The Hardware Crisis: Why Your Next Laptop Will Cost 20% More

While everyone talks about how cheap it is to use ChatGPT ($20 a month, what a bargain), there’s a bill nobody’s looking at: hardware.

The numbers are brutal — and getting worse (updated June 2026):

RAM: DRAM prices surged another ~95% in Q1 2026 alone; cumulative increases since mid-2025 are historic
SSDs: +100-200% vs. end of 2024; popular Gen 4 NVMe drives from Samsung and WD have doubled or tripled in price
HDDs: +46% over recent months
Laptops: +15-20% in 2026 (Dell, Lenovo, HP have all confirmed increases)
Smartphones: average selling price up ~14% to all-time highs (Gartner, Feb 2026)

The cause? AI data centers are devouring the world’s entire memory chip production. And there aren’t enough for everyone.

What’s happening

Between 2025 and 2029, $3 trillion will be invested in building AI data centers. Morgan Stanley estimates half of that investment will go to hardware.

To put it in perspective: each cutting-edge AI server needs obscene amounts of memory. A rack of NVIDIA H100 GPUs can carry more RAM than your entire office combined.

And memory manufacturers (Samsung, SK Hynix, Micron) can’t ramp up production fast enough. Chip factories take years to build. Demand has exploded in months.

Result: shortage. And when there’s shortage, prices go up.

How it affects you

If you’re buying a laptop

That laptop that cost $800 in 2024 now costs $920-960 or more. Same model, same specs. It’s just that the RAM and SSD inside cost the manufacturer much more — memory now accounts for roughly 35% of a laptop’s manufacturing cost, up from 16-20% historically.

High-end models (with more RAM and storage) are hit hardest. A work laptop with 32GB of RAM and 1TB SSD may have gone up $200-300. Dell, Lenovo, and HP have all announced price increases of 15-20% for 2026.

If you’re buying a phone

Mid-range and high-end smartphones are already rising. The global average selling price has climbed ~14% to all-time highs (Gartner, Feb 2026), and mobile DRAM prices are up close to 70% since early 2025.

Manufacturers have two options: raise prices or reduce specs (less RAM, less storage). You’re seeing both. Budget Android makers are hit hardest; Apple and Samsung have long-term supply contracts that give them partial insulation.

If you work with servers

This is where it really hurts. Server hardware is in critical shortage. Lead times have lengthened. Prices have risen. And cloud providers are absorbing all available stock.

If your company needs to expand on-premise infrastructure, prepare for long waits and revised budgets.

If you build your own PC

Gaming enthusiasts and DIY builders are feeling it especially. A 32GB DDR5 RAM kit that cost under $100 in mid-2025 now costs $375 at minimum — popular kits from Corsair or Crucial are well over $400. NVMe SSDs have risen in parallel: the best Gen 4 drives have doubled or tripled in price.

If you have components in your cart, don’t wait. This is the “AI tax” in full effect.

The irony

Here’s what’s ironic: AI is sold as technology that “democratizes” access to advanced capabilities. Anyone can use GPT-4 for $20 a month. Amazing.

But that democratization has a hidden cost distributed among everyone. The data centers needed for you to chat with an AI are consuming resources that make all hardware more expensive.

It’s an invisible cross-subsidy. You pay $20 a month for AI, but you pay $150 more for your next laptop. And $50 more for your next phone. And so on.

When will it get better?

Not soon. As of mid-2026, the outlook keeps getting pushed further out.

The new semiconductor factories being built (TSMC in Arizona with a $165B investment, Intel in Ohio, Samsung in Texas) won’t be fully operational until 2027-2028. TrendForce sees no real NAND relief before end of 2027. SK Hynix is talking about a structural shortage lasting through 2030.

Until then, the shortage continues. Prices stay high. And competition between AI data centers and the rest of the market continues.

Some signals to watch:

Positive: new factories coming online, more efficient memory technologies, slowdown in data center construction.

Negative: more AI investment (NVIDIA put another $2B into CoreWeave in January 2026), AI will consume 20% of all DRAM production in 2026 — and that share is growing.

My bet stands: prices will stay high through 2026 and 2027. Possible stabilization in 2028 if new manufacturing capacity arrives on schedule. But don’t expect to return to 2024 prices.

What you can do

If you need to buy hardware soon

Buy now. Analysts don’t expect meaningful price relief before late 2027 at the earliest. If you can afford the expense now, do it. And if you have DDR5 RAM or an NVMe SSD in your cart, don’t leave it there.

If you can wait

Wait until 2028. I know it sounds like a long time, but if your current equipment works, there’s no rush. Prices should stabilize when new manufacturing capacity (TSMC Arizona, Samsung Texas) comes online.

If you manage IT budgets

Revise your projections. Hardware will cost more than you budgeted. Better to adjust now than get surprised later.

If you’re considering cloud vs on-premise

Cloud becomes relatively more attractive when hardware prices rise. Hyperscalers have long-term contracts with manufacturers that give them better prices. You don’t have that advantage.

But watch out: cloud prices are rising too. You don’t escape entirely.

The elephant in the room

Nobody talks about this at AI conferences. Nobody mentions it in product launches. Nobody includes it in ROI analyses.

But it’s real. AI has an infrastructure cost that’s being externalized to the rest of the economy. We all pay a little more for our hardware so AI data centers can be built.

I’m not saying it’s bad. AI has real benefits. But it would be nice if the complete cost were on the table, not hidden in the price increase of your next laptop.