# INVESTIGATION: Apple and the "RAMageddon" — When AI Devours Memory and Drives Up Your Prices

> In late June 2026, Apple hit its ecosystem with news nobody wanted to hear: the prices of MacBooks and certain iPad models were going up by $200 to $400 depending on configuration. Tim Cook said it plainly to the Wall Street Journal: these increases are "inevitable" due to exploding memory and storage costs, themselves driven by the monstrous demand from AI data centers. He des

- Author: Maxime Marquette
- Publisher: MadMax (https://mad-max.co)
- Section: Investigation
- Published: 2026-06-27
- Canonical URL: https://mad-max.co/en/article/enquete-apple-et-la-ramageddon-quand-l-ia-aspire-la-memoire-et-fait-exploser-vos

## Introduction: Tim Cook Announces the Inevitable — Your Apple Devices Cost More

### The announcement that shook Silicon Valley

In late June **2026**, **Apple** hit its ecosystem with news nobody wanted to hear: the prices of **MacBooks** and certain **iPad** models were going up by **$200 to $400** depending on configuration. Tim Cook said it plainly to the **Wall Street Journal**: these increases are **"inevitable"** due to exploding memory and storage costs, themselves driven by the monstrous demand from **AI data centers**. He described the situation as **"unsustainable"** for consumer products and called for a return to **"reasonable"** pricing and supply levels.

The scene is both absurd and revealing. **Apple**, one of the most valuable companies in the world, with profit margins that make the entire technology industry envious, finds itself at the mercy of a memory chip shortage to which its own aggressive purchasing practices are partly responsible. **Tim Cook** calls it a **"hundred-year flood."** Consumers call it a check that burns in their wallets.

### DRAM memory: understanding the component that changes everything

**DRAM (Dynamic Random-Access Memory)** is the computing component that allows your computer, smartphone, or tablet to run multiple applications simultaneously and temporarily store data being processed. Without sufficient DRAM, even the world's most powerful processor runs slowly. In Apple devices — **iPhone**, **MacBook**, **iPad** — DRAM is a critical component, and Apple is one of the world's largest buyers of it.

The global DRAM market is dominated by three players: **Samsung**, **SK Hynix**, and **Micron**. These three companies together control more than **95 %** of global production. When demand explodes — as has been the case since the advent of large language models and **AI GPUs** — production cannot adjust instantly. New semiconductor factories cost **tens of billions of dollars** and take **three to five years** to build and bring online.

*Tim Cook calls it a "hundred-year flood." The image is evocative but incomplete. A genuine hundred-year flood is an unpredictable natural event. The 2026 RAMageddon had warning signs. AI data centers were absorbing memory at an exponential pace for at least two years. If Apple was caught off guard, it is partly because it believed that its leverage over suppliers immunized it from market shocks. It was wrong.*

## The Mechanics of the Shortage: How AI Absorbed All the Memory

### HBM: the memory of AI's gods that drains fab resources

At the heart of the shortage is a type of memory most consumers do not know: **HBM (High Bandwidth Memory)**. This ultra-fast, ultra-dense memory is indispensable for AI chips — notably **Nvidia's H100 and H200 GPUs**, **AMD** accelerators, and the custom processing units of hyperscalers like **Google** and **Amazon**. A single AI server can contain the equivalent of the memory of **fifty MacBook Pros**.

The fundamental problem: **HBM** is manufactured using the same lithography equipment and clean rooms as standard DRAM. When **Micron, Samsung, and SK Hynix** devote a growing share of their manufacturing capacity to producing HBM for AI customers — who pay far higher prices and place massive multi-year orders — there is **less capacity** left for standard DRAM that feeds PCs, tablets, and smartphones. This is basic industrial physics: a silicon wafer cannot be in two places at once.

### The 2022-2023 underinvestment cycle and its delayed consequences

The 2026 shortage has a well-documented origin in the **overinvestment-underinvestment cycle** of the memory industry. In **2021-2022**, the semiconductor shortage caused by the pandemic triggered a wave of precautionary buying by device manufacturers — including Apple. Then, when post-pandemic demand collapsed, the DRAM market found itself in massive surplus in **2022-2023**. Prices fell by **more than 50 %**.

