Morningstar DBRS: How an AI Investment Slowdown May Affect Tech Subsectors

The AI infrastructure buildout is accelerating rapidly, with major technology companies committing unprecedented levels of capital to data centers, processors, networking equipment, and power infrastructure. While this investment wave reflects strong confidence in long-term AI adoption, our research highlights the growing risks associated with such large-scale spending, including infrastructure overcapacity, declining returns on invested capital, power constraints, and increasing reliance on debt financing.

The analysis examines how a potential slowdown in AI-related revenue growth, moderating capital expenditure, or rising data center vacancy rates could affect credit quality across the broader AI ecosystem.

While the largest hyperscalers—including Amazon AMZN, Meta Platforms META, Google parent company Alphabet GOOG/GOOGL, and Microsoft MSFT—appear well-positioned to absorb such pressures because of their scale and financial strength, a slowdown could have significant implications for smaller and more specialized participants whose business models depend heavily on sustained AI infrastructure demand.

Key Highlights

• The AI buildout is gaining momentum in the United States, with technology companies increasing their capital expenditure.
• Historical infrastructure and technology cycles show that extraordinary capex growth eventually decelerates.
• A similar slowdown in AI investment would likely have repercussions throughout the broader ecosystem, including for data center operators, equipment manufacturers, semiconductor suppliers, and AI infrastructure financing vehicles.
• Smaller participants in the AI ecosystem may face deteriorating credit quality, thanks to their greater customer concentration, more limited financial flexibility, and weaker access to capital.

“The AI infrastructure expansion currently underway is one of the largest technology investment cycles in recent history, but history suggests that periods of extraordinary capital spending eventually moderate. A slowdown in AI investment could create excess capacity, weaken returns on capital, and place pressure on credit fundamentals across several technology subsectors. While large hyperscalers appear capable of absorbing these risks because of their scale, balance-sheet strength, and cash flow generation, more specialized participants—including AI-focused cloud providers, data center developers, semiconductor suppliers, and frontier model developers—could face significantly greater challenges if demand growth slows or infrastructure utilization falls below expectations,” says Morningstar DBRS vice president Julie Hung.

Most Resilient Sectors

Hyperscalers/Large Cloud Providers

• Hyperscalers risk committing excessive capital to data centers, processors, networking, and power before AI demand becomes sustainable.
• More efficient models or slower enterprise adoption may leave AI infrastructure underutilized.
• Parallel investment in Nvidia systems and proprietary processors also creates duplication, potential stranded capacity, and uncertainty regarding long-term infrastructure requirements.

Semiconductor Foundries

• Foundries face capacity, customer concentration, and architecture transition risks.
• Leading-edge production requires significant upfront investment and depends on demand from a limited number of chip designers.
• Product delays, inventory corrections, or a shift toward custom processors and more efficient inference hardware may reduce utilization and place pressure on margins and returns on capital.

Semiconductor Equipment Manufacturers

• Equipment manufacturers are likely to experience a delayed impact from any AI investment correction.
• Foundries, memory producers, and packaging companies may postpone or cancel orders after determining that sufficient capacity already exists.
• Long lead times increase the risk that equipment ordered during a shortage is delivered after demand, process requirements, or product architectures have changed.

Sectors with Moderate Risks

Data Center Developers/Operators

• Data center developers face construction delays, power constraints, permitting risk, cost overruns, and financing expenses before projects generate revenue.
• Once operational, facilities require sufficient utilization and pricing to cover power, cooling, maintenance, leases, and debt service.
• A mismatch between long-term financing obligations and the shorter competitive life of installed hardware may leave operators servicing debt on economically outdated systems.

Advanced Packaging Providers

• Advanced packaging suppliers risk expanding capacity based on temporary bottlenecks and aggressive demand forecasts.
• If AI chip demand moderates, specialized facilities may become underutilized and difficult to repurpose.
• Changes in product design, memory integration, process technology, or packaging architecture may reduce the competitiveness of recently installed capacity.

Networking/Optical Suppliers

• Networking and optical suppliers face excess inventory, rapid standard transitions, price competition, and customer insourcing risk.
• Concentrated demand from AI clusters may expose suppliers to data center construction cycles and shifts toward more distributed or communication-efficient architectures.
• Capacity optimized for a specific interconnect technology may lose relevance as standards and customer preferences evolve.

Memory Manufacturers

• Memory manufacturers face a shortage-investment glut cycle in high-bandwidth memory and server dynamic random-access memory.
• Aggressive capacity expansion may turn current scarcity into oversupply, leading to inventory accumulation, production cuts, falling prices, and write-downs.
• Rapid product transitions and customer qualification requirements create risk that investments lose competitiveness before generating adequate returns

High-Risk Sectors

Specialized Graphics Processing Unit Clouds / AI-Dedicated Cloud Providers

• Specialized cloud providers face high customer concentration, substantial leverage, and dependence on a narrow range of hardware options and workloads.
• The loss or repricing of a major contract may sharply reduce utilization and cash flows.
• Operators reliant on short-term contracts or spot pricing are particularly exposed to falling rental rates while fixed debt and lease obligations remain unchanged.

Large Language Model Developers / Frontier AI Labs

• LLM developers may enter long-term compute commitments before establishing durable revenue and cash flow.
• Falling token prices, open-model competition, and inference-efficiency improvements may compress margins while infrastructure costs remain fixed.
• Developers without proprietary data, distribution, workflow integration, or locked-in customers face the greatest risk of failing to recover training and compute investments.
• By contrast, leading AI model providers benefit from strong competitive positions, high customer demand for their frontier models, strategic partnerships with hyperscalers, and increasing workflow integration with enterprise customers.