Phoney Problem

Smartphones are getting pricier thanks to AI-driven jump in dynamic RAM prices. But there’s a way out

This is a strange time in smartphone history. For almost 20 years, we’ve expected more bang for our buck year after year. Yes, phones got costlier, but their capabilities – processor, camera, battery, storage, etc – improved faster. Today’s cheapest smartphone is far superior in every way to 2007’s best. But 2026 is upending this “law” amid an AI squeeze.

As AI hyperscalers continue pouring money into data centres, demand for hardware – dynamic RAM especially – has soared. And so have prices. Per a Deloitte report, DRAM prices doubled in the first quarter of 2026, and are likely to double again before the year is out. Just three big firms make RAM for our devices, and last Dec one of them shut down its consumer-facing division to focus on supplying hyperscalers, for whom no price is too high. This has implications for ordinary consumers.

Early this year, smartphone sales were expected to drop 2% due to rising hardware prices. A recent estimate puts the drop at 6%, with five months remaining in 2026. Forget improvements, you’ll now pay a lot more for last year’s performance. Computers have also become expensive, with RAM’s share in overall cost doubling. The really bad news is that this crisis won’t end soon, even if the AI bubble pops. It’s because setting up new lines to make RAM for our devices needs many billions of dollars and 3-5 years.

This is a problem for India and other developing countries. Many students and workers might not be able to afford devices they need. Others will delay necessary upgrades. That means a hit to learning and productivity. But there are two lessons here that GOI shouldn’t miss. One, we’ve lost an opportunity, again. When demand for smart devices exploded during the pandemic, squeezing supply of less pricey chips used in cars and appliances, India wasn’t in the chip game at all. Six years later, when we’re getting ready to make those workhorse chips, AI has squeezed supply of high-end consumer chips, which aren’t on our manufacturing horizon yet. They should be, along with AI hardware. Secondly, chips from a few years ago are perfectly capable, but growing OS demands make them obsolete. It’s time to develop an open, less resource-intensive OS that can run on older hardware, to make affordable devices for developing countries. With its software capabilities, India is well placed to do so.