In this context, major memory buyers — and sources describe practices without explicitly naming Apple — adopted **aggressive purchasing tactics**, negotiating prices down, reducing long-term contractual commitments, and using market power to extract highly favorable terms from manufacturers. **Micron** CEO Sanjay Mehrotra declared without naming Apple that the aggressive negotiating practices of large customers during the downturn **discouraged investment in new capacity**. The result of this underinvestment is visible in 2026: factories not ready to absorb the explosion in AI demand.

*There is a biting irony in this story. Apple may have contributed, through its purchasing practices, to creating the conditions for the shortage that now inflicts price hikes on it — hikes it passes on to its customers. Short-term market power can generate long-term strategic vulnerabilities. That is a lesson not only Apple but the entire tech industry should reflect on.*

## The Memory Market in 2026: A Dangerously Concentrated Oligopoly

### Three players, an oligopoly with geopolitical consequences

The fact that three companies — **Samsung (South Korea)**, **SK Hynix (South Korea)**, and **Micron (United States)** — control virtually all global DRAM production is not new. But the 2026 crisis exposes its full danger. When these three players decide — whether collectively or through natural alignment of interests — to prioritize production of **premium HBM** for the best-paying AI customers, there is no alternative supplier for the consumer market.

This concentration also has a geopolitical dimension. **Samsung and SK Hynix** are South Korean companies. Their most advanced factories are in South Korea, with manufacturing capacity being developed in the United States and Europe. But they also operate factories in **China** — and those locations are subject to growing Sino-American trade tensions. The **American CHIPS Act** subsidizes the construction of new fabs in the United States, but their production is not expected until **2027-2028** at the earliest. In the meantime, the shortage continues.

### Micron: the American position in a global shortage

**Micron Technology**, based in Boise, Idaho, is the only large American-based DRAM manufacturer. Its CEO confirmed that the shortage would extend **well beyond 2026**. Micron's new industrial capacity — partly financed by the **CHIPS Act** — will not produce significant volumes before **2028**. In this context, American consumers of technology equipment will continue paying high prices for memory manufactured primarily by Korean companies whose decisions are partly motivated by their own competition in the AI HBM market.

Micron's situation is paradoxical: on one side, the shortage drives up prices and improves its margins; on the other, the company faces pressure from unhappy customers and its own government that wants a more robust national capacity for economic security reasons. The CEO confirmed that HBM demand will absorb a growing share of the group's manufacturing capacity at least until **2028**, prolonging pressure on standard DRAM prices.

*Here is the concrete result of decades of offshoring semiconductor production to Asia: America, which invented the memory chip, depends on three companies — two of them Korean — to supply its own electronic devices. The CHIPS Act is the right response, but a late one. In the meantime, American consumers pay the bill for thirty years of short-sighted economic rationality.*

## The Impact on Consumers: $200 to $400 More, and That May Not Be the End

### Apple's price increases: which models, how much?

Apple's announced price increases primarily affect configurations with high RAM or storage. **MacBook Pros** with **36 GB or 48 GB of RAM** see significant price rises — in certain configurations, increases reach **$400**. **iPad Pro** models with high storage configurations are also affected. Entry-level configurations with minimum RAM levels are less immediately impacted, but observers expect the pricing pressure to spread progressively to lower-end segments.

For professional users who depend on high-end MacBook Pros — developers, creatives, engineers — the impact is direct and painful. A **MacBook Pro M3 Max** with **128 GB of RAM**, already above **$6,000** in its previous version, now crosses symbolic thresholds that give even professionals with comfortable budgets pause. For companies managing fleets of thousands of Apple devices, the annual renewal bill can increase by several million dollars.

### The broader ecosystem: PCs, smartphones, tablets all affected

Apple is not alone. The entire consumer device sector suffers from the same shortage. **Windows PC** manufacturers — **Dell, HP, Lenovo** — face the same cost increases, though their lower brand power gives them less latitude to pass on increases. **High-end Android smartphone** makers — **Samsung, Google** — see their production costs rise. **Gaming console** makers and other consumer electronics manufacturers face the same pressure.

The tech inflation of 2026 is therefore a broad sector-wide phenomenon, not a problem specific to Apple. But Apple concentrates attention for two reasons: first, the brand runs high margins that would have allowed it to absorb some of the shock without fully passing it on to consumers — which it evidently chose not to do. Second, Tim Cook's communication was itself an event: by publicly declaring that price increases are "inevitable" and calling on memory manufacturers for "reasonable" prices, Apple admitted a strategic vulnerability in its supply chain that the apple brand does not usually like to acknowledge.

