The AI boom has a hidden tax, and your next PC or phone is paying it
Samsung's chip profits are soaring while it warns the shortage will last through 2028. Buyers are covering the AI buildout through pricier devices, and brands should say so.
Photo: Qurren / Wikimedia Commons (CC BY-SA 3.0)
Who pays for a data center the size of a small city? Shareholders, yes. Cloud customers, sure. But if you plan to buy a laptop or a phone in the next two years, you should assume you're on the invoice too.
Samsung reported on Thursday that its July-to-September operating profit hit 107.4 trillion won (€72.2 billion), up from 12.2 trillion won a year earlier, according to Reuters via the Irish Times. Nearly all of it, 106.5 trillion won, came from chips. The same company said its high-bandwidth memory supply is sold out for next year and that the market will stay tight through 2028. Its own mobile division, meanwhile, fell to a loss of 800 billion won because memory got more expensive.
The shortage is a transfer of wealth, not an accident
Read that last sentence again. One of the world's biggest memory makers is earning a fortune on AI, and the division that sells phones to you is bleeding because of it. Samsung is effectively both the beneficiary and the victim of its own boom.
Samsung says AI-driven demand will "far exceed the industry supply in comparison with 2026." SK Hynix went further, saying its DRAM and HBM chips are sold out through 2029 and that the shortage will persist to 2030. These are two companies with every incentive to sound confident, so we'd treat the exact end dates as sales talk. The direction, though, is hard to argue with: capacity looks to be pointed at data centers first, and everything else gets what's left.
What's left costs more. A ZDNet France snippet attributes a 350% jump in RAM costs to Microsoft and OpenAI infrastructure soaking up global chip stocks. We can't verify that figure from the text we have, so take it as a headline number, not a benchmark. But the pattern matches what Samsung itself describes.
Your device is the shock absorber
Here is the uncomfortable economics, at least as we read it. Big AI buyers locked in supply early, and Samsung says it now has long-term contracts with 10 major customers, is negotiating with 10 more, and wants such deals to cover 60% to 70% of its capacity by 2030. It is also extending them to mobile and automotive customers. That helps those buyers eventually, but we suspect that in the meantime device makers, who work on thin margins and short product cycles, are the ones absorbing the volatility, and often passing the bill on to you.
Apple's new touchscreen MacBook Pro is a vivid example. 01net reports that the top 16-inch configuration, announced October 28, costs €16,264, and that €7,700 of that goes solely to moving from 2 TB to 16 TB of storage. Almost nobody will buy that machine, and it would be unfair to treat it as the typical price. But it shows how steeply memory is priced when it's scarce. TechCrunch noted that this "RAMaggedon" had already forced Apple to raise prices on its MacBooks, Macs and iPads.
There's one reason for mild optimism, and we'd flag it as such. Reuters notes that the pace of memory price gains is slowing, partly because higher chip prices are weighing on demand from phone and PC makers. In plain terms: buyers are starting to say no. That's the market working, but it's a rough way to find the ceiling, because it works by making devices unaffordable first.
Brands should stop blaming "market conditions"
Our argument is simple. When a laptop gets pricier, or ships with less RAM, or a budget model quietly vanishes, the explanation should be on the product page. Not buried in "component costs," but stated: memory is scarce because AI data centers are buying it all, and we're passing part of that on.
That honesty would cost vendors little and give buyers something useful: a reason to trust that the increase isn't plain margin padding. It would also put the real debate where it belongs. Is the AI buildout worth subsidizing through everyone's hardware budget? Reasonable people can say yes. But they should get to say it knowingly.
Our practical advice, as opinion and not a forecast: if your current phone or PC works, keep it; the worst pricing is likely still ahead if Samsung's outlook holds. If you must buy, favor the configuration you'll actually need over the cheapest one, since memory and storage upgrades are exactly where the premium seems to be landing. And if a vendor raises prices, ask them why. The honest answer is right there in Samsung's earnings.
No topics yet: start the first one.
More stories
Apple's touchscreen Mac wasn't a change of heart, just a change of timing
Apple mocked touchscreen laptops for years, then built a $2,899 one. The real question isn't hardware: it's whether macOS was finally designed for fingers.
Soldered RAM is a bad bet, and vendors should stop pushing it
Soldering memory turns a cheap later upgrade into a guess about your future needs, and with prices this high, that guess is the buyer's problem, not the vendor's.
Nvidia's tensor cap looks like a pricing decision, and Linux just exposed it
Open-source Linux modules double the RTX 4090's FP16 tensor throughput without adding hardware, which suggests the limit was about segmentation, not silicon.
DuckDB 2.0 Makes the Case for Keeping Analytics Close to the Data
DuckDB 2.0 does not reject the cloud; it argues that local execution, embedded analytics and selective remote scale belong in the same data architecture.



