The question of how much RAM to choose when buying a computer—referring to working memory, not solid-state storage—has shifted dramatically. A decade ago, the mainstream answer was at least 8GB, with 16GB being the preferred choice. Few could have predicted that by 2026, memory would become a sort of luxury, with brands selling 8GB laptops and even some manufacturers considering reviving 4GB RAM computers.

In essence, this is an extreme case driven by soaring memory prices. Major brands are scrambling to reduce the RAM capacity in their products, trying to squeeze out some profit from rising costs to prevent their products from being swallowed by losses. For the average user, however, shrinking memory capacity leads to a comprehensive degradation in the experience of devices like phones and PCs. While there are many ways to cut costs, targeting memory is clearly not the most rational choice.

**Phones Return to the 8GB Era, 4GB PCs Seem Unreal**

Before this wave of memory price hikes, there was a long period of low prices. At that time, phone manufacturers were equipping devices with large memory and storage, with 16GB+1TB combinations becoming selling points for many popular models. I recall mid-to-high-end models with second-tier flagship chips and 16+1TB configurations could be priced around 3000 yuan. For instance, the realme GT Neo5 16+1TB I bought in 2023, featuring a Snapdragon 8+ chip and 150W fast charging, was just over 3000 yuan—hard to imagine today.

But in 2026, some phone products have regressed to the 8GB era. For example, the recently released Honor 600 Yuanqi Edition starts with an 8+256GB configuration. Samsung's mid-range A57 model released in March offers 8+256GB and 12+256GB versions, with the latter priced 400 yuan higher.

Apple is no exception. Last year's iPhone 17 came standard with 8GB RAM, and this year's iPhone 17e still features 8GB. However, iPhones have historically had lower memory capacities than Android models, and the memory price surge has indeed delayed Apple's plans to upgrade phone RAM.

Of course, most new phones start with 12GB RAM. Excluding flagships, series like OPPO Reno16, Honor 600 Super Edition, and Pro Edition also offer at least 12GB. Correspondingly, the prices of mid-range models have quietly risen, generally exceeding 3000 yuan, especially after recent collective price adjustments by various brands.

The PC sector is similar. Previously, Apple, to support full-device compatibility with Apple Intelligence, generously equipped the base models of MacBook Air and Mac mini with 16GB RAM. However, this year's MacBook Neo only offers an 8GB RAM version, with no option to upgrade to 16GB even for an additional fee. Many believe this laptop is a product to clear inventory of the A18 Pro chip, which, originally used in the iPhone 16 Pro, integrates only 8GB of memory.

Meanwhile, brands like Microsoft, HP, Lenovo, and Honor have all launched notebook products with 8GB RAM this year. There's even more egregious news. Recently, PC brand Acer unveiled the Snapdragon notebook Aspire Go 15, with its spec sheet describing memory as 'up to 8GB RAM.' This phrasing seems to hint at a potential 4GB RAM version.

Acer's new Aspire Go 15 notebook, equipped with Qualcomm's Snapdragon C chip, targets the $300 entry-level notebook market. Consequently, its overall configuration feels heavily scaled back. For instance, it features a 15.6-inch large screen but only 1080P resolution. Its core selling points are low power consumption and long battery life.

Even so, launching a 4GB model on an entry-level notebook released in 2026 feels surreal. After all, 4GB RAM PCs were popular over a decade ago. My own 2015 MacBook Air started with 4GB, while the Windows camp had generally moved to 8GB, quickly entering the 16GB era in the following years.

Unsurprisingly, ordinary users have heavily criticized this move. In the comments section of IT Home, some users stated that 8GB RAM computers are barely usable, while 4GB notebooks are completely unusable. Others sarcastically suggested pre-installing Windows 7 or XP instead of Windows 11.

On platforms like Xiaohongshu, most people believe that for Android phones, one should avoid 8GB RAM, starting at 12GB and preferably 16GB. Although large-memory phones are now more expensive, the experience is significantly better than with small-memory models.

**Memory Reduction Leads to Degraded Daily and AI Experience**

Smartphone users have long grappled with the pain point of constantly expanding app sizes and perpetually insufficient storage. Today, apps from major internet companies often start at four to five hundred megabytes and quickly balloon to several gigabytes after some use. Mobile games are particularly heavy on storage, with titles ranging from 40-50GB to 100GB becoming increasingly common.

These increasingly bloated apps erode not just storage space but also RAM. Super-apps from large companies often integrate complex functional modules, aggressively competing for presence in the user's phone memory to meet KPIs. Often, opening just a few apps fills up the memory, leading to background app killing even on devices with large RAM capacities.

Taking my current Android primary phone as an example, with 12GB RAM, it's just barely sufficient for daily use. Opening too many apps leads to lag and background killing. Using an 8GB RAM phone today would make it difficult to guarantee smooth daily performance. In my view, among current smartphone hardware configurations, imaging and performance are somewhat excessive, while the bottleneck has shifted to memory. An 8GB RAM phone is like a barrel with a missing plank, significantly reducing its capacity.