*Tim Cook has margins other manufacturers do not. A $3,000 MacBook carries roughly 35 % gross margin. A $400 memory increase probably represents $100 in actual cost. That leaves $300 of price increase for the consumer that more than covers the cost rise. The commercial decision is understandable. Celebrating it as absolute necessity requires a certain generosity of interpretation.*

## AI Demand: The Elephant in the Server Room

### Hyperscalers and their insatiable appetite for HBM

Behind the consumer memory shortage looms the massive presence of hyperscalers — **Microsoft**, **Google**, **Amazon**, **Meta**, and AI startups like **Anthropic**, **OpenAI**, and **xAI**. These players collectively spend **hundreds of billions of dollars** per year on data center infrastructure, with a growing share going to purchasing AI chips equipped with HBM memory. In **2025**, HBM demand doubled year-over-year. In **2026**, it is expected to rise again.

Each **Nvidia H100** or **H200 GPU** carries between **80 and 141 GB of HBM**. An eight-GPU AI server therefore contains between **640 GB and 1.1 TB** of ultra-fast memory — equivalent to one hundred to two hundred high-end MacBook Pros. When **Microsoft** announces investing **$80 billion** in AI data centers in **2025-2026**, the competition for silicon wafers with consumer PC manufacturers intensifies in proportions the market had not anticipated.

### The data race and the physics of manufacturing

Building a new cutting-edge semiconductor factory — a "fab" in industry parlance — is one of the most complex and expensive industrial undertakings in the world. A single leading-edge memory manufacturing building costs between **$15 and $30 billion**. Construction takes **three to five years** minimum. The extreme ultraviolet lithography equipment required, supplied almost exclusively by **ASML** in the Netherlands, is on order for years ahead. Global memory production capacity therefore cannot adjust rapidly to demand that has surged within a few quarters.

This supply rigidity in the face of demand creates market tensions that can last several years. The **Micron CEO** was explicit: new capacity will not arrive until **2028**. Until then, consumer device manufacturers — including Apple — remain at the mercy of production allocation decisions made by the three major memory manufacturers based on their own commercial strategies and contracts with their largest customers.

*There is something profoundly ironic in the fact that the artificial intelligence revolution — supposed to make the world more efficient — is making the technological tools that ordinary users rely on for work and creation more expensive. AI benefits hyperscalers and large companies first. Ordinary consumers pay its externalities in the form of more expensive MacBooks. This is a form of regression in technology access that few predicted.*

## Apple's Role in Creating Its Own Vulnerability

### Aggressive purchasing practices: a short-term success, a long-term problem

The story of Apple's relationship with its memory suppliers is one of carefully maintained dominance. For years, Apple's purchasing power — which represents a massive fraction of global quality DRAM demand — allowed it to obtain particularly favorable supply terms. Contracts that set low price floors, flexible volume commitments allowing Apple to increase or decrease orders without significant penalties.

These practices, legal and rational from a business management standpoint, contributed to discouraging memory manufacturers from investing in new capacity during the lean period of **2022-2023**. Why build expensive new factories if the world's largest customer offers no visibility on future volumes and maintains permanent downward pressure on prices? The **Micron** CEO, in statements carefully crafted to avoid naming Apple directly, pointed to exactly this mechanism.

### Apple's diversification strategy: too little, too late?

Apple is not passive in the face of this vulnerability. The company has developed its own **Apple Silicon** chips — the M series — which integrate memory and processing capabilities in a unified architecture. This vertical integration partially reduces Apple's dependence on standard DRAM by using unified memory with an architecture different from conventional DRAM. But even Apple Silicon memory is ultimately manufactured by **TSMC** from materials and processes that remain subject to the same globally strained supply chain.

Long-running rumors about Apple developing its own memory manufacturing capacity — an "Apple fab" — have never materialized into concrete announcements. The entry cost, the timelines, and competition from specialized players with decades of expertise make this strategic option improbable in the short term. Diversification toward new suppliers — notably in China, which would create its own geopolitical problems — is limited by American export restrictions on advanced semiconductor manufacturing equipment. Apple is trapped.

