously fluctuated with the cycles of smartphones and PCs, are now being pushed to the core of AI infrastructure.**

For mid-to-low-end smartphone manufacturers, the challenge is not just rising memory chip prices. The more critical issue is that **the same wafers and supply are being locked in by data centers at higher prices, with longer contracts and greater certainty. The narrative of high value-for-money—large storage at low prices and high specifications at low costs—is becoming increasingly difficult to sustain.**

**Memory production capacity is shifting en masse away from consumer electronics, signaling the end of the golden age for budget phones.**

In February 2026, market research firm IDC released a forecast: Global smartphone shipments in 2026 will decline by 13%, marking the steepest drop in history. Shipments in regions such as Africa and the Middle East could plummet by more than 20%, with the sharpest cuts occurring in the cheapest low-end models on the market.

## 01 Budget Phones Suddenly Become Unaffordable to Produce

Over the past few decades, mobile phones transformed from luxury items into basic devices affordable for everyone, largely relying on the continuous price declines of core components like chips, memory, and flash storage.

However, this long-term downward cost curve is now being pulled upward by AI.

How significant is this turning point? We need to look at a longer timeline to see it clearly.

In 1985, an American buying an IBM PC AT had to pay $6,000. Adjusted for the U.S. CPI, this is roughly equivalent to nearly $20,000 in purchasing power today. Yet, this expensive "behemoth" could only process 900,000 instructions per second.

Forty years later, on the streets of Lagos or Nairobi, ordinary people can easily buy a Tecno Spark Go from Transsion for $30 to $120. Its price is less than a fraction (0.3%) of that IBM computer, yet its computing speed reaches billions of operations per second, with performance thousands of times greater.

In the history of human commerce, no commodity has experienced such a dramatic collapse in costs. It was precisely these cheap devices that allowed hundreds of millions of the world's poorest population to access the internet for the first time.

For many years, the vast majority of memory chips went to smartphones and laptops. But in the last two to three years, artificial intelligence (AI) has begun to devour memory resources with extremely high profits, forcing a large-scale shift of memory production capacity from consumer electronics to feed AI. This has caused the cost of manufacturing phones to rise significantly compared to previous years. As a result, the budget phones that once delivered computing power and connectivity to the poorest corners of the world can no longer survive.

Ultimately, a smartphone is essentially a miniaturized computer. It has everything a computer does: a processor, memory for temporary data storage, a hard drive for data retention even when powered off, and circuit boards connecting various components. It is just smaller, with the addition of a touchscreen and radio modules.

Over the past few decades, processor performance has improved exponentially according to Moore's Law. However, no matter how fast the processor is, it needs a constant stream of data "fed" to it. This task falls to memory, specifically DRAM (Dynamic Random Access Memory).

However, an awkward problem exists: The improvement speed of DRAM has far lagged behind the evolution speed of processors.

In the 1980s and 1990s, processor speeds soared by 60% annually, while DRAM speeds increased by only 7% per year. Computer scientists coined a term for this speed gap: the "Memory Wall." For decades, researchers in computer architecture have spent much of their energy trying to bypass this speed gap between processors and DRAM.

## Why Is DRAM So Difficult to Manufacture?

Because the essence of a memory chip is actually a super-large array of storage units. Each unit contains a transistor and a capacitor, with the capacitor responsible for holding the electric charge that represents data. Although transistors can be made smaller and smaller, shrinking capacitors is extremely difficult. Once a capacitor becomes too small, it becomes harder to retain the charge steadily; the charge easily leaks, disappears, or gets interfered with by adjacent components.

To make DRAM more efficient, manufacturers have had to attempt various increasingly unusual structures, making modern DRAM manufacturing extremely complex and expensive. Today, building a state-of-the-art DRAM wafer fab casually requires an investment of $15 billion to $20 billion, not including the billions more needed for equipment like lithography and etching machines. Moreover, completing construction is not the end; it takes several years of repeated trial and error adjustments before yield rates are barely competitive enough for the market.

Furthermore, a more important issue is that memory is a "commodity," meaning it is interchangeable. Unlike Intel processors, which cannot be directly plugged into Apple devices as they are generally custom-made, DRAM chips follow a set of industry standards. Chips produced by any manufacturer can be directly inserted into devices from any other manufacturer.

The combination of "extremely high capital barriers" and "interchangeability usable by anyone" forms a fatal mix. Looking back at the entire history of DRAM development, it has been a series of life-and-death cycles from "boom to bust." Whenever demand in a certain sector pushes up prices, all manufacturers rush to invest and expand production. However, overinvestment inevitably leads to oversupply, followed by a price avalanche.

In this industry, every downturn cycle is a matter of life and death because production costs are simply too high. Even Intel, which was the leader in the memory sector in the early 1970s, could not withstand the pressure in the 1980s and had to exit to focus on processors. Texas Instruments and IBM also withdrew from the race in the 1990s. Later, Germany's Qimonda went bankrupt in 2009, and Japan's Elpida, once the world's third-largest DRAM manufacturer, fell into bankruptcy in 2012.

