Why It’s Costing More to Invest in the Gulf

The Gulf is still spending billions on AI. The question is no longer whether the money is coming. It is whether investors are still willing to accept the same risk to earn it.

On October 7, Gulf stock markets gave an early warning. Saudi Arabia’s TASI fell 0.5% to 10,542 points, Abu Dhabi’s ADX lost 0.3%, Dubai’s DFM fell 0.2%, and Qatar’s index declined 0.5%. None of these moves is large enough to suggest that investors are suddenly leaving the Gulf, and that is not what is happening. The important point is that markets are showing how quickly regional risk can move into asset prices, at a time when the Gulf is simultaneously committing enormous amounts of capital to AI infrastructure.

For AI investors, this matters because data centres are not ordinary technology investments. They are enormous physical assets that can require billions of dollars before generating meaningful returns, and their economics depend on electricity, cooling, fibre connections, physical security and uninterrupted operation. If the cost of protecting those assets rises, the investment case changes with it. Capital has not disappeared from the Gulf. The price of that capital, and the return required to justify the risk, is what is beginning to change.

The Stock Market Is Only the Beginning

A falling stock index does not mean investors have suddenly decided that Gulf AI projects are bad investments. Public markets price hundreds of things at once, including oil, interest rates, corporate earnings, liquidity and regional sentiment. But markets are useful because they provide the first visible price of uncertainty. Saudi Arabia is a good example: TASI closed at 10,542 on October 7, much closer to the bottom of its roughly 10,194–11,782 annual range than to the top, while September had already been difficult for the Saudi market, with TASI falling 6.2% during the month. That does not prove that investors are repricing AI data centres specifically, but it does tell us that the broader market is demanding more compensation for risk.

Private infrastructure investors do not operate in a completely separate universe. If public markets become more nervous, private markets eventually have to answer the same question: what return is enough to justify putting billions of dollars into a physical asset in the Gulf? This is where the story becomes more important than the daily movement of an index. A 0.5% decline in a stock market can disappear within days, but a permanent increase in insurance costs, financing costs or required returns can remain embedded in an infrastructure project for decades.

The Cost of Protecting an AI Data Centre Is Rising

Insurance is one of the clearest places where this risk becomes measurable. Carnegie has estimated that insurance for $100 million of military-risk exposure in the region rose to around $5 million, compared with roughly $250,000 previously. For a small project, a higher premium may be manageable. For a multi-billion-dollar AI campus, it becomes part of the economics of the asset itself. If protecting a facility becomes significantly more expensive, the developer cannot simply make the risk disappear. It has to absorb the cost, pass some of it to customers, invest more heavily in security and redundancy, or accept a lower return.

This is why insurance has become an investment issue rather than a technical detail. Large data centres already require enormous amounts of capital before they generate meaningful operating revenue. Add higher insurance premiums, additional security and redundant infrastructure, and the amount of capital required to produce the same level of usable computing capacity increases. The equity investor is particularly exposed because equity absorbs much of the uncertainty around construction, utilisation and the long-term value of the asset. A project can remain strategically important while becoming financially less attractive.

The Real Problem Is Not Demand. It Is Capital Efficiency.

There is still enormous demand for AI computing, and the Gulf remains exceptionally well positioned to capture it. Saudi Arabia, the UAE and Qatar have capital, energy, land, government support and increasingly strong relationships with global technology companies. The problem is not finding reasons to build more computing capacity. The problem is building it at a cost that still produces an attractive return after every new layer of risk has been priced in.

A new data centre needs power, cooling, fibre, backup systems and physical protection. It also needs a plan for what happens when one part of that system fails. Recent disruption to cloud infrastructure in the UAE and Bahrain demonstrated why redundancy matters. AWS experienced outages affecting infrastructure in the region, forcing customers to rely on alternative availability zones and regions.

The lesson for investors is straightforward: one giant facility may be efficient, but it is not necessarily resilient. Developers can respond by building additional sites, alternative network routes and backup systems, but every layer of resilience requires more capital. Efficiency lowers the cost of computing; resilience lowers the cost of disruption. Investors now have to decide how much of both they are willing to pay for.

Why Lenders May Be Getting More Interesting

There is another shift worth watching: who actually wants to finance these projects. Infrastructure investors have shown growing interest in lending against large AI data-centre projects, while equity investors face a much more complicated proposition. For a lender, a data centre backed by strong customer contracts can look attractive. If a major technology company has committed to using the capacity and the project generates predictable cash flow, a lender can structure its exposure around those payments. Equity is different. The equity investor owns the uncertainty: construction can cost more than expected, insurance can rise, equipment can become obsolete, GPU economics can change, a customer can reduce its commitment, and a new generation of computing technology can alter what the facility actually needs.

That difference could become increasingly important in the Gulf. The region can continue attracting billions of dollars into AI infrastructure while becoming less attractive to certain types of equity investors. A headline saying that $5 billion is being invested tells you surprisingly little about the actual risk being taken. The useful questions are different: how much is equity, how much is debt, who carries the downside, who has the contractual cash flow, and who gets paid first if something goes wrong? A billion dollars of debt and a billion dollars of equity may finance the same building, but they represent very different bets on its future.

The Winners Will Not Be the Same as Before

Higher infrastructure risk does not mean everyone loses. Some businesses could become more valuable precisely because the cost of resilience is rising. Power producers and energy infrastructure companies are obvious beneficiaries because reliable electricity is becoming one of the most valuable inputs into AI capacity. Telecom and fibre operators have another advantage: if developers want multiple network routes rather than a single connection, connectivity becomes part of the resilience strategy. Lenders can benefit as well, particularly if developers become less comfortable funding projects entirely through equity. Large technology companies with strong balance sheets and long-term commitments can also gain negotiating power because a data-centre project backed by a credible customer is easier to finance than one built speculatively and waiting for demand to appear.

The pressure falls on a different group: developers whose economics depend on one location, one network route, uncertain customers or aggressive assumptions about future demand. Their problem is not that AI demand is disappearing. Their problem is that the cost of being wrong has increased. When insurance, redundancy and financing costs rise simultaneously, a project that looked highly profitable on paper can suddenly require a much higher utilisation rate or customer commitment to deliver the same return. That is why the next phase of the market will reward infrastructure that is not merely large, but financially and physically difficult to disrupt.

What Investors Should Watch Next

This is where the Gulf AI story becomes particularly useful for investors. The next data-centre announcement should not be judged by the size of the headline investment alone. Investors should look at the power supply and who controls it, the quality and duration of customer commitments, the number of alternative network routes, the project’s financing structure and the actual cost of insuring the physical asset. These details reveal far more about the quality of an AI infrastructure investment than another announcement about billions of dollars being committed to the region.

The same principle applies to companies building the infrastructure. Investors should start distinguishing between businesses that simply benefit from the growth of AI and businesses that control the bottlenecks required to make AI infrastructure work. Reliable electricity, high-capacity fibre, cooling, land, financing and long-term contracts are becoming more valuable because they determine whether a data centre can actually deliver the computing capacity it promises. In a market where everyone wants to build, the scarce assets may not be the buildings themselves. They may be the infrastructure and contracts that make those buildings economically viable.

The Gulf Is Not Becoming Less Attractive. It Is Becoming More Selective.

This distinction matters. The Gulf still has a powerful investment proposition: large pools of sovereign capital, access to energy, ambitious digital strategies and governments willing to support major technology infrastructure projects. Global technology companies are not