Tech • IA • Crypto
This new chip is driving up PC prices
9/10TL;DR
The surge in AI demand has redirected global memory production toward high-performance chips, sharply increasing consumer hardware prices and likely prolonging shortages until at least 2028.
KEY POINTS
AI Boom Reshapes Hardware Markets
Since late 2022, the rapid adoption of generative AI has driven an unprecedented demand for computing power. Modern AI systems rely on transformer architectures that require massive memory bandwidth, pushing demand for specialized components far beyond previous levels. This shift has caused prices for consumer hardware such as RAM, GPUs, and SSDs to rise by two to four times in many cases.
HBM Memory Becomes Critical Bottleneck
At the center of the الأزمة is High Bandwidth Memory (HBM), a vertically stacked memory essential for AI workloads. Unlike traditional RAM, HBM involves complex 3D stacking processes with extremely low manufacturing tolerance. A single defect can render an entire unit unusable, drastically reducing yields and limiting supply.
Production Shift Away from Consumers
Manufacturers have redirected production toward higher-margin enterprise clients. AI-focused memory has grown from just 1.5% of total output to nearly 100% in three years, effectively crowding out consumer-grade components. Some brands have even exited the retail market, signaling a structural shift rather than a temporary imbalance.
Highly Concentrated Supply Chain
The global memory market is dominated by three players: Samsung, SK Hynix, and Micron. Their limited number and technical barriers to entry give them significant pricing power. Building new capacity requires investments of around $10 billion per фаб and up to 24 months before production begins.
Complex Global Dependency Network
The supply chain involves multiple critical actors: ASML provides lithography machines, designers like Nvidia and AMD create chips, and TSMC handles fabrication and packaging. Memory suppliers must coordinate closely with all of them, making the system highly interdependent and difficult to scale quickly.
Tech Giants Drive Demand Surge
Companies such as Meta, Amazon, and Microsoft are purchasing hundreds of thousands of GPUs to train and run AI models. The size of these models has exploded, growing from around 300 billion parameters in earlier systems to over 1.5 trillion in newer ones, multiplying hardware requirements.
Unusual Purchasing Dynamics
Due to shortages, some suppliers have shifted responsibility downstream. Buyers must now secure memory directly before purchasing GPUs, reversing traditional supply roles. In extreme cases, companies have agreed to pay double the expected price upfront just to secure supply.
Consumers Bear the Cost
End users are absorbing the impact through higher prices across a wide range of products. Even unrelated devices such as projectors or storage drives are affected because they rely on similar components and manufacturing processes.
Historical Concerns Over Price Coordination
The memory industry has faced scrutiny before. In the early 2000s, major manufacturers were fined for operating a price-fixing cartel, with some executives receiving prison sentences. The current concentration of power has renewed concerns about limited competition.
China Faces Technological Barriers
While Chinese firms are developing alternative AI chips, they lag in HBM manufacturing. The precision required for vertical memory stacking represents a significant barrier, with development cycles estimated at 10 years or more, limiting short-term competition.
“War Economy” Dynamics Emerging
The market increasingly resembles a wartime economy, where critical resources are prioritized for strategic industries. AI has become a central battleground, diverting materials and production capacity away from consumer goods.
Limited Relief Before 2028
Although new factories are planned, industry estimates suggest shortages will persist until at least 2028. Prices may stabilize but are unlikely to return to previous levels due to sustained demand and structural changes.
CONCLUSION
The global race for AI dominance is fundamentally reshaping the semiconductor industry, concentrating resources on high-performance computing at the expense of consumers and locking in higher prices for years to come.