The Artificial Intelligence revolution is producing an unexpected economic side effect: Chipflation, a term coined by Morgan Stanley.
For decades, the technology industry operated on a powerful assumption—computing would become cheaper, faster and more accessible. Moore’s Law, manufacturing efficiencies and economies of scale steadily reduced the cost of computing power. But the economics of artificial intelligence are beginning to challenge that model.AI is creating an extraordinary demand for advanced processors, high-bandwidth memory (HBM), networking equipment and data-centre infrastructure. Semiconductor manufacturers are struggling to expand capacity at the same speed. The result is a new form of inflation—not necessarily visible immediately in consumer price indices, but increasingly visible in the cost of computing infrastructure. Laptop prices have increased tremendously in India; overnight the base Mac Mini (Apple) has increased from Rs 59,900/- to Rs 94,900/- while the high-end MacBook Pro saw its price tag increase by one lakh rupees. Apple’s official explanation points to a severe and unprecedented global memory shortage. Silicon Valley says that the insatiable hunger for AI Data Centres is eating up global supplies of DRAM and NAND flash storage, pushing component costs four times. Unfortunately, this is a fact. The entire AI setup relies on massive server farms running complex LLMs, which require an enormous amount of high-bandwidth memory. This has pushed the prices of memory chips as suppliers like Samsung and Micron have shifted production priorities away from consumer electronics to drive margins from enterprise AI clients. TrendForce estimates that conventional DRAM contract prices rose sharply in 2026 as suppliers redirected capacity towards HBM and server applications. It forecasts another 58–63% quarter-on-quarter increase in conventional DRAM contract prices in the second quarter of 2026. For the third quarter, it expects a further 13–18% increase. NAND Flash prices are also under pressure. This is important because HBM is integral to modern AI accelerators.
The other side of chipflation is the extraordinary demand for AI accelerators. Hyperscalers, cloud providers, technology companies and Governments are building AI computing capacity at unprecedented scale. The competition is not merely for today’s GPUs; companies are trying to secure access to successive generations of increasingly powerful accelerators. The strategic significance of GPUs has consequently changed. This has created a fundamental economic challenge for data-centre operators: the cost of computing capacity is becoming an increasingly important determinant of the economics of AI. A company that cannot secure sufficient GPUs and memory may not merely experience higher procurement costs. It may lose time-to-market in deploying AI products. That creates a willingness to pay a premium for supply. And hence we have Chipflation.
As of May 2026, the Government of India said 12 semiconductor manufacturing units and 24 semiconductor design companies had been approved for fiscal support under the Semicon India Programme. The IT Hardware PLI scheme had generated cumulative production of ₹23,993 crore by May 2026. India has also moved from announcements towards actual production. A semiconductor facility in Sanand, Gujarat, began commercial production of intelligent power modules in March 2026, with eventual capacity expected to reach 6.33 million units a day. These are important milestones. But Chipflation tells us that capacity alone is not enough. India needs a complete ecosystem. India’s semiconductor mission must evolve from a programme for making chips into a broader strategy for securing the economics of the digital economy. Because the next industrial revolution may be powered by artificial intelligence—but its competitiveness will ultimately depend on something much more physical: who has the chips, who can afford them, and who can make them.
What do you think?
Indian Chamber of Commerce