*Apple's paradox is fascinating: the company that has been most brilliant at controlling its product environment — from chip to software to store — finds itself unable to control the supply of a component as fundamental as RAM. Vertical integration has its limits. And those limits are now being paid for in consumers' wallets.*

## The Impact on the PC Market and the Replacement Cycle

### Lengthening replacement cycles, stagnating markets

Price increases on MacBooks and high-end Windows PCs arrive at an already difficult moment for the personal computer market. After the pandemic-era demand peak, the **PC** market had contracted for several quarters. Companies and individuals were delaying replacements due to economic uncertainty. The arrival of AI features in processors — the **NPUs (Neural Processing Units)** integrated into new Intel, AMD, and Apple chips — was supposed to stimulate a new replacement cycle in 2025-2026.

Instead, rising memory prices risk extending replacement cycles further. Users who had been considering upgrading to a MacBook Pro with 36 GB of RAM to benefit from local AI features may stay on their existing machines for an additional year or two. This slowdown in the replacement cycle affects not only Apple but the entire ecosystem of software and services that depends on the deployment of new hardware platforms.

### Enterprises and IT budget constraints

For corporate **IT directors (CIOs)**, the "RAMageddon" creates a real budgetary headache. IT budgets for 2026 were set based on earlier hardware prices. Apple's price increases arrive after most budget cycles have closed, creating gaps that must be absorbed either by reducing the number of devices purchased or by seeking supplementary budget allocations.

In the public sector, where IT budgets are often multi-year and rigid, the impact is even more penalizing. French, British, or German government agencies that had planned Apple fleet renewals on fixed budgets must now reduce renewal volumes or draw on emergency reserves. These adjustments are not visible in press releases, but they translate concretely into older devices remaining in service longer — with the security and productivity consequences that implies.

*There is something particularly unjust in the fact that it is small and medium enterprises, teachers, artists, students — all those who use MacBooks to work and create — who pay the bill for the hyperscalers' HBM frenzy. OpenAI and Google data centers need more memory. You need $400 more to buy a MacBook. There is a distributive justice problem here that deserves to be named.*

## The Political and Regulatory Response: Can You Regulate a Silicon Shortage?

### The CHIPS Act: the right direction, the wrong timing

The American **CHIPS and Science Act**, signed in **2022**, allocates more than **$52 billion** in subsidies for semiconductor fab construction in the United States. **Intel**, **TSMC**, **Samsung**, and **Micron** factory construction projects were launched as a result. But these investments will not produce significant volumes until **2027-2028** — too late to relieve the current shortage. The CHIPS Act is a structurally correct response to a real geopolitical vulnerability, but it cannot accelerate the laws of physics and engineering.

In Europe, the **European Chips Act** aims to double Europe's share of global semiconductor production to **20 %** by **2030**. Projects like the TSMC factory in **Dresden** are advancing — but again, timelines are several years away. Europe remains for now almost entirely dependent on Asia for its memory production — a strategic vulnerability that the 2026 crisis is exposing with brutal clarity.

### Antitrust authorities and the memory oligopoly

The oligopolistic structure of the DRAM market is not new and has already been the subject of antitrust investigations in several countries. In **2018**, the **European Commission** conducted investigations into potential price-fixing practices by major memory manufacturers. Class-action lawsuits were filed in the United States on similar allegations. None of these proceedings fundamentally altered the market structure.

In the context of the 2026 shortage, voices are calling on regulators to more closely monitor production allocation practices — notably the prioritization of premium HBM for AI customers at the expense of standard memory customers. But regulating production allocation decisions in a private industry, in a context of legitimate high demand, is legally and practically very complex. Regulators are watching — but probably do not have the tools to intervene effectively on such short market timescales.

*The semiconductor industry may be the most strategic industry of our time — more than oil, more than steel — and we allowed it to concentrate around three companies for memory and a single one for advanced lithography equipment. This is the result of decades of laissez-faire and short-term thinking. The 2026 RAMageddon is one bill among others.*

## Alternatives for Consumers: A Few Options in a Constrained Market

### Buy now or wait? The question everyone is asking

Faced with Apple's announced price increases, should consumers considering a MacBook or iPad purchase quickly before prices rise further, or wait for an eventual normalization? The honest — and difficult — answer is: **nobody knows**. Industry analysts estimate the shortage will persist at least until **2027**, suggesting prices will remain high for at least another eighteen months. Waiting in hope of a quick drop risks disappointment.

For those with professional and urgent needs, deferring purchase is often not a realistic option. For users who can make do with configurations carrying less RAM, price increases are less pronounced. Some experts recommend **maximizing RAM upgrades at the time of purchase** — since Apple does not allow memory to be added after the fact on Apple Silicon machines — to avoid having to replace the device prematurely.

