AI WAS SUPPOSED TO MAKE EVERYTHING CHEAPER. WHAT IF IT MAKES THINGS MORE EXPENSIVE?

 

BIZ MASALA SPECIAL REPORT

AI WAS SUPPOSED TO MAKE EVERYTHING CHEAPER.
WHAT IF IT MAKES THINGS MORE EXPENSIVE?

The AI boom is creating a new economic tension: enormous productivity potential on one side, and intense demand for chips, electricity, equipment and skilled capacity on the other.

By Murthy ASR  |  Biz Masala  |  7 October 2026

THE QUESTION NOBODY EXPECTED FROM THE AI BOOM

Artificial intelligence was sold to the economy as a productivity machine. More output from the same workforce. Faster software. Better logistics. More efficient factories. Lower costs. Eventually, perhaps, cheaper products.

But a new warning is emerging from an unlikely place: central banking. San Francisco Federal Reserve President Mary Daly has said that rising AI demand for computer chips could create inflationary pressure beyond the technology sector, particularly if chip supply remains constrained. She pointed to companies seeking forward contracts for memory chips and redesigning products to use fewer chips. cite Reuters, 6 Oct 2026 / Axios, 6 Oct 2026

That does not mean AI has suddenly become an inflation machine. It means the transition to an AI-heavy economy has an infrastructure bill—and somebody has to pay it.

EDITORIAL ILLUSTRATION — chip demand is no longer only a technology-sector story.

THE AI BOOM HAS A PHYSICAL SIDE

The public conversation around AI often happens in software: models, applications, agents and productivity. The economic build-out happens in the physical world. It needs advanced processors, memory, networking equipment, cooling systems, buildings, transformers, power generation and thousands of specialised workers.

That is where bottlenecks can become inflationary.

The San Francisco Fed warning matters because it describes a mechanism familiar from the post-pandemic chip shortage. If one industry absorbs a large share of constrained components, other industries can be forced to compete for the same supply. Autos, appliances and electronics are obvious examples.

The difference this time is that AI investment is still expanding rapidly. Marvell, for example, recently raised its fiscal 2028 revenue outlook to about $20 billion, citing strong AI data-centre demand. cite Reuters, 6 Oct 2026

AND THEN THERE IS ELECTRICITY

EDITORIAL ILLUSTRATION — AI infrastructure ultimately runs on electricity, not algorithms alone.

A data centre is an industrial customer with a very large appetite for electricity. As AI workloads become more computationally intensive, power availability becomes an economic constraint.

Research from the Federal Reserve Bank of Dallas estimates that even a modest US data-centre build-out could raise annual PCE inflation in 2030 by roughly 0.04 to 0.13 percentage points under plausible assumptions. The effect could be larger if renewable generation expands more slowly than expected. cite Dallas Fed, 2026

That is a small number at the headline level—but central banks care about small persistent pressures, especially when they arrive alongside oil, tariffs or other supply shocks.

INDIA: THE NEXT BATTLEFIELD IS THE POWER GRID

India has a particularly interesting version of this story because it wants both sides of the AI equation: rapid adoption and rapid infrastructure creation.

The World Bank now expects India to grow 7.1% in FY27, up from its earlier 6.6% forecast, while explicitly highlighting AI as a development opportunity. Private AI investment in India rose from $1.2 billion in 2024 to $4.1 billion in 2025, according to the World Bank. cite World Bank, 6 Oct 2026

But the physical footprint is also expanding. Government projections indicate that AI data centres could add 26.3 GW of load to India's power system by FY2031–32—almost double the earlier 13.56 GW estimate. cite Ministry of Power reporting, July 2026

Source: Government projection reported by Indian Express and Business Standard, July 2026. Editorial chart by Biz Masala.

THE AI INFLATION PIPELINE

The key point: the inflation risk comes from bottlenecks and timing—not from AI itself.

WHAT COULD GET MORE EXPENSIVE?

Editorial framework: potential transmission channels, not a forecast of price increases.

Semiconductors sit closest to the first pressure point. But the chain can travel through construction, electrical equipment, transformers, cooling systems, data-centre land, specialised labour and electricity.

