Could the AI Boom Keep Mortgages Expensive for Years?
AI Data Centres Are Swallowing Money and Power — Now Interest Rates Could Feel It
The Price of the AI Revolution
Artificial intelligence may seem a long way removed from the monthly mortgage payment leaving a British household’s bank account. But the extraordinary amount of money now being poured into data centres, computer chips, power generation and AI infrastructure could become another force keeping global interest rates higher than borrowers became accustomed to before the pandemic.
The connection is no longer merely theoretical. The Bank of England has explicitly warned that large and sustained investment in AI infrastructure can place upward pressure on real interest rates, while its latest evidence from businesses shows the AI expansion colliding with constraints in electricity, grid connections, semiconductors, transformers and skilled labour.
The AI Revolution Is Becoming an Investment Supercycle
What began as a software revolution is rapidly becoming a physical construction boom. AI needs enormous amounts of computing power, and computing power requires chips, servers, cooling equipment, electricity, substations, transmission infrastructure, buildings and land.
The transformation is already visible in the United States. The Federal Reserve said business fixed investment increased at an annualised rate of 11% in the first quarter of 2026, with most of that strength apparently connected to infrastructure required to support AI services. Construction spending on new data centres has surged since 2022, while announced future projects have risen dramatically.
Federal Reserve data presented in July suggested capital expenditure associated with AI — including software, data centres, high technology and power — contributed about 1.36 percentage points to US GDP growth in the first quarter of 2026.
That is an extraordinary economic force for a technology whose mass commercial adoption is still relatively young.
Why Investment Booms Can Push Interest Rates Higher
Interest rates are ultimately the price of money. When companies, governments and households all want access to capital, the amount borrowers are prepared to pay for it matters.
The AI build-out is creating an enormous new source of demand. Technology companies need capital to finance computing infrastructure; utilities require money to expand electricity networks; semiconductor manufacturers need new fabrication capacity; and governments increasingly want the infrastructure required to remain competitive in AI.
The Bank of England's Financial Policy Committee has now identified precisely this possibility. It said large, sustained investment in AI infrastructure could place upward pressure on real interest rates where the supply of savings does not increase sufficiently to meet that demand.
That matters far beyond Silicon Valley.
Government bond yields influence borrowing costs throughout financial markets. In Britain, the rates lenders can offer on fixed mortgages are strongly influenced by expectations for future interest rates and wholesale market pricing. An AI-driven shift towards structurally higher global real interest rates could therefore eventually reach households refinancing an ordinary semi-detached house thousands of miles from the nearest hyperscale data centre.
The Bank of England Is Already Watching the Pressure
The issue is not that artificial intelligence has suddenly become the primary reason British mortgage rates are elevated. It has not.
Energy shocks, inflation, wage growth, government borrowing, monetary policy and global bond markets remain enormously important. Bank Rate stood at 3.75% following the July 2026 meeting, while the Bank has continued to confront inflation above its 2% target.
But AI is emerging as another variable inside an already complicated inflation equation.
The Bank's July Monetary Policy Report said strong demand for AI-related components was producing sector-specific price pressure that was feeding into UK import prices. Its network of business contacts separately reported shortages and capacity constraints involving electricity, grid connections, semiconductor chips, transformers and skilled workers.
That is the crucial distinction. Generative AI software can be copied almost infinitely at extremely low marginal cost. The physical infrastructure required to run it cannot.
Electricity Could Become One of AI’s Biggest Bottlenecks
The scale of the energy requirement helps explain the problem.
The International Energy Agency projects global electricity consumption from data centres could roughly double to around 945 terawatt-hours by 2030. Data-centre electricity use is projected to grow by around 15% annually between 2024 and 2030 in its base case, while electricity consumption from accelerated servers — the hardware most closely associated with AI — is expected to grow much faster.
Building a data centre can be relatively quick. Expanding an electricity system is not.
New transmission lines, generating capacity, grid connections and other infrastructure can require lengthy planning and construction. If computing demand grows faster than the physical economy's capacity to supply power and equipment, prices can rise before new capacity catches up.
The Bank of England is already hearing evidence of exactly that tension. Businesses say limited electricity capacity and slow grid connections are constraining expansion and contributing to higher costs across construction, energy infrastructure, IT hardware and operating expenses.
The AI boom therefore contains a strange contradiction: a technology designed to make businesses dramatically more efficient may initially require such a huge physical investment programme that parts of the economy become more expensive.
The Productivity Paradox
There is another side to the argument — and it could eventually overwhelm the inflationary one.
AI could make workers and businesses substantially more productive. Companies could produce more with the same number of employees, automate repetitive work, reduce administrative costs and improve decision-making.