After decades of brutal consolidation and survival of the fittest, only three major memory manufacturers remain globally, controlling over 90% of DRAM production. Two are in South Korea: SK Hynix and Samsung. The remaining one is Micron Tech in the United States.

## 02 AI Devours Smartphone Memory Share

SK Hynix, Samsung, and Micron Tech have survived to this day precisely because they maintained restraint in capacity expansion during past crises.

The memory giants that fell in the previous DRAM reshuffle—Elpida and Qimonda—left a profound warning for the entire industry: Idle wafer fabs can kill a company, but unmet market demand will not. Therefore, when HBM order volumes soared in 2024 and early 2025, these three manufacturers cautiously chose not to expand production blindly.

It was not until memory prices skyrocketed at an unprecedented rate in 2025 that they began constructing new factories targeted at HBM. However, these factories will not produce chips until 2027 or 2028 at the earliest. Even with new factories, they are still avoiding over-expansion. In December 2025, Samsung executives emphasized "prioritizing long-term profitability over rapid capacity expansion."

Since they are not significantly expanding production, the only way to meet the overwhelming demand for HBM is to forcibly redirect wafers from DDR and LPDDR production.

Tech media Tom's Hardware reported in late 2025: "With total wafer startup rates flat and packaging production lines locked in, for every wafer manufacturers allocate to HBM production, corresponding capacity is removed from commodity DRAM."

By the end of 2025, SK Hynix had allocated 30% of its wafer capacity to HBM, almost entirely diverted from DDR and LPDDR. Micron Tech went further, directly halting production for its consumer brand Crucial in December 2025, stopping all shipments to consumers, and redirecting all capacity to AI and enterprise markets.

As a result, the memory available for traditional computers (DDR) and smartphones (LPDDR) shrunk sharply, causing prices to skyrocket wildly. From the first quarter of 2025 to the first quarter of 2026, within just one year, the price of LPDDR4 standard memory for phones rose by 250%, and LPDDR5 rose by 220%. In Germany, the price of DDR5 for personal computers surged by 414% in one year.

Memory instantly became the most inaccessible and expensive component in consumer electronics. In the total material cost of a budget Android phone, the proportion of memory jumped directly from around 15% to 50%.

This cost pressure hit marginal consumers the hardest.

Some budget phone companies have long survived by purchasing previous-generation cheap components on the spot market, assembling Android phones at extremely low costs, and then selling them at very low prices. Their profit margins are usually single digits, extremely thin, relying entirely on huge shipment volumes to survive.

Take Transsion as an example. It shipped 105 million phones in 2024 (compared to Apple's 230 million), capturing 48% of the African smartphone market single-handedly. However, when memory prices doubled and skyrocketed, this business model of "small profits and quick turnover" collapsed instantly. Smartphones priced below $100 faced "permanent economic unviability"—meaning each unit produced resulted in a loss.

To survive, manufacturers had to pass the costs on to consumers. Phones originally sold for $50 were forced to rise to $120 or even higher. For low-income consumers who are extremely price-sensitive, this directly exceeded their affordability, leading many to stop buying.

In early 2026, Transsion announced that its net profit for 2025 had dropped sharply by 54%, and helplessly cut its annual shipment target by 40%.

In the Indian market, the smartphone market segment priced below $100 shrank by 59% year-on-year in the first quarter of 2026. The pressure from memory costs forced the entire market into an awkward situation of "forced premiumization."

In Africa, 81% of smartphone shipments in 2025 were priced below $200. However, with the collective surge in phone prices, many African consumers found themselves unable to afford smartphones at all.

## 03 Apple Is Also Feeling the Pressure

The ruthless squeeze on capacity by HBM has not only blocked a large number of low-income consumers from the smartphone market but also caused tangible pain for tech giants higher up the DRAM food chain, who usually wield significant influence.

Samsung's own consumer electronics division is now unable to secure favorable long-term LPDDR agreements from its internal memory division. It was forced to reduce memory capacity and ship the Galaxy S26 at higher costs. Samsung executives even pessimistically warned that the mobile business might experience its first annual net loss in history. Dell also helplessly raised the prices of its laptops by 15% to 20% in December 2025.

Even Apple, which previously possessed immense bargaining power over Korean memory manufacturers due to its massive purchase volume, often signing long-term agreements to lock in prices for years, now faces a reversal of fortunes. The leverage has shifted entirely to the memory manufacturers. When Apple's long-term agreements expired in January 2026, memory manufacturers flatly refused to renew them, stating they would only sign short-term contracts of up to one quarter.

By February, to ensure the iPhone supply chain did not falter, Apple had to agree to pay a premium of up to 100% to Samsung for LPDDR5X memory used in iPhones.