This is why the debate is larger than whether an AI chip becomes more expensive. The real question is whether AI investment starts competing with other parts of the economy for scarce resources.

THE CENTRAL-BANK PROBLEM

Monetary policy is designed mainly to influence demand. It is much less powerful against a shortage of chips or transformers.

That creates a difficult policy distinction. If inflation is caused by a temporary supply shock, a central bank may choose to look through it. If the shock persists, spreads to wages and prices, and changes inflation expectations, policymakers become more uncomfortable.

That is precisely the issue Daly raised: AI-related demand, higher oil prices and tariffs could either fade separately—or reinforce one another. If they persist together, inflation can become harder to dismiss as temporary. cite Reuters, 6 Oct 2026

India is facing its own version of the dilemma. Markets were already watching the RBI closely on October 7 amid expectations of a possible 25-basis-point rate increase, with crude above $100 a barrel adding to the inflation conversation. cite Reuters, 7 Oct 2026

BUT HERE IS THE OTHER HALF OF THE STORY

There is a powerful counterargument—and it should not be ignored.

AI can eventually lower the cost of producing almost everything. Software development can become faster. Customer service can become more automated. Supply chains can become more predictive. Energy systems can be optimised. Factories can use AI to reduce downtime and waste.

Federal Reserve officials have made this point too: productivity gains could ultimately raise living standards and allow stronger economic growth without equivalent inflation. The problem is timing. The infrastructure bill arrives now; productivity gains may take longer to diffuse through the economy.

In other words, the AI economy may have an uncomfortable transition period: **higher capital spending and input pressure first, broader productivity gains later.**

INDIA'S OPPORTUNITY: DON'T JUST IMPORT THE BOOM

For India, this is where industrial policy matters.

If India simply imports AI servers, advanced chips, power equipment and cloud capacity, the AI boom can create a large external technology bill. If India builds more of the surrounding value chain—electronics, components, power infrastructure, cooling, networking, software, data-centre services and eventually semiconductor capacity—the same boom can become an industrial opportunity.

The World Bank's latest India update argues that India is well positioned for AI because of its technical workforce, globally integrated IT sector and digital public infrastructure. That is the demand-side advantage. The harder challenge is building enough physical infrastructure to support it. cite World Bank, 6 Oct 2026

THE BIZ MASALA VERDICT

AI is not the new oil shock. But AI can amplify an oil shock, a chip shortage or a power constraint.

That is the real economic story.

The first phase of AI was about models. The second is about infrastructure. And infrastructure has a price.

The countries that win the AI race will not necessarily be the ones with the biggest models. They may be the ones that can supply chips, electricity, networks, cooling, talent and capital fast enough to keep the cost of intelligence from rising faster than its productivity.

For India, the opportunity is even bigger: become not merely a large consumer of AI, but a competitive producer of the infrastructure and services around it.

The ultimate test is simple: **Can the productivity dividend arrive before the infrastructure bottleneck becomes the inflation story?**

QUICK TAKEAWAYS

·       AI demand is already influencing chip procurement and capacity decisions.

·       Data-centre expansion can raise electricity demand and, in some markets, power prices.

·       India's projected AI data-centre load has risen sharply in government estimates.

·       The near-term inflation risk is about bottlenecks; the long-term deflationary potential is about productivity.

·       Central banks have limited tools against physical supply constraints.

·       India's strategic opportunity is to build the value chain around AI, not merely consume imported AI infrastructure.

SOURCES & NOTES

·       Reuters, 6 October 2026 — Fed's Daly: need for more hikes hinges on what happens with shocks.

·       Axios, 6 October 2026 — Fed's Daly on AI chip demand, tariffs, energy and inflation.

·       Federal Reserve Bank of Dallas / Energy Economics Research Network, 2026 — Data-center boom and electricity-price inflation analysis.

·       World Bank, India Development Update, 6 October 2026 — FY27 growth forecast, AI investment and India's AI opportunity.

·       Indian Express / Business Standard, July 2026 — Government projection of 26.3 GW additional AI data-centre load by FY2031–32.

·       Reuters, 7 October 2026 — Indian markets and RBI rate-decision context.

© 2026 Biz Masala | Murthy ASR
Original editorial illustrations and charts prepared for this edition.