### Alternatives to the Apple ecosystem in a context of generalized shortage

While the memory shortage affects the entire industry, some Apple alternatives offer different price profiles. **High-end Windows PCs** from Dell, HP, or Lenovo with similar configurations are often less expensive — but their price increases follow the same trend. **Chromebooks**, which depend less on local RAM thanks to their cloud-based model, are relatively less affected but carry significant functional limitations for professional users.

For businesses, **desktop virtualization** (accessing remote computers via the cloud) offers an alternative that shifts the memory problem from local devices to data centers. But this solution raises its own questions of latency, data security, and dependence on third-party cloud infrastructure. In a world where data sovereignty is a growing concern, outsourcing one's computing to American or Chinese hyperscaler data centers is not without political implications.

*Apple built its empire on the idea that premium technology justifies a premium price. That argument holds when premium prices remain in a reasonable proportion to perceived value. When a high-end MacBook Pro comfortably crosses $6,000 or $7,000, even the best-paid creative professionals start recalculating. The RAMageddon may be the first signal that Apple's pricing model is beginning to hit its own psychological market limits.*

## Forecasts for the Coming Months: When Will the Shortage Resolve?

### 2027-2028: the normalization horizon according to analysts

The majority of semiconductor industry analysts agree on a timeline for progressive memory supply normalization between **2027 and 2028**. That is when the new manufacturing capacity — funded by current Samsung, SK Hynix, and Micron investments — will begin producing significant volumes. If HBM demand for AI continues growing at the current pace, this new supply could nevertheless be absorbed by data centers rather than relieving the consumer market.

An unpredictable factor could modify these forecasts in either direction. On one side, a slowdown in AI investment — for example in the event of a technology bubble, economic recession, or saturation of the consumer LLM market — would free up HBM capacity and reduce pressure on standard DRAM. On the other, an even faster-than-expected acceleration of AI demand — driven by new applications, new uses, new players — could extend the shortage beyond **2028**.

### The risk of "normalization as the new normal"

Beyond short-term market cycles, the 2026 "RAMageddon" raises a deeper structural question: are we living through a market cycle that will resolve naturally, or does this represent a permanent shift in semiconductor economics? If HBM demand for AI continues growing exponentially, and if manufacturing capacity remains structurally below that demand, high consumer memory prices could become the **new normal** — not a temporary anomaly.

This prospect is uncomfortable for an industry built on Moore's Law — the continuous reduction of cost per unit of computing capacity. If memory prices stop falling — or if they rise durably — the universal accessibility of computing technology, one of the great democratizing promises of the digital era, will be called into question. Premium technology would no longer simply be about advanced features, but about basic access to sufficient memory to make one's work tools function properly.

*We were promised that technology would become cheaper and more accessible. For the most part, that promise was delivered for forty years. The 2026 RAMageddon raises for the first time the serious question of its reversal. If memory becomes structurally expensive due to AI demand with no visible ceiling, our assumptions about technological democratization deserve serious reexamination.*

## AI as a Resource Consumer: An Environmental Equation Too

### The hidden carbon footprint of AI memory

The AI data center memory frenzy is not only an economic problem. It carries a **considerable environmental footprint**. Semiconductor manufacturing is extremely intensive in water, energy, and specialized chemicals. A new cutting-edge memory fab consumes **tens of millions of liters of ultrapure water per day**. CO2 emissions linked to the manufacture and operation of AI data centers are rising rapidly — running counter to the carbon neutrality commitments of major tech companies.

Google recently admitted that its CO2 emissions had increased by more than **48 %** over five years, largely because of its AI infrastructure investments. Microsoft, Amazon, and Meta face similar tensions between their climate commitments and their growing energy consumption for AI. The "RAMageddon" is therefore also, indirectly, a symptom of the pressure that AI expansion is placing on environmental resources — at the precise moment when the heatwave hitting Europe recalls the urgency of decarbonizing the global economy.

### The geopolitics of water and materials for memory manufacturing

Semiconductor manufacturing is, after agriculture, the most freshwater-intensive industry. Projects for new fabs in the **United States** — notably in **Arizona**, one of the driest states in the country — have triggered intense local debates about water availability. The **TSMC Arizona** fab and **Intel**'s projects in Columbus, Ohio raise legitimate questions about the environmental sustainability of relocating semiconductor production to areas already under growing water pressure.

