Iran War Shocks $2 Trillion AI Investment; Infrastructure and Hardware Bear the Brunt
According to Bloomberg columnist Parmy Olson, the outlook for $2 trillion in funding commitments from the Middle East has been clouded. Under the dual impact of soaring energy costs and chip supply chain pressures, the combined infrastructure investment of over $1.15 trillion by hyperscalers is bearing the brunt, while AI software layer companies such as OpenAI and Anthropic—shielded by the moat of enterprise contracts and lower energy dependence—remain relatively less affected
The Middle East conflict is reshaping the AI investment landscape, tearing this seemingly monolithic tech feast into two distinctly different halves.
According to the latest views of Bloomberg columnist Parmy Olson, funding commitments from the Middle East once reached $2 trillion, serving as a vital cornerstone for the U.S. AI race. However, as the United States becomes involved in this conflict, the prospects for related investments are being cast in shadow. Meanwhile, soaring energy costs are significantly driving up data center operating expenses, and the chip supply chain is also under pressure due to the volatile situation in the Middle East.
The ripple effect of this conflict is more likely to split the AI market in two rather than destroy it entirely. She stated that hyperscalers, represented by Alphabet, Amazon, and Microsoft, are bearing the brunt, while AI application layer companies such as OpenAI and Anthropic are minimally affected.
AI Software Layer: Enterprise Contracts Build a Moat
Parmy Olson pointed out that the AI market is essentially composed of two distinct parts: the high-cost infrastructure business and the relatively low-cost software business.
On the software layer, Anthropic's annualized revenue has more than doubled in the past three months to $19 billion; OpenAI's annualized revenue is approximately $25 billion. Enterprise clients, consumers, and government agencies across sectors like finance and life sciences continue to pay for subscriptions and access. Unlike previous hype cycles of the metaverse and cryptocurrency, this growth momentum is considered more sustainable.
AI software companies also benefit from sticky enterprise contracts. Customers are less likely to cancel contracts due to geopolitical uncertainties and are more inclined to maintain subscriptions in an effort to cope with potential economic fluctuations by improving organizational efficiency.
Furthermore, although AI software companies rely on data center operations, they do not directly bear the pressure of rising energy costs. OpenAI and Anthropic's daily operations primarily rely on "inference"—running existing models to respond to user queries. This consumes far less energy than training new, cutting-edge models. The latter requires thousands of GPUs to run continuously for weeks or even months, consuming an extremely high amount of energy, and can be temporarily postponed.
Hyperscale Cloud Providers: Under Pressure from Both Energy and Capital
In contrast, hyperscale cloud providers such as Amazon, Alphabet, Microsoft, Meta, and Oracle are in a more precarious position. Parmy Olson noted that these companies have collectively invested over $1.15 trillion in infrastructure construction. This massive expenditure is highly dependent on cheap and stable energy supply, particularly natural gas.
According to data from the International Energy Agency, natural gas accounts for approximately 40% of the electricity source for data centers in the United States, making it the most significant single energy source. The Middle East conflict is driving up energy prices, directly impacting the operating costs of related companies.
However, Alphabet and Amazon still have recurring cloud subscription revenue as a financial buffer.
Chip Supply Chain: TSMC and NVIDIA Face the Highest Risks
The chip supply chain is also highly exposed to the situation in the Middle East. TSMC manufactures almost all of NVIDIA's high-end chips, and about one-third of its fuel for production comes from the Middle East, with most of its helium originating from Qatar. Helium is crucial in semiconductor manufacturing for cooling and protecting silicon wafers. According to Olson's analysis, after the drone attack on QatarEnergy's Ras Laffan Industrial City by Iranian drones last week, helium production was compressed, and a full recovery of chip capacity could take months.
In this landscape, NVIDIA may be the company facing the highest risk. This company, with a market capitalization exceeding $4 trillion, primarily derives its revenue from selling chips to hyperscale cloud providers. Any factor that slows down the construction of large data centers will directly impact its orders.
Unlike Alphabet and Amazon, NVIDIA has no recurring revenue source as a buffer, relying solely on chip sales to sustain its business. It faces a double blow: on one hand, production is hindered; on the other, the prospect of large orders previously agreed upon with the Middle East is uncertain. Last November, the U.S. government approved NVIDIA to sell 70,000 of its most advanced chips to the UAE and Saudi Arabia. This transaction now appears even more uncertain. NVIDIA declined to comment.
Olson's analysis points out that energy resources and capital from the Gulf region have long been significant drivers of the AI infrastructure boom. Regardless of how strong the revenue growth is in the AI application layer, as long as the war continues, the prospects for the underlying infrastructure will become increasingly fragile.
