Summary of "Por qué la RAM está tan cara"
RAM Price Surge: Causes and Implications
The video explains the dramatic surge in RAM prices, analyzing technological, market, and economic factors that impact the memory industry and computing costs worldwide.
Key Technological Concepts and Industry Analysis
RAM Price Explosion
- RAM prices have nearly quadrupled in the last year.
- Example: 32GB DDR5 6000MHz modules rose from $80 to $364.
- This increase significantly raises the cost of computers and laptops globally, especially affecting affordability in regions like Latin America.
RAM’s Role in Computing
- RAM serves as temporary working memory essential for:
- CPUs (serial computing)
- GPUs (parallel computing for graphics and AI)
- Storage devices
- While the rise in GPU and AI demand was anticipated, the sharp increase in RAM costs was unexpected.
Market Structure
- The RAM market is an oligopoly dominated by three major companies:
- SK Hynix
- Samsung
- Micron
- These companies are primarily based in South Korea and Taiwan.
- A rising Chinese company, CXMT, is emerging but mainly produces older RAM technology.
Manufacturing Technology
- RAM and chip manufacturing rely on extreme ultraviolet (EUV) lithography.
- EUV is a highly advanced process using lasers and plasma to restructure silicon wafers at the atomic level.
- This technology is extremely complex and expensive.
China’s Efforts
- China is heavily investing in developing its own EUV lithography technology, often referred to as its “Manhattan Project” for chip manufacturing.
- The goal is to reduce dependence on Korean and Taiwanese manufacturers.
- Currently, China produces many wafers of older RAM technology but little cutting-edge memory.
AI Impact on RAM Demand
- The explosion of AI models—from billions to trillions of parameters—has drastically increased RAM demand for both training and inference in data centers.
- AI’s hunger for memory chips surpasses all previous computing cycles, creating a significant supply-demand imbalance.
RAM Inventory and Economic Cycles
- RAM inventory has dropped sharply from 31 weeks in early 2023 to only 8 weeks by late 2025, signaling severe shortages.
- This scarcity parallels past tech cycles (e.g., webcams during the pandemic, internet boom, smartphone peak), but the current AI-driven cycle is much larger.
Investment and Factory Constraints
- Building new chip factories is costly and slow.
- Existing factories cannot be repurposed for new RAM technology.
- New factories for advanced RAM are expected to come online around 2027–2028, meaning price relief is not imminent.
- The cost of EUV machines has tripled—from $120 million in 2017 to approximately $400 million now.
Physical Limits of Miniaturization
- RAM and chip technology are approaching physical limits due to quantum effects (stochastic quantum noise).
- These effects prevent further shrinking of transistors with current lithography methods.
Apple’s Unified Memory Architecture
- Apple avoids the RAM shortage by integrating RAM directly into its chips (unified memory architecture).
- This approach combines CPU, GPU, and AI neural engines on a single silicon wafer, reducing dependency on separate RAM chips.
Shift in Chip Manufacturing Revenue
- TSMC’s revenue from smartphones dropped from 38% in 2023 to 29% in 2025.
- AI and high-performance computing (HPC) now account for 58% of TSMC’s revenue.
- Nvidia and AI companies pay premiums for chip production, competing with Apple.
Memory Wall in AI
- AI systems face a “memory wall” due to insufficient RAM supply.
- This limits the size and context length of AI models during training and inference.
Second-Order Effects
- Intel is leasing its factories to RAM manufacturers instead of focusing on its own chips.
- Apple may start producing its own chips in Intel’s factories.
- RAM shortages are increasing costs in other sectors such as:
- Automotive (smart car systems, autonomous driving)
- Gaming consoles (delaying new PlayStation and Xbox releases)
- Other PC components like power units and coolers are also affected due to increased data center demand.
Emerging Technologies
- New memory types such as RERAM, FRAM, and MRAM are in early development stages and may offer future alternatives.
Potential End of Custom-Built PCs
- The severe RAM shortage driven by AI demand could mark the end of affordable, custom-built gaming or AI PCs.
- This would impact enthusiasts and small-scale builders significantly.
Reviews, Guides, and Tutorials
- The speaker recommends Platzi’s software engineering fundamentals course for a deeper understanding of chip manufacturing and related technologies.
Main Speaker / Source
- The video is presented by a technology educator affiliated with Platzi.
- It references internal industry data and financial reports from companies like TSMC and SK Hynix.
- Insights into AI projects such as OpenAI’s Stargate are also discussed.
Summary
RAM prices have skyrocketed due to oligopolistic control, extreme manufacturing complexity, and unprecedented AI-driven demand. The shortage affects the entire computing ecosystem—from PCs and smartphones to cars and data centers. Factory investments and new technologies will take years to alleviate the crisis, with significant economic and technological ripple effects expected in the near future. Apple’s integrated chip approach offers a partial solution, but the broader market faces ongoing supply constraints and rising costs.
Category
Technology
Share this summary
Is the summary off?
If you think the summary is inaccurate, you can reprocess it with the latest model.
Preparing reprocess...