India’s smartphone shipments fell 10% year-over-year in the April-June quarter, the steepest June-quarter drop in six years. According to TechCrunch AI, citing data from Counterpoint Research, the cause isn’t weak demand or a saturated market. It’s memory chips. The same RAM and storage components that AI data centers are hoovering up by the truckload.
This is the clearest evidence yet that the AI buildout has a consumer price tag, and regular people are paying it.
What’s actually happening
Samsung, SK Hynix and Micron have been shifting production toward high-bandwidth memory, the specialized chips that go into AI accelerators. HBM is far more profitable per wafer than the standard memory in phones and laptops. So capacity moved. Prices for everyday memory went up.
India got hit harder than anyone. Around 60% of its smartphone market sits below ₹20,000 (roughly $210), where every extra dollar of component cost shows up immediately in the sticker price. China, by comparison, saw shipments dip just 2%.
The numbers Counterpoint shared with TechCrunch AI:
- Sub-₹15,000 segment: shipments down 45% year-over-year
- Handset prices up between 4% and 68%, depending on model
- Samsung: only major brand with growth, up 2%
- Apple: down 3%, mostly supply constraints rather than demand
- Chinese brands: combined share at its lowest Q2 level since 2020
The two-tier split
What stands out is how cleanly the pain sorted itself by price point. Premium buyers barely flinched. Prachir Singh, a senior analyst at Counterpoint, told TechCrunch AI that financing makes expensive devices manageable, so high-end shoppers absorbed the increases. Budget buyers didn’t have that cushion. They’re delaying upgrades instead, stretching replacement cycles from about 3.5 years to four.
Kiranjeet Kaur, associate research director at IDC, framed it as a shift from volume-led growth to value growth. Fewer phones sold, more revenue per phone. That sounds fine on a spreadsheet and terrible if your entire business model is cheap handsets at scale.
OnePlus just proved the point. This week the brand said it would stop launching new products in Europe and North America while keeping its India business. Counterpoint data shows China accounted for 74% of OnePlus shipments in Q1, up from 59% a year earlier. India dropped to 19% from 30%. That’s a company retreating to wherever it can still turn a profit.
Tarun Pathak, Counterpoint’s VP of research, explained why the sub-brand model breaks first: “Sub-brands normally have overlaps and shared resources, and you need a minimum base to justify the cut-throat margins. Profitability is the key to deciding market operations.”
Why this matters beyond phones
India has 700 million-plus smartphone users and functions as a bellwether for price-sensitive markets globally. When entry-level demand collapses there, it’s a signal about what component scarcity does to the bottom half of any consumer hardware category. Laptops, tablets, smart TVs, IoT devices. They all use the same memory.
IDC expects the shortage and elevated prices to run until at least the end of 2027, though Kaur noted the pace of increases should moderate as higher prices become normal. She also flagged a compounding problem for India specifically: “For Indian consumers, it is a double whammy as the weaker currency makes imports costlier.”
What to do with this
If you’re building or selling hardware:
- Model your BOM assuming memory stays expensive through 2027, not for two more quarters
- Watch the entry tier first. It’s the canary for component squeeze
- Financing is now a product feature, not a payment detail. Kaur called it “central to affordability”
- If you run multiple sub-brands, do the volume math again. Shared-cost structures stop working fast at thin margins
If you’re deploying AI infrastructure, note the second-order effect: your compute costs are subsidized by scarcity somewhere else in the supply chain. That’s a story regulators in price-sensitive markets will eventually notice.
The memory crunch was predicted months ago. India is where the prediction stopped being theoretical. Full details are at the original source.