AI Boom to Push Inflation Higher Before It Cools It: What the RBA Warning Means for the Future of AI

For the past few years, the story about artificial intelligence has been mostly about what it can do. It writes code. It answers questions. It draws pictures. It sorts through mountains of data in seconds. The big question for most people has been whether it will take their job.

Now a new question is landing on the table, and it comes from the world of money rather than machines. In November 2026, a senior Reserve Bank of Australia official named Hunter made a point that is easy to miss but hard to un-hear: the AI boom will put near-term pressure on the Consumer Price Index.

That one sentence changes the frame. It says that AI is no longer just a tech story. It is a macro story. And the way it plays out over the next few years will shape how fast AI gets built, who pays for it, and what it ultimately costs to use.

The Warning, in Plain English

Start with the basics. The Consumer Price Index, or CPI, is the main tool a country uses to measure inflation. It tracks the price of a big basket of everyday things, food, rent, electricity, petrol, haircuts, doctor visits, and shows whether that basket is getting more expensive over time.

Central banks watch CPI closely. When prices climb too fast, they raise interest rates to cool things down. Higher rates make borrowing more expensive, which slows spending, building, and hiring.

What the RBA's Hunter signaled is that the AI build-out is now big enough to show up in that basket. The rush to build data centers, power them, and connect them is creating real demand for real things, electricity, land, construction crews, cables, cooling systems, and computer chips. When a lot of money chases a limited supply of those things, prices go up. And some of those price jumps land directly on households and businesses.

The point here is directional, not a precise forecast. It is not a claim that AI will send inflation to the moon forever. It is a warning that the first chapter of the AI era looks inflationary, not deflationary, and that policymakers, investors, and business leaders should plan for that.

Why Building AI Looks a Lot Like Building a Railway

There is a useful comparison here. When railways were built in the 1800s, the immediate effect was not cheaper travel overnight. The immediate effect was a boom in demand for steel, timber, coal, and labor. Prices rose before they fell. It took years for the productivity payoff to arrive.

AI infrastructure follows the same arc. Before a model can answer a question for a fraction of a cent, someone has to:

• Build enormous buildings full of servers
• Connect those buildings to power lines and water supplies
• Fill them with chips that are made in a small number of factories worldwide
• Hire electricians, engineers, and technicians who are already in short supply
• And keep it all running with massive amounts of electricity, day and night

None of that happens instantly. Supply is slow and lumpy. Demand is fast and hungry. That gap is exactly where inflation lives.

The Inflation Channels to Watch

1. Electricity Becomes the New Bottleneck

This is the most direct link to CPI, because power prices sit inside the household basket. Data centers are power-hungry. When a region adds a large number of them quickly, it can tighten the local grid. That can push up wholesale power prices, which eventually flow through to retail bills, sometimes with a lag.

The second-order effect matters even more. High demand for power pulls forward spending on new transmission lines, substations, and generation. Those projects are expensive, and the costs are usually recovered through customer bills. So even households that never touch an AI tool can end up sharing the cost of the infrastructure that runs it.

2. Construction and Skilled Trades

Data centers are construction projects. They compete for concrete, steel, and, most importantly, people. Electricians, welders, HVAC specialists, and site managers are not an unlimited resource. When a wave of large projects lands at once, wages in those trades get bid up, and timelines slip.

That pressure can spill into housing and other building work, because the same crews are being pulled in a different direction. This is one of the quieter ways an AI boom can end up affecting the cost of living.

3. Chips and Hardware

The supply chain for advanced computing hardware is narrow. A handful of factories produce most of the world's leading-edge chips, and memory has its own tight market. When demand surges, prices for components can spike, and lead times stretch.

Hardware costs hit businesses first, but they do not stay there. Cloud providers pass costs through. Software companies that rent computing power adjust their prices. Eventually, the cost of the machine shows up in the cost of the service.

4. Services and the Cost of Getting Things Done

There is also a demand-side channel. The AI build-out creates well-paid jobs, engineers, data specialists, sales teams, consultants. Those workers spend money. In places with tight housing and limited services, that extra spending can nudge rents and local prices higher.

Add in the fact that many businesses are currently paying twice: once for the old way of doing things, and once for the new AI tools layered on top. That cost has to go somewhere, and often it goes into prices.

The Flip Side: Cheaper, Eventually

Here is the part that often gets lost. The RBA's warning is about the near term. The word "near" is doing a lot of work in that sentence.

Over a longer horizon, AI is widely expected to push the other way. If a company can do the same work with fewer hours, or serve more customers with the same team, its cost per unit falls. Falling costs are disinflationary. Productivity is the closest thing economics has to a magic trick: the same money buys more output.

