Analysts have warned that Gulf investors committing billions of dollars to artificial intelligence infrastructure could face challenges in capturing long term financial returns despite the rapid expansion of the sector. As governments and sovereign wealth funds across the Gulf accelerate investments in data centres, semiconductor technologies, and artificial intelligence platforms, concerns are emerging that the substantial capital required to build computing infrastructure may not translate into equivalent commercial value. The discussion has gained momentum during the latest financial reporting season as technology companies continue increasing capital expenditure while investors closely monitor free cash flow, infrastructure utilization, and long term profitability. Market volatility affecting major technology companies and South Korean semiconductor manufacturers has also highlighted concerns about whether current investment levels could eventually result in excess capacity.
According to estimates published by technology publication The Information, the United Arab Emirates, Saudi Arabia, Kuwait, and Qatar have announced artificial intelligence investment plans exceeding 300 billion dollars across data centres, semiconductor technologies, and AI model development. In 2024, Abu Dhabi established MGX, a dedicated artificial intelligence and technology investment company chaired by Sheikh Tahnoon bin Zayed Al Nahyan. Last month, MGX announced the close of its first dedicated artificial intelligence fund with commitments totaling 49 billion dollars. Mubadala and G42, which founded MGX, also maintain significant exposure to artificial intelligence investments, while institutions including Qatar Investment Authority and Abu Dhabi Investment Authority are expected to hold similar interests. Oscar Mackereth, Partner at Cerno Capital, said investors should carefully evaluate who assumes the risks associated with hardware becoming outdated, unused infrastructure capacity, declining prices, and future equipment replacement. He explained that these risks can be transferred through long term agreements, but warned that challenges arise when project sponsors finance infrastructure while the financial benefits are realized elsewhere. Danni Hewson, Head of Financial Analysis at AJ Bell, added that although investment in artificial intelligence is widely expected, there is a possibility that infrastructure could be built faster than market demand develops, while rapid advances in technology may reduce the long term usefulness of facilities constructed today.
The discussion comes as major global technology companies continue reporting record levels of spending on artificial intelligence infrastructure. Alphabet increased its projected capital expenditure to as much as 205 billion dollars this year and reported negative free cash flow of 5.9 billion dollars after quarterly capital spending reached 45 billion dollars. Microsoft, which is a minority investor in OpenAI, reported a 70 percent annual increase in quarterly capital expenditure to 41 billion dollars, while its free cash flow declined by 23 percent to 19.6 billion dollars. Meta also expects to invest up to 145 billion dollars this year to expand data centres and other artificial intelligence infrastructure. Mackereth noted that free cash flow provides insight into financial pressure created by heavy investment, although it does not necessarily measure the long term return generated by those assets. He explained that purchasing artificial intelligence servers requires immediate cash expenditure, reducing free cash flow, while accounting practices spread the cost across the expected lifespan of the equipment. Some technology companies have extended server depreciation periods from approximately three years to six years, lowering annual depreciation expenses. He also emphasized that negative free cash flow does not automatically indicate financial weakness if future returns significantly exceed the cost of capital. However, he cautioned that current demand for computing capacity and high rental prices demonstrate limited supply rather than confirming that data centres and semiconductor infrastructure will remain profitable throughout their operational lifetime.
The report also highlighted concerns surrounding South Korean semiconductor manufacturers that supply memory chips used by hyperscale cloud providers. Despite reporting strong financial performance, including a sixfold increase in quarterly operating profit and margins exceeding 80 percent, SK Hynix experienced a significant decline in its share price after announcing plans to increase capital expenditure by approximately 50 percent to at least 31 billion dollars. Investors also expressed concerns about expanding competition from Chinese memory manufacturers. Together, SK Hynix and Samsung Electronics account for nearly half of the Kospi index, contributing to increased market volatility as investor sentiment shifted between optimism and caution. While both the Nasdaq and Kospi recorded strong rebounds following positive cloud computing results from Amazon, analysts noted that long term investment decisions depend on whether artificial intelligence infrastructure is expected to generate direct commercial returns or support broader national economic objectives. Mackereth observed that if Gulf governments view affordable computing capacity as a strategic objective rather than a direct revenue source, periods of infrastructure oversupply may be considered acceptable. He added that the more important question for Gulf investors is not whether computing resources become less expensive, but whether the businesses benefiting from that computing capacity will ultimately be located within their own economies.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
