Why the cost of laptops, phones and game consoles keep rising and how AI is making it worse
In the past, the price of technology would decrease over time. Production lines have matured, components have achieved economies of scale, and demand has shifted towards more modern designs; This meant the one-year-old laptop got a price drop for Christmas or the four-year-old games console got a new redesign that sold for half the price.
When a device’s price changes these days, it almost always goes up.
Features of popular consumer laptops have increased over the past four years, but not in keeping with price increases approaching $1,000 in some cases. Samsung’s Galaxy S26 smartphone $300 more than S22.
It is becoming more common to see price increases in the middle of the cycle; This means the price of the same model increases because the manufacturer cannot wait until the next product launch to adjust the cost. On Thursday, Sony increased the price of the PlayStation 5 in Australia to $1,000. It was launched in 2020 at $750.
It’s been a long time since the old tech buying advice to “wait for a price drop or sell-off” actually worked, but now we’re at a point where that advice is being reversed. If you’re likely to want an upgrade before 2028, you should buy today if possible.
Explaining the AI tax on RAM
The price of everything is increasing, but when it comes to computers, the biggest increase was in RAM. RAM, often referred to as memory, is a fundamental part of every computing device and stores data related to actively used applications so that the CPU can access this data at very high speed. It is difficult to manufacture, as most of it is produced by two major suppliers in South Korea.
The appetite has grown over the years, with the same types of chips used for PC RAM sticks becoming an integral part of graphics processors, phones and even cars. But the insatiable appetite of new AI data centers has strained production into uncharted territory.
Tech giants are spending billions of dollars building new facilities with extraordinary amounts of high-bandwidth memory, so RAM manufacturers have shifted their production lines to serve them. These aren’t the same chips installed in your new smartphone or laptop, but they use the same manufacturing infrastructure.
As a result, niche high-speed consumer RAM is becoming even more niche (see the Xbox Ally
Reportedly, the high-bandwidth memory used for AI hyperscalers uses approximately three times the wafer capacity compared to consumer DRAM. Translated, this means that every gigabyte of memory produced for the tech giants takes up three gigabytes of production space that could go into your devices. By the way, hyperscaler facilities are measured in petabytes, not gigabytes. That’s one million gigabytes.
The Stargate project alone, OpenAI’s plan to open massive data centers in the US that are currently at risk, was recently estimated to take up 40 percent of global RAM capacity.
What does this mean for your purchasing power?
Prices have been rising for years, and industry watchers have been warning before, but while there is time to prepare, there is no quick way to increase RAM production capacity. New facilities being built may not produce consistently high volumes for years, and AI data centers always need more, so they can eat them too.
So for every gigabyte you got, a company had to take a dump truck of money to Samsung or SK Hynix to get some line capacity.
As these manufacturers prioritize high bandwidth, remaining capacity is likely to shrink rather than increase. Or, as latest analyst report In Counterpoint’s words, “the market is witnessing a full-throttle uptrend across all segments”.
That’s why many new devices will currently be their cheapest for the next few years, even if they are more expensive than their 2024 counterparts.
In the near term:
- The price of RAM sticks and graphics cards used to build DIY PCs has doubled and continues to rise.
- Some non-volatile memories, such as SD cards, are also affected.
- Products that are replaced each year, such as laptops and smartphones, will become more expensive each year and may see increases mid-cycle.
- The launch of the new game console is being delayed and the prices of existing models are likely to increase.
What further complicates the issue, especially for laptops, is that AI also increases the minimum acceptable amount of RAM per product.
Microsoft required laptops to have at least 16GB (among other requirements) to be branded as Copilot+ PCs and to be able to access Windows AI features. This pretty much killed the sub-$1,000 basic 8GB laptop market, but Apple recently released one.
At this time, buying old stock that retailers already own or from trusted second-hand dealers may provide some relief. But these prices will also rise.
Is there any advantage?
If you’re just stocking up on RAM manufacturers or are looking forward to more powerful chatbots.
A raving proponent of generative AI might say that high costs are the price we pay to enter a new era of computing. The hardware is more expensive, but thanks to helpful AI, the time cost of each task is lower and the capabilities get better all the time, for free. But none of this is actually true. Investment in AI has resulted in breakthroughs in key areas such as disease diagnosis and language translation, but consumers are unlikely to fund this through a device tax; These are commercial developments.
Some argue that switching to artificial intelligence is the solution to the problem as well as the cause. A device with artificial intelligence can use system resources more efficiently, meaning a less capable machine can produce intelligent results where you previously needed brute force. So maybe your next computer may have less RAM and the money can go to Silicon Valley to train the next generation of models and everything will be fine. However, this is far from certain.
For example, AI-powered rendering is too energy-intensive and prone to bugs and quirks to replace Photoshop. You still need RAM for this. And chatbots aren’t reliable enough to keep you from opening 20 Chrome tabs for research. More RAM.
Nvidia caused a minor uproar when it recently introduced the latest version of its DLSS technology (deep learning supersampling), which is designed to make video games run faster relative to the machine’s horsepower by displaying them at lower resolutions and having AI clean up the data. DLSS 5 appeared to do this, but it also seemed to take creative license with faces, inventing details like wrinkles, makeup, and hair texture to make characters look like AI-generated photographs. Additionally, the demonstration required two RTX 5090 graphics cards; they are currently worth about $7,000 each.
We are a long way from AI being as important to a computer as RAM. But we cannot choose which one we will pay for.
Get news and reviews on tech, gadgets and games in our Tech newsletter every Friday. Sign up here.
Tim Biggs is a writer covering consumer technology, gadgets, and video games.Connect with: X or email.