The Bank of England's business contacts are already reporting productivity gains from AI in software development, finance, administration, customer service, professional services and content creation. Some businesses are reducing labour requirements for routine tasks.
Greater productivity can reduce firms' costs and expand the amount an economy can produce without generating inflation. That could ultimately allow central banks to maintain lower policy rates than otherwise.
But the timing matters enormously.
A Bank of England working paper published in August 2026 found that permanently stronger productivity growth can increase expected incomes, investment and consumption — potentially raising the economy's natural real interest rate. If demand accelerates before the extra productive capacity fully materialises, inflationary pressure can actually increase during the transition.
In other words, AI making the economy richer does not automatically mean interest rates fall.
It could initially do the opposite.
Why UK Mortgage Holders Could Feel an American AI Boom
This is where the story becomes relevant to British homeowners.
The epicentre of the infrastructure boom may be the United States, but global capital markets are interconnected. A persistent rise in US real interest rates or Treasury yields can influence government borrowing costs and financial conditions elsewhere.
Britain is already experiencing the consequences of changing market-rate expectations. The Bank of England said in July that increases in market interest rates had fed through into rates offered to households and businesses, with quoted two-year fixed mortgage rates considerably higher than they had been earlier in the year.
The newest official mortgage data underline the pressure. The effective interest rate on newly drawn UK mortgages increased from 4.35% in June to 4.45% in July 2026, while mortgage approvals for house purchases fell to 56,100.
The Bank has projected that a little over five million households could experience higher mortgage repayments by the end of 2028. Around 750,000 households paying mortgage rates below 3% were expected to roll off fixed deals during 2026, facing an average increase of approximately £170 a month.
AI is plainly not responsible for all of that. The recent Middle East energy shock and the resulting movement in market rates have been much more immediate factors.
The longer-term question is whether the AI investment cycle adds another reason for global borrowing costs to remain elevated after today's inflation shocks have disappeared.
AI Is Increasingly Being Built With Borrowed Money
There is another development worth watching: debt.
The first phase of the AI boom was heavily financed by the enormous cash flows and equity valuations of the world's biggest technology companies. But the infrastructure requirements are becoming so vast that external borrowing is playing an increasing role.
The Bank of England described the pace of AI infrastructure investment as historically unprecedented and said the use of credit markets accelerated rapidly during the first half of 2026. It warned that growing reliance on public debt markets, private credit, leveraged finance and structured products could make AI increasingly interconnected with the wider financial system.
Estimates cited by the Bank indicate that more than half of external data-centre financing requirements between 2026 and 2028 could ultimately be financed using debt. It also cited estimates suggesting more than $2 trillion of aggregate funding could be required for AI chips over the coming five years.
Those forecasts are uncertain. But they illustrate the scale of capital being demanded by the technology.
The Other Scenario: AI Ultimately Makes Money Cheaper
None of this means AI guarantees permanently expensive mortgages.
There is a powerful scenario pointing in exactly the opposite direction.
If artificial intelligence significantly increases productivity, businesses may be able to produce more without proportionally increasing wages, employment or other costs. Supply could expand, unit labour costs could fall and inflationary pressure could weaken.
The Bank of England has previously said stronger-than-expected productivity growth from AI could temporarily reduce firms' unit labour costs and permit a looser monetary-policy stance, all else being equal.
AI could also eventually improve government finances by increasing economic growth and tax receipts. Faster productivity growth would make existing public debt easier to support relative to the size of the economy, potentially reducing some pressure on government bond markets.
So two enormous forces are pulling in opposite directions.
The AI construction boom demands capital, electricity, workers and physical infrastructure and could push real interest rates upward.
The AI productivity boom could make the economy more efficient, expand supply and eventually suppress some inflationary pressures.
Which one dominates — and when — could become one of the defining economic questions of the next decade.
Cheap Mortgages May Not Return to the Old Normal
For homeowners, perhaps the biggest mistake would be assuming mortgage rates inevitably return to the extraordinarily low levels seen during the 2010s and the pandemic era.
The AI boom is only one reason that assumption could prove wrong. Ageing populations, higher government debt, defence spending, energy investment, deglobalisation and enormous infrastructure requirements could all increase competition for capital over the coming years.
Artificial intelligence adds potentially trillions more.
The irony is difficult to miss. AI may eventually deliver one of the biggest productivity improvements since the internet, reducing costs throughout the economy. But getting there requires building an immense new industrial system first — data centres, chips, grids, power stations and network infrastructure — and somebody has to finance it.
For millions of mortgage borrowers, that could mean the path back to genuinely cheap money takes considerably longer than expected.
