Nine of the largest technology companies have committed to around $3 trillion of AI spending that does not appear on their balance sheets, five times the capital expenditure they reported over the past year. Morgan Stanley expects AI-related debt issuance to reach $570 billion globally in 2026, more than double last year.

AI may prove to be one of the most important technologies of this generation. However, borrowing to fund its development has reached enormous levels, and the revenue that would justify current valuations has not turned up yet.

Strain is already showing. Banks spent months trying to spread the risk of billions of dollars of loans they made to build data centers leased to Oracle in Texas and Wisconsin, which clogged their balance sheets and made the next projects harder to finance. While some argue that pressure on the wider US bond market has little to do with the AI borrowing binge, we might be testing the limits of how much debt AI can support.

Nobody knows whether this ends in steady growth, a slow deflation, or a crash. But having a plan for the latter is certainly worthwhile. Here are the ways an AI downturn would reach your business, and what to do about it while everything is calm.

How a bursting AI bubble could reach your business

Your suppliers come under pressure

Your business depends on software companies you have never thought about as borrowers. Private credit funds had lent over $500 billion to software-as-a-service companies by the end of 2025, 19 percent of all their direct loans, and the same BIS review recorded software stocks falling almost 30 percent between October 2025 and February 2026 as investors questioned which business models AI would disrupt.

Reuters reported in February that software companies were delaying debt deals as lenders grew more cautious. A supplier does not have to fail to cause you a problem. Prices go up. Support gets thinner. A product you built a process around gets discontinued or sold to somebody with different plans.

Credit becomes harder or more expensive

Data centers are being funded through bonds, bank loans and private credit at the same time, and the BIS has described how that structure connects the biggest technology companies to private credit funds, insurers and the banks lending against those vehicles. Banks are already looking further afield for ways to fund AI-related borrowing.

Losses in one part of a credit market make lenders more careful in every other part of it. A profitable company with a clean payment history may still find its facility renewal much harder work in 2027 than it was in 2024.

Your customers become more cautious

Follow the money one step back from your invoices. If a large share of your revenue comes from technology companies, or from people whose bonuses and share options depend on technology valuations, a fall in AI investment reaches you through their budgets before it reaches you anywhere else.

Agencies lose retainers and consultants lose projects. Recruiters, events companies and restaurants in the cities where that money concentrates feel it within a quarter. Your business may have no connection to AI at all while your best customers depend on it entirely.

You depend too much on one AI provider

Work out how much of your business would stop if one AI company changed its terms tomorrow. They raise the price, or retire the model you built on, and either decision is routine product management for them.

The more valuable a workflow becomes to you, the more expensive it is to move. Dependency builds because the tool works well. Nobody notices it happening.

5 ways to protect your business from an AI crash

Audit your critical suppliers

Write down the suppliers your business would struggle to operate without. Check four things for each one. Whether you can export your data, whether a workable alternative exists, how long a move would take, and whether you are paying years in advance for something you could pay for monthly. Then decide what you would do if one of them changed its prices, removed a product or disappeared. That decision is cheap to make today and expensive to make in a hurry.

Reduce dependence on a single AI provider

Keep the data and the instructions that make your AI systems valuable somewhere your company controls. Prompts, business logic, customer information, the descriptions of how your processes work. Where you can, build so a different model slots in without a rebuild. Use AI heavily where it helps you and keep the material that makes it useful in your possession.

Give yourself more financing headroom

Know the date every loan and credit facility comes up for renewal, and start the conversation months before it. Companies get better terms when they arrange finance before they urgently need it. Keep enough liquidity that a slow quarter stays a slow quarter. A business plan that only works while borrowing is cheap and easy is a bet on conditions you do not control.

Stress-test your customer base

Give your favourite AI tool your client list and last year's revenue by account, and ask it to group your customers by how exposed they are to technology spending. Then cut your revenue from the most exposed group by 20 percent and see what happens to your year. You do not need a financial model for this. You need to know whether too much of your revenue depends on the same part of the economy.

Keep investing in useful AI

Use AI everywhere it reduces your costs, improves what you sell or lets your people do more. A slowdown in investment could eventually make computing power, software and skilled people cheaper than they are today. Businesses in a strong position at that point get to buy what everybody else is selling. Building your systems now, on material you control, puts you on that side of it.

Protect your business from an AI bubble that could burst

Nobody can predict the future. Growth may continue, spending may slow gradually, or parts of the market may take much larger losses than anyone currently expects. You do not need to forecast any of those outcomes to act on them.

Check your suppliers, your customers, your financing and your technology dependencies over the next month, and your business is stronger in almost any economic environment that follows. The founders who look at all of this and do nothing are making a forecast too.

Get my free AI playbook for ambitious founders looking to scale.