Throughout 2025, the price of the 12GB LPDDR5X chips used in the iPhone 17 Pro rose by a staggering 230%. Burdened by this, Apple had to announce that the release of the standard iPhone 18 would be delayed until spring 2027, while the Mac Studio launch was dragged from summer to autumn.

Worse still, in the fourth quarter of 2026, Nvidia is set to launch its next-generation Vera Rubin platform. This is a rack-level AI supercomputer that combines the Rubin GPU and Vera CPU, specifically designed for large-scale AI training and inference. This Vera CPU will generate terrifying demand for LPDDR memory.

Industry forecasts predict that by 2027, a single Vera Rubin platform is expected to consume more LPDDR memory than the combined total of Apple and Samsung, the two smartphone giants. A report from JPMorgan Chase even gave a startling prediction: By 2027, memory costs could account for 45% of the total component cost of an iPhone, whereas currently, this ratio is only about 10%.

This means that within a year, Apple must face a painful choice: either cut into its own profit margins or wield the axe of significant price hikes against consumers.

The only variable at present may come from the Chinese market. Local Chinese memory manufacturers like Changxin Memory Technologies are rapidly expanding their scale and have already captured over 30% of China's LPDDR market. However, objectively speaking, as long as the bottomless memory black hole of global AI data centers persists, the cruel economic laws of DRAM shortage remain unsolvable. After all, those hyperscale cloud providers offer far more generous prices than budget phone manufacturers. Even Changxin Memory Technologies has planned to redirect about 20% of its capacity to HBM production.

In short, this wave of memory price hikes impacts price-sensitive populations and markets first. Low-income consumers, users in emerging markets, and small and medium-sized enterprises relying on low-cost electronic products will feel the pressure of rising costs first. But it won't be long before this "persistent high fever" of pressure is transmitted to each of us through expensive bills.

Risk Warning and Disclaimer

The market involves risks, and investment should be approached with caution. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment decisions made based on this content are the sole responsibility of the investor.

### Related Stocks

- [MU.US](https://longbridge.com/en/quote/MU.US.md)
- [PSI.US](https://longbridge.com/en/quote/PSI.US.md)
- [07709.HK](https://longbridge.com/en/quote/07709.HK.md)
- [SOXQ.US](https://longbridge.com/en/quote/SOXQ.US.md)
- [SOXL.US](https://longbridge.com/en/quote/SOXL.US.md)
- [XSD.US](https://longbridge.com/en/quote/XSD.US.md)
- [MULL.US](https://longbridge.com/en/quote/MULL.US.md)
- [SMH.US](https://longbridge.com/en/quote/SMH.US.md)
- [DRAM.US](https://longbridge.com/en/quote/DRAM.US.md)
- [FTXL.US](https://longbridge.com/en/quote/FTXL.US.md)
- [SOXX.US](https://longbridge.com/en/quote/SOXX.US.md)
- [688036.CN](https://longbridge.com/en/quote/688036.CN.md)
- [AAPL.US](https://longbridge.com/en/quote/AAPL.US.md)
- [SSNGY.US](https://longbridge.com/en/quote/SSNGY.US.md)
- [DELL.US](https://longbridge.com/en/quote/DELL.US.md)
- [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md)
- [JPM.US](https://longbridge.com/en/quote/JPM.US.md)
- [INTC.US](https://longbridge.com/en/quote/INTC.US.md)
- [TXN.US](https://longbridge.com/en/quote/TXN.US.md)
- [SMSN.UK](https://longbridge.com/en/quote/SMSN.UK.md)
- [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md)
- [JPM-M.US](https://longbridge.com/en/quote/JPM-M.US.md)
- [JPM-C.US](https://longbridge.com/en/quote/JPM-C.US.md)
- [JPM-D.US](https://longbridge.com/en/quote/JPM-D.US.md)
- [JPM-L.US](https://longbridge.com/en/quote/JPM-L.US.md)
- [8634.JP](https://longbridge.com/en/quote/8634.JP.md)
- [JPM-K.US](https://longbridge.com/en/quote/JPM-K.US.md)
- [JPM-J.US](https://longbridge.com/en/quote/JPM-J.US.md)

## Related News & Research

- [Micron joins $1 trillion club as AI race powers memory chip boom](https://longbridge.com/en/news/287651738.md)
- [Micron or Sandisk: One AI Memory Stock Looks Much Easier to Justify, Says Top Investor](https://longbridge.com/en/news/287805571.md)
- [Where Will Sandisk Stock Be in 5 Years?](https://longbridge.com/en/news/286990821.md)
- [BREAKINGVIEWS-SK Hynix rally lobs retail investors into AI fire](https://longbridge.com/en/news/287767060.md)
- [Beyond Nvidia: Micron, SK Hynix and the AI Stocks Reshaping the $1 Trillion Club](https://longbridge.com/en/news/287757743.md)