The situation is similar for PC products. Many may not have noticed that Windows 11 has significantly increased memory requirements compared to older systems. Although Microsoft officially states that Windows 11 can be installed with a minimum of 4GB RAM, the company recently stated in a document that 16GB is the baseline for Windows 11, with 32GB being the 'carefree choice.'

The increased memory usage in Windows 11 largely stems from Microsoft's strong push for WebView2 technology. Many built-in system applications integrate the Chromium-based Edge browser engine (WebView2). Put simply, these Microsoft-developed software have a built-in Google Chromium browser. Google Chrome's product philosophy is 'trading space for time,' aggressively occupying computer memory and disk cache to improve operational efficiency.

Following Microsoft's lead, many third-party software have adopted similar strategies. In users' daily experience, this translates to Windows 11 feeling significantly more memory-hungry. Typically, a PC with 16GB RAM will see Windows 11 itself consuming 4-5GB even without any third-party software running. Even on a 32GB RAM computer, memory usage can approach half under low load.

Of course, Windows 11's memory usage is dynamic. If a user's device has only 8GB or even 4GB RAM, the system's memory footprint will be lower. However, in 2026, expecting a 4GB RAM computer to run Windows 11 smoothly is completely unrealistic. This is why Acer's subtle hint at a 4GB RAM budget notebook was met with collective ridicule and criticism from netizens.

Beyond daily performance, on-device AI implementation could also fail due to insufficient memory. As mentioned earlier, Apple previously increased the base RAM for Mac series products to 16GB specifically to prepare for AI. Similarly, the base RAM for iPad and iPhone has reached 8GB, with even budget models like the iPhone 16e featuring 8GB.

The memory redundancy reserved for AI on phones and PCs serves two main purposes: on-device AI and cloud-based AI. On-device AI is straightforward—running models and AI functions using local chip compute power, where memory is crucial. For PCs, the most suitable AI chips are, of course, Nvidia GPUs with their own VRAM, but they are expensive and not present in every device. Therefore, RAM often becomes the bottleneck for local AI performance.

Admittedly, 16GB RAM is still insufficient for running large local models but can handle some niche tasks with lower computational demands. Currently, on-device AI applications on phones are mainly in image processing areas like photo object removal and offline translation. On PCs, it's possible to run some relatively smaller-parameter large models. On-device AI is also applied in gaming scenarios like AI super-resolution and frame generation.

Even if phones and PCs adopt cloud-based AI solutions, memory consumption will still increase. As AI becomes a system-level, real-time, always-responsive feature in future devices, it will inevitably require a portion of memory resources to be occupied long-term. For example, Apple's future upgraded Siri, infused with Gemini, will not only be a smarter voice assistant but also a central hub capable of retrieving vast amounts of local files and data, which will significantly increase its consumption of local hardware resources like memory.

In summary, if users receive electronic devices like phones and PCs with significantly reduced memory capacity, it will inevitably lead to a substantial degradation in the overall experience. This, in turn, makes people even less willing to upgrade to new devices.

**When Will the Memory Price Surge End?**

This wave of memory price increases has created a mixed scenario. On one side, there's the surreal sight of SK Hynix employees receiving huge bonuses; on the other, device manufacturers and ordinary users suffer. The skyrocketing memory prices ultimately stem from supply-demand imbalance. The surging demand primarily comes from AI giants investing heavily in AI infrastructure, requiring a continuous supply of memory. On the supply side, capacity from memory giants like Samsung and SK Hynix cannot be increased in the short term, and they naturally prefer to prioritize producing higher-margin products like HBM memory.

As for terminal brands, they can only either consume previously stockpiled memory or purchase it at high prices. Apple has already used its financial power to sign long-term, high-price contracts with Samsung and SK Hynix and leveraged its massive cash flow to sweep the memory market. Other brands are streamlining their product lines, prioritizing allocating memory resources to higher-margin flagship and mid-range models where memory cost sensitivity is lower.

So, when will this memory price surge stop? The solution lies in addressing both demand and supply. Currently, the arms race in the AI field continues. However, AI infrastructure building cannot go on forever. After all, few current AI giants are profitable; even a powerhouse like OpenAI struggles with monetization. The moment the AI arms race slows down is when memory prices will fall. But when that moment will arrive remains uncertain.

On the supply side, a few giants like Samsung and SK Hynix control the core production capacity for memory chips. Theoretically, surging demand should prompt them to expand capacity. But the reality is that building new memory fabs takes a long time. More importantly, the memory industry has long been cyclical, with prices repeatedly hitting rock bottom. If Samsung and SK Hynix invest huge sums to expand capacity now, they face future risks from falling memory prices. Therefore, a safer choice for them is to prioritize producing products urgently needed by AI giants, like HBM, while selling standard memory at high prices while prices remain elevated.

For us as ordinary users, choices are limited. However, voting with our wallets by refusing to purchase products with reduced memory is something we can still do.