But productivity gains are slow to arrive and unevenly spread. A firm has to redesign its workflows, retrain staff, fix the mistakes the new tools make, and rebuild its processes around them. That takes years, not quarters. So the realistic shape of the AI era is a J-curve: costs and prices rise first, then fall later, and the fall only arrives if the build-out actually delivers the promised gains.

That "if" is the whole ballgame. Money spent on AI infrastructure is inflationary right now, no matter what. The disinflation only comes if the investment turns into real output.

What This Means for the Future of AI

If the AI boom is inflationary in the short run, then interest rates matter more to AI than most people assume. AI is one of the most capital-hungry industries ever built. Data centers cost enormous sums, and much of that money is borrowed.

Here is the chain reaction to watch:

• Higher CPI leads to higher rates. Central banks respond to inflation with tighter policy.
• Higher rates raise the cost of building. Borrowing gets more expensive, and projects that looked profitable at lower rates stop making sense.
• Some projects get delayed or cancelled. Not the biggest players, perhaps, but the second and third wave of builders.
• Supply catches up with demand. Slower building eases pressure on power, labor, and hardware.
• Prices stabilize, but so does capacity. The boom cools, and the winners are the ones who built efficiently.

This suggests the future of AI will not be a straight line of ever-cheaper intelligence. It will be a cycle. The cost of using AI may fall more slowly than the early hype promised, because the physical world, power, land, chips, people, sets a floor under how fast it can grow.

It also changes the competitive map. In a world where electricity is scarce and expensive, energy becomes a competitive advantage. Companies that can secure long-term power contracts, build near cheap generation, or design chips that sip rather than guzzle power will have a real edge. Efficiency stops being a nice-to-have and becomes a survival trait.

And it makes AI's benefits more uneven. If computing is expensive, the biggest users get the best prices. Smaller firms and public institutions may find the tools costlier than expected, which slows adoption exactly where the social payoff could be highest, schools, clinics, local government.

What Businesses Should Do Now

For leaders planning around AI, the practical takeaway is not "slow down." It is "plan for a bumpier cost curve." A few concrete moves:

• Model energy as a line item, not an afterthought. If your AI plans depend on cheap power, test what happens if power costs rise faster than you assumed.
• Stress-test your ROI at higher rates. Rerun your business case with borrowing costs above today's level. If the project only works in a low-rate world, it is fragile.
• Lock in what you can. Long-term power agreements, reserved compute capacity, and multi-year supplier contracts reduce the risk of price shocks.
• Invest in efficiency before scale. A model that uses less compute and less power is a hedge against exactly the pressure the RBA is warning about.
• Watch the CPI print, not just the model release. Inflation data is now a leading indicator for AI economics. It tells you what the next round of infrastructure will cost.
• Budget for the double-running period. Assume a stretch where you pay for old systems and new ones at the same time. That cash flow gap sinks more AI programs than bad technology does.

What It Means for Households and Society

For ordinary people, the message is simpler and less comfortable: the AI revolution is not free, and some of the bill arrives through the mailbox.

Electricity bills are the clearest channel. So is rent in cities where data center construction and tech hiring are concentrated. Neither of these means AI is bad. It means the costs and the benefits land in different places, on different timelines, for different people.

That creates a political question as much as an economic one. If power prices rise because of AI demand, regulators will face pressure over who pays for grid upgrades. If construction crews are pulled away from housing, local governments will be asked why. Central banks, meanwhile, will have to judge whether an AI-driven price bump is a one-off investment surge or something more persistent, a genuinely hard call, because the two look identical in the early data.

The best outcome is a policy mix that speeds up the supply side: faster grid connections, more power generation, more construction capacity, and more chip supply. Supply is the antidote to inflation. Building slower is not.

The Bigger Picture

The most important thing about the RBA's warning is what it signals about the stage AI has reached. When a technology starts showing up in a national inflation measure, it has stopped being a niche experiment. It has become part of the physical economy, with all the friction that implies.

That is, in a strange way, good news. It means the investment is real. Real buildings are going up. Real power is being consumed. Real people are being hired. The AI era has moved from slide decks into concrete and copper.

But it also means the easy assumptions are over. Cheap compute is not guaranteed. Cheap power is not guaranteed. Low interest rates are not guaranteed. The future of AI will be shaped as much by electricity grids and central banks as by model architectures.

For anyone building, buying, or betting on AI, the smart move is to treat the next few years as a period of expensive growth. Build for efficiency. Plan for higher costs. And keep an eye on the CPI, because it has quietly become one of the most useful signals for understanding where AI goes next.