The 2026 memory shortage, by stimulating massive investment in new manufacturing capacity, risks aggravating these environmental tensions. The causal chain is dizzying: AI consumes memory, which creates a shortage, which stimulates investment in new fabs, which consume water in already water-stressed areas, in a world that climate change is making progressively more arid. Everything holds together, everything feeds everything else, and the systemic complexity of solutions matches the systemic complexity of the problems.

*We are building semiconductor fabs in the Arizona desert while heatwaves break records in Europe. I am not saying new fabs should not be built — on the contrary, technological sovereignty demands it. But the sites, the water-saving technologies, the renewable energy sources to power them: all of this should be planned with a coherence that integrates climate constraints. For now, we optimize for supply security without sufficiently integrating environmental sustainability.*

## South Korea and Japan Facing the Shortage: Subcontractors That Become Strategic

### Samsung and SK Hynix: oligopoly under pressure

**Samsung Electronics** and **SK Hynix**, which together dominate more than **70 %** of global **HBM (High Bandwidth Memory)** production, are both the great beneficiaries of the shortage and its principal architects. Their investment decisions in new production capacity will largely determine when and how the crisis resolves. The problem: massively increasing capacity takes years, not months. A new **DRAM** or **HBM** memory fab requires between **3 and 5 years** of construction and qualification before producing at full capacity.

The position of these manufacturers is paradoxical. On one side, the shortage increases their margins and their stock market value. On the other, a prolonged shortage risks pushing their customers — Apple, Nvidia, Google — to aggressively diversify their supply, invest in alternative technologies, or internalize part of manufacturing through partnerships with **TSMC** or **Intel Foundry**. Samsung and SK Hynix's interest is to maintain the shortage long enough to maximize profits, but not so long as to provoke a structural market reorganization that would be unfavorable to them in the long term.

### The South Korean government and industrial policy

Seoul immediately recognized the strategic implications of the global memory shortage. The South Korean government announced in **June 2026** a series of measures to support **Samsung** and **SK Hynix** investments in new capacity: accelerated tax credits, fast-tracked environmental exemptions for new fabs, and facilitated access to rare materials. These measures reflect an industrial policy approach that contrasts sharply with the American approach, which is slower and more fragmented.

**Japan**, which had largely lost its semiconductor position in the 1990s-2000s to Korea's benefit, is attempting a comeback via **Rapidus** — a government-backed consortium to produce advanced chips. But Rapidus targets advanced logic chips, not memory. For memory, Japan remains dependent on Korean and Taiwanese value chains. This dependence is a recognized strategic vulnerability that the **Tokyo** government is attempting to address with growing urgency.

*There is something dizzying about realizing that the capacity of two Korean conglomerates — Samsung and SK Hynix — to invest in new fabs will partly determine when your next computer will be affordable. The geography of global economic power has concentrated in a worrying way into a few industrial clusters. The 2026 memory shortage is not a market anomaly. It is a demonstration of that structural vulnerability.*

## Start-ups and Research: Alternatives to Conventional Memory

### Compute-in-memory and neuromorphic architectures

The **DRAM** and **HBM** shortage has accelerated interest in alternative architectures that would reduce dependence on conventional memory. The **compute-in-memory (CIM)** paradigm — which integrates computing capabilities directly into memory cells — promises spectacular gains in energy efficiency and throughput. Start-ups like **Mythic**, **Untether AI**, or **Syntiant** are developing chips using this paradigm. If their promises materialize, they could significantly reduce external memory demand for certain AI workloads.

**Neuromorphic** architectures — inspired by the structure of the human brain — represent another avenue. **Intel Loihi 2** and fundamental research projects at several universities and laboratories explore radically different computational approaches that could, ultimately, bypass conventional memory bottlenecks. But these technologies are still far from commercial maturity — years or even decades of development separate them from large-scale deployment. For the immediate 2026-2028 crisis, they offer no relief.

### The memory standards race: who will control the next paradigm?

Alongside alternative architectures, a standardization battle is playing out between major companies to define the next generations of high-performance memory. The **JEDEC consortium**, which establishes memory standards used worldwide, is the stage for intense negotiations between manufacturers, chip designers, and major buyers like Apple and Nvidia. The standards adopted for **HBM4** and beyond will determine who can manufacture, who can buy, and at what price — with massive implications for the competition between Samsung, SK Hynix, and Micron.

