Artificial intelligence was supposed to make computers smarter. It may first make them more expensive.
Mumbai: The latest pressure point is memory. AI data centres are consuming enormous quantities of HBM, server DRAM and enterprise storage, while memory manufacturers are increasingly allocating production capacity toward those higher-margin products. The unintended consequence is beginning to surface elsewhere: PCs, smartphones and even gaming hardware are competing for what remains.
Nobody has introduced an official “AI tax” on consumer electronics.
The semiconductor market appears perfectly capable of inventing one by itself.
As of late August 2026, the average spot price for a 16GB DDR5-4800/5600 desktop module had reached about $230, while a comparable 16GB DDR4-3200 module averaged roughly $162.20. By September 2, PC DRAM contract-price expectations had been revised upward again, despite comparatively weak consumer demand.
Your Laptop Is Competing With A Data Centre
The underlying problem is not simply that AI servers use memory. They use extraordinary amounts of it.
Modern AI infrastructure depends heavily on high-bandwidth memory (HBM) around accelerators, but the appetite extends to conventional server DRAM, SSDs and increasingly memory-heavy CPU systems used for agentic AI.
Micron says memory content per server has doubled in three years, while rapidly increasing AI context lengths are placing further pressure on capacity and bandwidth.
Memory companies naturally follow the money. Samsung says its memory business is concentrating on high-value products including HBM4, DDR5 and SOCAMM2, with AI infrastructure investment expected to keep server demand robust through the second half of 2026. It also expects supply constraints to continue despite efforts to increase production.
Which leaves the ordinary laptop asking a rather unfortunate question: Do I really need 32GB?
Apparently, the data centre has already answered on your behalf.
Consumers Could Pay In Price Or Specifications
The impact does not necessarily mean every ₹70,000 laptop suddenly becomes ₹1 lakh. Memory represents only part of a device’s total component cost.
Manufacturers have other options.
They can absorb some of the increase and sacrifice margins. They can raise prices. Or they can quietly ship machines with less memory than originally planned.
The last option is already becoming relevant. The latest industry analysis says PC and smartphone buyers are moving toward lower-capacity memory configurations to manage costs as suppliers favour server applications.
Conventional DRAM contract prices were forecast to rise another 13–18% quarter-on-quarter in Q3 2026. Smartphone makers are also expected to pass higher memory costs into retail pricing, although weakening consumer demand is limiting how far those increases can go.
That produces an awkward possibility: tomorrow’s “AI PC” may arrive with increasingly sophisticated AI features while compromises are made elsewhere to keep the sticker price socially acceptable.
Technology loves progress. Preferably with an asterisk.
Gaming Hardware Is Not Entirely Safe Either
Graphics memory is experiencing its own squeeze.
Demand for GDDR6 and GDDR7 has not been uniformly strong, but manufacturers have been reallocating capacity toward other profitable memory products. That has kept graphics-memory supply constrained enough for prices to rise alongside the wider DRAM market.
For gamers, that matters because GPUs already carry substantial silicon and cooling costs. Higher graphics-memory pricing gives board makers one more expense to absorb—or introduce to the customer with the enthusiasm traditionally reserved for “premium editions.”
The Good News Is Manufacturing Is Finally Responding
Shortages create pain, but they also create investment.
Micron recently increased its planned US manufacturing and technology investment to more than $250 billion through 2035, citing surging memory demand in the AI era. The company ultimately wants around 40% of its DRAM production located in the United States.
Samsung and other manufacturers are also accelerating advanced-node transitions and expanding capacity.
That investment should eventually increase supply, improve memory technology and create faster, more efficient products for both data centres and consumers.
The word doing heroic amounts of work there is eventually.
There Is One Natural Limit: Consumers Can Say No
The industry cannot raise consumer-electronics prices indefinitely.
Analysts already describe PC and smartphone customers as approaching an affordability limit. Notebook retail prices are rising, smartphone production plans are becoming more conservative, and weaker demand is starting to temper further memory-price increases.
That may ultimately provide the market’s most effective correction.
If a new laptop costs too much, people keep the old one.
If 32GB becomes prohibitively expensive, manufacturers settle for 16GB.
And if consumers stop upgrading, suppliers eventually rediscover how fond they were of the consumer market all along.
The AI boom is unquestionably accelerating semiconductor investment and producing extraordinary advances in computing. The downside is that resources are finite, fabrication plants take years to build, and the highest bidder currently tends to live inside a data centre.
So your next PC may still become smarter.
Just don’t be surprised if intelligence arrives with less RAM, a higher price—or both.
Read More: DeepSeek’s 160,000-Chip Bold Bet
