DDR5 Just Hit 5x Last Year's Price, AMD Is Eating Intel's Lunch, and Laptops Are Getting Absurd

Rough week to be shopping for a new build. If you've been sitting on a wishlist, brace yourself — memory prices are doing something genuinely ugly right now, and the storage side isn't far behind. But before you panic, there's also a surprisingly good wave of laptops landing, plus AMD quietly flexing on Intel where it matters. Let me untangle what's real and what's just noise.

The Memory Market Is On Fire, and Not the Good Kind

Here's the number that made me double-check my sources: DDR5 is now roughly five times more expensive than it was a year ago. That's not a typo. TechRadar's been tracking this and the curve is just... up, with no plateau in sight. And it's not just RAM — an exec in the storage space is predicting SSD pricing pain that could run all the way to 2030. Unverified industry rumor territory for the timeline, but the direction is hard to argue with.

What's driving it? The usual suspects: AI training demand hoovering up fab capacity, plus manufacturers pivoting production toward high-margin data-center silicon. From my perspective, this is the same story we saw in the 2017-18 memory crunch, except the demand side is now basically a vacuum cleaner.

The practical takeaway: if you were planning a build in the next few months, buy your memory and SSD now, not later. Waiting to "see if prices drop" has historically been a losing bet in a cycle like this. And if you're on DDR4 still, honestly, holding tight is looking smarter every week.

AMD Is Quietly Clawing Back Desktop Share

Meanwhile, over in CPUs — AMD hit 34.9% desktop CPU share in Q2 2026, per TweakTown's numbers. That's a real gain against Intel, and it's happening even while the overall desktop market shrank more than 20% year over year. The shrinking pie is Intel's bigger problem, because AMD's grabbing a bigger slice of a smaller plate.

The interesting subplot is the AI inference angle. Several analysts are now arguing AMD could outperform Nvidia in the inference market over the next few years — training is Nvidia's castle, but inference is where the volumes get huge and the cost-per-token wars happen. I've seen the price-to-performance math AMD's been pushing on MI350-class parts, and the argument isn't crazy. Nvidia's Q2 FY27 earnings drop on Aug 26, and there's chatter about SpaceX's 10 GW data center plans potentially being worth $120-230B in revenue to Nvidia — that's the kind of number that moves stock prices and GPU prices in weird ways.

Laptops: The Weirdest, Most Interesting Week

This is where it gets fun, because the laptop space is genuinely all over the map right now.

The long-battery Windows dream is here. HP's OmniBook 3, a Snapdragon X machine aimed straight at the MacBook Neo, is claiming up to 40 hours of battery life at a competitive price. If that number holds up in real use — and that's a big if, because these claims never survive contact with Chrome — it's the first Windows laptop I'd genuinely consider for all-day field work.

The workstation crowd got fed. Dell's Pro Precision 5 16s AMD pairs the Ryzen AI 9 HX PRO 475 (12 cores, 24 threads) with Radeon 890M graphics and a 60 TOPS NPU. StorageReview put it through its paces and it's a legit mobile workstation, though that Radeon iGPU is going to feel thin if you actually need CUDA.

Framework's still doing its thing. The Laptop 16 got updated components with the RTX 5070 12GB, one-piece keyboard, and a haptic touchpad. As a repairability-first machine, it's still the most interesting laptop nobody's buying in volume. The 5070 is a sensible step up from the previous GPU options, and the one-piece keyboard is a genuine quality-of-life upgrade over the old modular design. It's not cheap, but it's the only laptop where "upgrade later" is a real sentence.

ASUS is absolutely firing on all cylinders. The ProArt 16 pairs a Ryzen AI 9 H 465 with an RTX 5070, 64GB LPDDR5X, 2TB SSD, and a 2.8K 120Hz OLED — a direct MacBook Pro challenger for creators who need actual Windows GPU power. Then there's the ROG Zephyrus Duo (2026) going global with a Panther Lake CPU, up to an RTX 5090 laptop GPU, and dual 1,100-nit OLED displays. That's a ridiculous machine and an even more ridiculous price, but if you do 3D work on the road, there's nothing else like it.

And Lenovo's ThinkPad T1g Gen 9 brings Intel Panther Lake with vPro, 64GB of LPCAMM2 RAM, and up to the RTX 5070 12GB. That LPCAMM2 is the future of laptop memory — modular, upgradeable, way better than soldered RAM. Pricing starts at $4,109 with the RTX 5060, which is... a lot. But it's a flagship workstation, so nobody's buying it for the value.

If you're shopping in the mid-range, the ASUS V16 with Core 7 + RTX 5070 + 1TB SSD is on sale at a solid discount right now, and honestly that's where the sweet spot is this week.

The RTX 5090 Finally Met Its Match

Here's a fun one for the DIY crowd: Samsung's first 6K gaming monitor apparently pushes even the RTX 5090 beyond its comfort zone. The review headline says it plainly — the flagship isn't fast enough for 6K gaming at high refresh. That's the reality check the "just buy the biggest GPU" crowd needed: resolution inflation is now outrunning even the top silicon.

For 99% of us, this doesn't change the buying calculus — 1440p and 4K are still extremely happy at current GPU levels. But it's worth knowing that the "future-proof" argument for spending $2,000+ on a GPU is getting weaker by the month. The other side of that coin: used 5080/5090-class cards are starting to look a little less insane as the 6K wave approaches.

The Foundry Wars Are Getting Expensive

On the industry side, two things are shifting prices everywhere. Samsung foundry is charging up to 15% more for the chips it makes, with insiders saying TSMC's fabs are completely booked up and overflow orders are landing on Samsung's doorstep. And TSMC's Arizona fab crossed 4% of group sales, a small but symbolic number as US chip economics reshape. Revenue topped NT$40 billion in Q2, though profit slipped below NT$19 billion — the ramp cost is real.

All of this flows straight into the consumer pricing you're seeing: when foundry capacity tightens and wafer prices climb, it takes a quarter or two, but it lands on the shelf price of everything from GPUs to SSDs. Keep this in mind if you're budgeting a build — the quotes you're getting today are the cheapest you'll see for a while.

By the way, if you're weighing whether to build now or wait — and the numbers are making your head spin — there's a straightforward Decision Calculator that lays out the trade-offs in plain terms. No account, just the math.