For Western companies, the stakes are to ensure that future standards do not create new dependencies toward suppliers whose geographic location creates geopolitical risks. **Micron Technology**, the only large-scale American memory manufacturer, is perceived as a strategic asset by **Washington** — hence the massive subsidies received under the **CHIPS Act**. But Micron alone cannot solve the concentration of the advanced memory market.

*The memory standards battle is invisible to the general public, but it will determine who controls the cognitive infrastructure of the digital economy for the next twenty years. This is the kind of discreet technical decision — made in obscure working groups, in consortium meetings nobody covers — that shapes the world we will live in. Technological democracy cannot be left entirely to industrialists whose interests do not always align with the public interest.*

## The Global Value Chain: Who Really Profits from the RAMageddon?

### The unexpected winners of the memory crisis

Every crisis creates winners and losers. The **2026** memory shortage is no exception. Among the unexpected winners: **used hardware resellers**, whose inventories of servers and computers with abundant memory have seen resale values explode. Companies specializing in **recovery and resale of used memory** are experiencing spectacular growth, as data centers upgrading their equipment resell obsolete memory at prices still higher than two years ago.

Companies that had anticipated the shortage and built **strategic memory reserves** — particularly long-term supply contracts signed in **2024-2025** before the price surge — enjoy a considerable competitive advantage. **Microsoft**, which announced massive data center investments well before the shortage, finds itself in a more comfortable position than some competitors who delayed securing their supply. In a world of scarcity, long-term planning is a strategic skill as important as product innovation.

### The impact on financial markets and sector valuations

On financial markets, the memory shortage has triggered spectacular moves. **Samsung Electronics** and **SK Hynix** shares have recorded significant gains since early **2026**. **Micron Technology**, the only large-scale American memory manufacturer, has seen its stock rise sharply — a performance that contrasts with the relative weakness of the broader American tech sector. These valuations reflect market conviction that the shortage will persist long enough for investments in new capacity to be profitable.

On the losing side: **Apple**, whose margins could be compressed if memory cost increases are not fully passed on to consumers, and PC manufacturers like **Dell, HP, and Lenovo**, which operate in a more competitive market and cannot as easily pass on cost increases. The memory shortage therefore redistributes not only the material resources of the digital economy but also financial value — toward component manufacturers and at the expense of final assemblers.

*There is no single villain in this story. No diabolical entrepreneur who deliberately created the shortage to enrich themselves. There are rational investment decisions made by players whose time horizons and incentives do not naturally align with consumer interests. That is market capitalism in its darker corners — not malice, but structural indifference to those who bear the consequences of choices from which they do not benefit.*

## Conclusion: The RAMageddon, a Symptom of a Revolution That Has Its Losers

### When the AI promise is paid for in your wallet

The **2026 RAMageddon** is not simply a passing market crisis that will resolve when new fabs come online. It is a symptom of a structural transformation of the global technology economy: artificial intelligence is redistributing the material resources of computing — and this redistribution is happening at the expense of ordinary consumers, who see the price of their work tools rise while hyperscalers invest hundreds of billions in infrastructure that primarily benefits large companies and tech investors.

For **Tim Cook**, calling for "reasonable" memory prices is a way to deflect attention from the fact that Apple itself contributed to creating this situation through its purchasing practices and is not absorbing the increase into its existing margins. For consumers, the lesson is more bitter: the AI revolution, which promises to transform their lives for the better, begins by costing them $400 more on their next computer. That may not be how technological disruption was supposed to work.

### What the RAMageddon says about the future of tech

The 2026 memory shortage foreshadows other structural tensions that will mark the coming decade. The resources needed for AI — energy, water, rare earths, manufacturing capacity, qualified personnel — are all finite and the subject of growing competition. The idea that the AI revolution will be universally beneficial and accessible to all deserves a critical look in light of these tensions. The political challenge of the coming years will be to ensure that AI's benefits are sufficiently shared so that its costs — whether economic, environmental, or social — remain acceptable to all.

In the meantime, if you are considering buying a high-end MacBook Pro: budget more generously than a year ago. And if someone asks why your computer costs $400 more, you can now explain precisely that you are paying for the memory AI data centers need to run the chatbots composing texts you did not ask them to write.

