Chipflation: What to know about ‘AI’s hidden price tag’

Memory chips have long been foundational to technological innovation. More than a trillion are manufactured globally every year to run everything from smartphones and modern cars to medical devices and national defence systems.

There is currently no replacement for these silicon chips in mass-market computing. This makes them both vital for future progress and strategically important to businesses and economies.

But as the boom in artificial intelligence (AI) continues – the US’s five biggest AI companies expect to spend more than $650 billion this year, largely on data centres, and nearly double what they spent in 2025 – this demand is creating a risk of what is known as ‘chipflation’.

Addressing the issue could, however, accelerate the shift toward more efficient compute.

In simple terms, chipflation is when memory chips stop getting cheaper over time, and become more expensive and even harder to find.

That definition is from Morgan Stanley, which recently released research warning that the effects of the phenomenon are starting to spread from data centres to the wider economy.

Some of the core drivers behind this include:

AI data centre investment: AI systems and data centres need huge numbers of chips, leading big tech companies to buy more than the industry can easily supply.

Prioritization of high-margin chips: AI chips can carry significantly higher margins, leaving chip manufacturers to focus on these instead of chips for everyday products, including phones, laptops and cars.

Long manufacturing lead times: Building and operationalizing new semiconductor fabrication plants takes years and creates a supply bottleneck that cannot be quickly rectified.

Constrained supply chains: Geopolitical tensions, export restrictions and fragmented supply chains continue to limit availability of components and drive up production costs.

This is not the first time chips have been under pressure. Pandemic-era disruptions showed how exposed semiconductor supply chains were to sudden shocks, when shortages of chips led to the shutdown of automotive production lines and doubled the secondary market prices for consumer goods such as Sony’s PlayStation 5 console.

Apple CEO Tim Cook has described surging memory costs due to today’s crunch as “a hundred-year flood”. Cook says he has never seen anything like it in four decades in the industry, as AI data centres soak up vast quantities of memory chips such as Dynamic Random-Access Memory (DRAM).

His company has announced it will raise prices of MacBooks and iPads. As Bloomberg's Mark Gurman notes: "The bill for the AI era has officially come due for Apple Inc. customers".

Morgan Stanley estimates that memory prices have risen roughly six-fold over the past year, as manufacturers prioritize higher-margin data-centre chips over those used in everyday devices. “What began as an AI infrastructure bottleneck is now spreading into hardware margins, device affordability, cloud costs, inflation and policy,” it warns.

Brands including Sony, Microsoft and Lenovo have responded by lifting prices on PCs, consoles and other devices, while others absorb part of the shock through thinner margins.

It means that the next time you come to replace a laptop or phone, you might be hit by a surprising price hike. Research firms expect global PC and smartphone shipments to shrink sharply this year, as these higher prices deter buyers and lengthen replacement cycles.

In response, chip manufacturers are adding capacity, but the cost and complexity of building new fabrication plants means meaningful relief is likely to take years. Analysts see supply constraints persisting well into the second half of the decade, with AI servers continuing to soak up memory and keep prices high.

All of this means chips are becoming one of the defining strategic assets of the AI economy, according to BlackRock.

As governments and technology companies continue to pour hundreds of billions of dollars into AI infrastructure, they are no longer just competing with one another, they are competing with the rest of the economy for the same finite resources.

This could present a test to the AI revolution: can the world build the infrastructure that will enable increasingly intelligent models without passing the costs – direct and otherwise – onto the rest of the economy?

AI is driving new waves of innovation and growth, according to the World Economic Forum report, Advancing Responsible AI Innovation: A Playbook, produced in collaboration with Accenture. But realizing its full potential lies in building and managing AI systems to maximize benefits while minimizing risks to people, society and the environment.

Chipflation could put pressure on this goal, centralizing the most powerful AI capabilities among a handful of companies that can afford the most expensive chips.

But it could also create opportunities – by driving innovation in smaller models for specific tasks that deliver high performance using a fraction of the compute. Apple and Microsoft have both developed small language models in recent months, with OpenAI's Sam Altman announcing "we're at the end of the era [of] giant models". This shift would bring cost and energy-efficiency benefits and help democratize access to language models.