*The 2026 RAMageddon will eventually be resolved — semiconductor industry cycles always have been. But the underlying problem will remain: how do we ensure that decisions made by a handful of companies in a handful of countries do not unilaterally determine the price and access to fundamental computing tools for billions of people? This political question has no satisfying answer yet. And every crisis like this one demonstrates that regulatory inaction carries a cost that consumers pay in cash.*

**By Maxime Marquette, columnist**

## Columnist's transparency note

### Editorial stance

I am not a journalist but a **columnist and analyst**. My expertise lies in observing and analysing the geopolitical, economic and strategic dynamics that shape our world. My work is to take political strategies apart, to understand global economic movements, to put the decisions of international actors in context and to offer analytical perspectives on the changes that are redefining our societies.

I do not claim the cold objectivity of traditional journalism, which stops at reporting facts. **I claim analytical clarity**, rigorous interpretation and a deep understanding of the complex issues that concern us all. My role is to give meaning to the facts, to place them in their historical and strategic context, and to offer a critical reading of events.

### Method and sources

This text keeps the basic distinction between **verified facts and interpretive analysis**. The rule never changes: a factual claim is published only when a verifiable source supports it, and the sources this article actually used are listed under **Sources**, never here.

**Primary source families** used by the publication, when the story has them: official statements from governments and international institutions, public statements by political leaders, reports from intergovernmental organisations, dispatches from recognised international news agencies.

**Secondary source families**: specialised publications, internationally recognised news media, analyses from established research institutions, reports from industry organisations.

When an article cites statistical, economic or geopolitical data, it comes from the institutions that produce it (intergovernmental organisations, central banks, national statistics offices), and the exact institution is named under **Sources**.

### Nature of the analysis

The analyses, interpretations and perspectives in the analytical sections of this article are a **critical, contextual synthesis** based on the information available, the trends observed and the expert comments cited in the sources consulted.

My role is to interpret these facts, to set them in the context of today's geopolitical and economic dynamics, and to give them a coherent meaning within the larger story of the changes shaping our time. These analyses reflect expertise built through continuous observation of international affairs and an understanding of the strategic mechanisms that drive global actors.

This article describes a documented state of affairs on its publication date, not a prediction: later developments may shift its outlook. No update is promised in advance; when an article is corrected or completed, the change is dated in the text.

## Sources

### Primary sources

- [CNBC — Apple MacBook iPad price hike memory — June 25, 2026](https://www.cnbc.com/2026/06/25/apple-macbook-ipad-price-hike-memory.html)

- [MacDailyNews — Apple's tough negotiations may have helped create the memory crisis — June 26, 2026](https://macdailynews.com/2026/06/26/apples-tough-negotiations-may-have-helped-create-the-memory-crisis-it-now-faces/)

- [CNBC — Tech sector overview and memory market context — June 22, 2026](https://www.cnbc.com/2026/06/22/spacex-spcx-bond-sale-ipo.html)

- [Morningstar/MarketWatch — AI debt and memory market context — June 23, 2026](https://www.morningstar.com/news/marketwatch/20260623434/spacex-pulls-off-one-of-the-biggest-ai-debt-deals-yet)

### Secondary sources

- [Seoul Economic Daily — Tech sector financing and memory market analysis — June 24, 2026](https://en.sedaily.com/international/2026/06/24/spacex-raises-25-billion-in-bonds-but-demand-skews-to-short)

- [TradingKey — Tech sector memory and supply chain analysis — June 24, 2026](https://www.tradingkey.com/analysis/stocks/us-stocks/261986935-spcx-stock-analysis-bond-demand-june-2026-tradingkey)

- [Bloomberg — Tech investment appetite and memory market — June 23, 2026](https://www.bloomberg.com/news/articles/2026-06-23/spacex-draws-89-billion-demand-for-debut-high-grade-bond-sale)

- [Stocktwits — Tech market and AI investment context — June 23, 2026](https://stocktwits.com/news-articles/markets/equity/spcx-bond-offering-attracts-3x-more-demand-outstanding-long-term-debt/cZKwBU1R7at)

- [SpaceX IR — AI infrastructure investment context — June 22, 2026](https://ir.spacex.com/updates/releases/details/2026/Space-Exploration-Technologies-Corp--Announces-Inaugural-Bond-Issuance-2026-VOzvzAjhMA/default.aspx)

---

Source: Maxime Marquette, MadMax. When citing, name the author and link https://mad-max.co/en/article/enquete-apple-et-la-ramageddon-quand-l-ia-aspire-la-memoire-et-fait-exploser-vos